TUTORIAL 2: WHAT IS MANAGEMENT?

The origins of a fast-food empire: Effectiveness and efficiency from the

beginning

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. McDonald’s long-term success is primarily due to the early technical and conceptual
skills of the McDonald Brothers.

B. McDonald’s long-term success is primarily due to Ray Kroc’s early effective

management of franchisees.

C. McDonald’s long-term success is primarily due to the ongoing management

functions of planning and controlling.

D. McDonald’s long-term success is mainly due to the ongoing management functions

of leading and organising.

Page 1 of 4
 The origins of a fast-food empire: Effectiveness and efficiency from the
beginning

Ray Krok in front of a McDonald’s restaurant. Photo licensed under CC BY

Ray Krok was a milkshake machine salesman, and by his own admission, things weren’t going very
well. Ray worked for manufacturer Prince Castle and the company was struggling to compete with
lower cost machines from their competitors. Ray was accustomed to doors slamming in his face and
his frustration was growing. The year was 1954 and Ray was waiting for the inevitable moment when
he would be fired. Luckily for Ray things were about to change. When checking in with his office he
was told of a restaurant that had bought 8 milkshake machines. Figuring this was either a mistake or
the greatest luck, Ray packed his car and drove to San Bernardino, California.
Richard and Maurice McDonald ran the most successful restaurant in San Bernardino. They had grown
up working at their father’s hot dog stand so were experienced. They had recently closed their store
to redesign and open serving burgers, french fries and milkshakes. While this was nothing particularly
radical, they were offering something different. They presented a highly rationalised and efficient
kitchen process which guaranteed burgers in 30 seconds, as opposed to the industry standard of 30
minutes. This was revolutionary for 1954.
Roadside hamburger restaurants were not as we know them today. In fact, Ray was quoted many
years later as saying “In my experience, hamburger joints are nothing but jukeboxes, pay phones,
smoking rooms, and guys in leather jackets. I wouldn't take my wife to such a place and you wouldn't
take your wife either”. You can start to imagine Ray’s shock when he finally arrived in San Bernardino
to see a parking lot full of cars, families sitting at tables or on the grass, and queues that were long but
moved quickly.
The restaurant was clean, modern, mechanised and the staff were well trained. Accepting the
opportunity to have behind-the-scenes tour, Ray was convinced that the brothers had the best
operation he had ever seen. Bespoke kitchen equipment ensured patties were cooked exactly the
same, sauces were distributed to each burger in exactly the same amount. The kitchen becomes a

Page 2 of 4
“symphony of efficiency” with each major food area moving seamlessly towards metal slides that
delivered the food toward the front counter. There was only one oddity in amongst this efficiency, all
of this equipment and kitchen space took up the whole restaurant interior. This means the public were
ordering and sitting outside of the restaurant.
After a sleepless night and a lengthy discussion with his wife, Ray returned to meet the brothers and
pitched the idea of franchising. They were initially reluctant but Ray convinced them this was the
future. Ray was hired as the first franchise manager for the McDonalds Brothers restaurant.
The next year the first franchise opened in Illinois, sadly it didn’t work out as the owners changed the
menu and the processes that made the restaurant so unique. Ray and the brothers worked hard to
ensure that things would work better in the future. They looked to selling single stores to middle class
people, as opposed to larger (and more expensive) territorial rights. The new franchise agreements
included contractual obligations to ensure same service, food preparation and most importantly the
layout of the “speedee service kitchen”, the most important feature of every McDonalds.
Over time Ray and McDonalds brothers clashed on how to grow the franchises. This frustration
eventually led to Ray raising $2.5 million USD to buy out the brothers.
Ray was now free to build the fast food empire he envisioned. He knew the secret was that all of the
food and drink needed to taste the same regardless of location. He recruited franchisees who bought
into the vision of working not for McDonalds but for themselves. He was quoted as saying “In business
for yourself, but not by yourself”. First and foremost, Ray advocated adherence to the McDonald’s
operating system that required franchisees to follow the core McDonald’s principles of quality,
service, cleanliness and value.
McDonalds Today

A modern McDonald’s restaurant. Photo licensed under CC BY-SA-NC

McDonald’s is regarded as the master of fast food service. One of the reasons for McDonald’s success
is its careful design of the fast-service system. Wherever you order a Big Mac meal in the world, you
can expect exactly the same quality of product and service. The burger is never over-cooked, the fries
are always crisp, the staff always smile and the restaurant is always clean and tidy. Nothing has really

Page 3 of 4
changed since Ray Kroc’s time. Considering that McDonald’s is operating more than 31 000 stores in
119 countries in the world and serve 47 million customers every day, it is amazing that they achieve
such a high consistency. So, how do they do it?
The recipe for McDonald’s success is a management system which makes sure nothing is left to
chance. The whole process of producing the products on the menu and how they are being prepared
and presented to the customers is carefully designed and orchestrated by the organisation’s
management to make sure the same high standards are achieved every time. It all starts with
McDonald’s having worked out the exact standards that each item on the menu has to meet. Raw
hamburger patties are carefully pre-packed and pre-measured and delivered to the stores in a frozen
form. They have to be of the same size and quality. For example, the basic hamburger patty is machine
cut, 1.6-ounce pure beef with no more than 10 per cent fat content. The same goes for the fries which
are delivered frozen in a pre-cut and partially cooked form to cut down on the time it actually takes
to deep-fry the fries in the store.
McDonald’s has worked out the most efficient way of preparing the food in the stores. In the past
hamburgers used to be cooked on one side until a buzzer and flashing light went off indicating it was
time for the person doing the hamburgers to flip the hamburger and cook the other side. This system
made sure the hamburgers were always cooked to perfection and not burnt on one side and under-
cooked on the other. However, McDonald’s decided a few years ago that this was not assuring 100
per cent accuracy and developed its cooking technology so that the hamburgers are now being cooked
on both sides at the same time. Not only did this achieve greater consistency in the cooking process,
but it also halved the time it takes to cook a hamburger. If you consider that millions of McDonald’s
hamburgers are cooked each day you can understand the time saving involved in refining the
technology and working out the best and most efficient way of doing the hamburgers. For example,
according to McDonald’s training program, the minimum cooking time for a regular patty is 37 seconds
and for a quarter pounder, it is 101 seconds.
Another example is in the production and handling of french-fried potatoes. French fries become
quickly soggy and unappetising. To be good they must be freshly made just before serving. Like other
fast food establishments, McDonald’s provides its outlets with pre-cut, partially cooked, frozen
potatoes that can be quickly finished in an on-premises deep-fry facility. The McDonald’s fryer is
neither so large that it produces too many French fries at one time (thus allowing them to become
soggy) nor so small that it requires frequent and costly drying.
Storage and preparation space and related facilities are expressly designed for and limited to the
predetermined mix of products. There is no space for any foods, beverages or services that were not
designed into the system at the outset. Thus the owners have no discretion regarding what they can
sell and the employees have virtually no discretion regarding how to prepare and serve any item of
food. The consistent high standard in the service is achieved by careful selection and training of staff.
McDonald’s training program is second to none in the hospitality industry and is designed to make
sure McDonald’s staff have the skills when it comes to food preparation and service. However, it
does not just end there. McDonald’s training also includes developing other skills such as
communication, time management, relationship building and leadership skills.

Additional material
Watch this film clip from the movie the Founder, which depicts a meeting between Ray Krok
and the McDonald Brothers. The video shows the McDonald brothers explaining how they
developed the Speedee Service System that allows them to make a burger in 30 seconds:
https://www.youtube.com/watch?v=jTageuhPfAM

Page 4 of 4
 TUTORIAL 3: ETHICS AND SOCIAL INCLUSION

ChatGPT sparks cheating, ethical concerns as students try realistic essay writing
technology

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements
A. When it comes to Generative Artificial Intelligence, managers of universities and
schools should take a utilitarian approach to ethical decision-making.

B. When it comes to Generative Artificial Intelligence, managers of universities and

schools should take a moral rights approach to ethical decision-making.

C. When it comes to Generative Artificial Intelligence, managers of universities and

schools should take a justice approach to ethical decision-making.

D. When it comes to Generative Artificial Intelligence, managers of universities and

schools should first consider issues related to diversity and social inclusion.

1
 ChatGPT sparks cheating, ethical concerns as students try realistic essay writing
technology

A new artificial intelligence chatbot that can generate realistic, human-like text is causing
intense debate among educators, with schools, universities and students divided about
whether it poses a threat to learning or will enhance it.

Key points:

• ChatGPT writes sophisticated essays and songs and answers questions

• Cheating and ethical concerns have been raised about the AI chatbot
• But some in the education sector say the technology should be embraced

Chat Generative Pre-Trained Transformer, known as ChatGPT, fluently answers questions

from users online and has the ability to write bespoke essays and exam responses.

Teachers are worried that students will use the tool to cheat and plagiarise, with some
universities moving quickly to rewrite exams, essay questions and integrity procedures.

Three states — New South Wales, Queensland, and Tasmania — have already banned
ChatGPT in public schools, and Western Australia's Education Department will next
week decide whether to form a similar policy, in time for the start of the school year.

'Helpful for initial draft': student guild

ChatGPT can quickly pump out a multitude of written responses — from explaining a
topic and writing speeches and computer code, to composing songs, poems, and short
stories.

The tool had over a million users sign up a week after its launch in November.

In Western Australia, Curtin University student guild president Dylan Botica said students
were quick to jump on board.

"For me, it's still a bit rudimentary in its early stages, but you can definitely see how it will
get better and be harder to detect," he said.

"It is really helpful to start with that sort of initial draft or getting some ideas on paper.

"I think other people see it as a tool that they're going to use.

[But] there have been a few students concerned their degrees won't mean as much if
everyone is using these tools."

2
'Tertiary experience' at risk

Mr Botica said universities needed to write assessments in a variety of ways and ensure
students were genuinely engaged in the learning process, in order to make them less
tempted to use AI.

"I don't think you're ever going to stop people from being able to use these services,
especially as they get more sophisticated," he said.

Curtin University student Ryan said he did not think ChatGPT was the answer, but
regulations were needed to ensure academic integrity.

"It undermines the tertiary experience of students coming out of university. Because if they
don't have that foundational knowledge, then they're probably not going to do as good a
job in industry," he said.

Fellow student Imari was apprehensive about using the tool.

"How much do you just trust this AI? Is it completely accurate? Is it taking from other
sources without you realising it?" they said.

Embrace technology: headmaster

While WA's Education Department mulls over how to respond to the technology, one
independent school in Perth has already made up its mind.

Scotch College headmaster Alec O'Connell said the department should be embracing the
technology, not banning it.

"I am not a great one for prohibition … I think it's better to look for ways to work with it.
Don't be scared, go find out more," he said.

Dr O'Connell said while screening for cheating in 2023 was complex, good teachers knew
their students well enough to know when they submitted work that was not their own.

"A while ago we would've been sitting here discussing Wikipedia. We had to work our way
through that as well," he said.

"We need to teach students the difference between right and wrong, and submitting work
that is not your own is morally incorrect.

Cheating concerns downplayed

A law and technology expert at the University of Western Australia (UWA), Julia Powles, felt
the cheating concern was "overblown".

"Ever since we've had the ability to search the web or access material on Wikipedia, people
have been able to draw on digital resources," she said.
3
"And if you're setting assessments that could be addressed simply by drawing on web
resources, then you may have a problem."

Associate Professor Powles said it was important to talk about technology, its ethics and
where the line was as a society.

"During COVID, we were forced to use lots of technologies, [such as] contact tracing," she
said.

ChatGPT has a range of capabilities, from writing essays to translation. (Supplied: ChatGPT)

"In education, we had tools — eye tracking [when students sat online] assessments — and
we really didn't look at the various compromises involved in those technologies when we
deployed them.

"We have the chance now. There is no rush."

She said many technologies, including ChatGPT, had a significant environmental and social
cost.

"Young people are curious about technology. But they should be curious too about the
implicit compromises of products developed by foreign companies that are scraping
material from all kinds of sources," she said.

4
Associate Professor Powles pointed to an investigation by Time magazine, which found the
multi-billion-dollar owner of ChatGPT, OpenAI, employed workers in Kenya for $2 an hour to
weed out the most abhorrent and sensitive content on the internet from the tool.

Workers reportedly had to sift through sexually explicit, racist, and offensive content for
hours a day, with many saying they experienced long-term mental health effects and PTSD
from the work.

"There is also a significant environmental cost in terms of computational intensity to train a

model like this," she said.

"Also, what does it mean for the sustenance of our creators and writers, if their works can
be taken for free without compensation and consent and regurgitated in a model like this?

"There is a corporate entity that's behind ChatGPT. They have their own commercial drivers
and they are backed by some of the biggest companies and most wealthy individuals in the
world, whose ends are not the same as those people of Western Australia."

Reference: Davis, A. (2023, January 26). ChatGPT sparks cheating, ethical concerns as
students try realistic essay writing technology. ABC News. Accessed 13 February 2023 from:
https://www.abc.net.au/news/2023-01-26/chatgpt-sparks-cheating-ethical-concerns-in-
schools-universities/101888440

5
 TUTORIAL 4: PLANNING AND CONTROLLING

Don’t just bet on the metrics – personal connection is the real key to managing remote
workers well

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. Focusing on relationships over metrics is important for managers of remote workers

because it overcomes the drawbacks of planning.

B. Focusing on relationships over metrics is important for managers of remote workers

because it is an effective approach to goal setting.

C. Focusing on relationships over metrics is important for managers of remote workers

because it overcomes the limitations of the control process.

D. Focusing on relationships over metrics is important for managers of remote workers

because it is a more effective way of using feedforward, concurrent and feedback
controlling.

1
 Don’t just bet on the metrics – personal connection is the real key to managing remote
workers well

Remote working has become much more common since the pandemic sent people home
from their offices and into the Zoom universe. This has put the onus on managers to find
new and better ways to engage with and motivate staff, and we’re still learning how best to
do this.

Perhaps the commonest advice given to managers is to provide remote staff with clear,
measurable goals. Set targets, monitor progress, measure results.

And at first glance, this makes perfect sense. After all, when staff aren’t in the office you
can’t keep an eye on them. Better to define outcomes and let them get on with it. You can
then measure achievement without micromanaging the process.

But our research suggests this advice, however well-intentioned, might not be entirely
correct. Indeed, if you follow it to the letter, it seems unlikely you’ll get the results you’re
expecting.

Some work is harder to measure

We began researching remote work before the pandemic by asking experienced managers
what they did differently with remote staff and what factors they felt were most important.

Earlier studies suggested managers would set clear goals and then monitor progress,
otherwise leaving staff to get on with it. Instead, the managers in our study told us they
tended to treat remote workers holistically – as people, not just staff. They focused on social
connections instead of just objective performance.

We believe these managers have learned to discount conventional advice and develop
practices that are more effective with remote staff. As such, their experience contains
valuable lessons for today’s managers.

One described spending one-on-one calls “just connection and socialising … [It] builds the
rapport and the connection [because] you lose opportunities for that being remote.”

One of the main reasons why these managers didn’t rely on metrics and measurable goals
with their remote staff is that a great deal of remote work is “knowledge work”. It’s often
difficult to quantify this kind of work – to know in advance what will be required or how long
it will take.

And because people often work in teams, it can be hard to untangle individual
contributions. At best, managers may have rough estimates of individual productivity. But if
someone doesn’t meet those estimates, they need more context to understand why – and
that relies on talking to people.

Hoping to only measure progress in unpredictable knowledge work the way we measure
progress with more predictable work is unlikely to succeed.

2
 An office allows managers to read people’s moods more easily, and this can be key to
evaluating performance. Getty Images
Nurturing relationships

The other reason these managers needed more than goals and metrics was the lack of
interpersonal contact. They could not unobtrusively observe how remote staff were coping.
As one manager put it, “You don’t have that visual of, oh, this person walked in today and
looks happy or they look sad.”

Another manager pointed out that with in-person interaction, “You can see people being
playful, hear the grunts or the moans or the sighs or whatever when they’re under pressure.
And likewise, you can see when they’re visibly, you know, feeling successful.” With remote
staff, “you just never get a sense of that”.

This day-to-day observation was important enough that the managers we spoke to devoted
considerable time and energy compensating. They increased communication with remote
staff and nurtured relationships.

Advising managers to “focus on clear objectives” with remote workers overlooks the
importance of relationships as the basis for understanding performance.

Furthermore, when people are stressed, disengaged or unsure they’re doing a good job,
their first instinct is not always to share that with their manager. Instead, they might try to
mask those feelings and keep up a good appearance – what one manager called “sticking on
a plastic smile for ten minutes”.

3
Intervention to avoid escalation

In a shared workspace it’s easier to pick up body language, changes in behaviour or working
relationships that might hint at deeper problems. That’s much harder when we
communicate primarily by email, text or short video calls.

And these cues can be very important. If someone misses a delivery deadline or is less
productive, for example, their manager is only likely to find out after it happens. Whereas
noticing a change in behaviour earlier might give the manager the chance to intervene and
address the problem before it escalates.

Overall, we recommend embracing interpersonal connections rather than relying on

measurable goals for what might be unmeasurable work. The fewer the opportunities for in-
person observation, the more important establishing good one-on-one relationships
becomes.

It will take more time, and maybe more emotional energy, but there’s no shortcut to
managing well.

Reference: Downes, R, Donnelly, N. &. Daellenback, U. (2022, November 28). Don’t just bet
on the metrics – personal connection is the real key to managing remote workers well. The
Conversation. Accessed 13 February 2023 from: https://theconversation.com/dont-just-bet-
on-the-metrics-personal-connection-is-the-real-key-to-managing-remote-workers-well-
193206

4
 TUTORIAL 5: LEADING

The value of virtue: 7 reasons why Volodymyr Zelenskyy’s crisis leadership has been
so effective
Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements:

A. Volodymyr Zelenskyy is an effective leader because of the way he has mobilised his
power in a crisis.

B. Volodymyr Zelenskyy is an effective leader because of his virtuous leadership traits.

C. Volodymyr Zelenskyy is an effective leader because of the way he has adapted his
leadership to the contingencies of a crisis.

D. Volodymyr Zelenskyy is an effective leader because of his charismatic leadership

style.

1
 The value of virtue: 7 reasons why Volodymyr Zelenskyy’s crisis leadership has been
so effective

The war in Ukraine would test even the most hardened political operator: millions forced
to flee their homes, thousands (including many civilians) killed or injured, evidence
of Russian war crimes mounting.

Yet Volodomyr Zelenskyy, a relative novice head of state, has not just risen to the challenge,
he has been widely praised and admired for his exemplary crisis leadership. So, what
explains this prowess?

Zelenskyy’s acting experience has been credited with his ability to connect powerfully with
different audiences, using facts and emotions to build support for the Ukrainian cause.

His commitment to serve his people has been called pivotal. He has been described
as charismatic – although this alone is no guarantee of success, given charismatic leaders
can still lead their nations to destruction.

And it’s Zelenskyy’s repeated displays of courage that seem to really strike a chord with
many. This leads us into the territory of character virtues, which we argue hold the key to
Zelenskyy’s abilities as a crisis leader.

Ancient wisdom for today’s world

Aristotle is credited with first proposing that virtues play a central role in forging a strength
of character that can navigate and weather life’s challenges with moral fortitude and
integrity.

Over the past few decades, scholars concerned with preventing unethical
leadership have developed Aristotle’s insights further, using modern social scientific
methods.

Recently, we drew on this knowledge to examine crisis leadership and how character
virtues guided 12 heads of state through that first, tumultuous wave of COVID-19. We’ve
used the same approach to analyse Zelenskyy’s leadership.

We closely examined an extended filmed interview with Zelenksyy by The Economist. Being
unscripted and more spontaneous than his pre-prepared speeches, it offered a clearer
insight into his character.

We found all seven of the key character virtues – humanity, temperance, justice, courage,
transcendence, wisdom and prudence – evident in Zelenskyy’s responses to the
interviewers’ questions.

2
Character virtues in action

The virtue of humanity relates to care, compassion, empathy and respect for others.
Zelenskyy demonstrates this primarily through his focus on protecting Ukrainians from
Russian aggression, but it even extends to his enemy’s suffering.

Zelenskyy expresses concern that Putin is “throwing Russian soldiers like logs into a train’s
furnace”, and laments that the Russian dead are neither mourned nor buried by their own
side.

This refusal to simply give way to hate and anger when speaking of his enemies also reflects
a second virtue, temperance – the ability to exercise emotional control.

Zelenskyy’s modesty also reflects this virtue – in the interview he shrugs off praise for being
an inspirational hero, preferring to keep to the main issues. Temperance serves to maintain
emotional equilibrium, thus enabling Zelenskyy to make difficult decisions in a level-headed
manner.

The virtue of justice means acting responsibly and ensuring people are treated fairly. It
involves citizenship, teamwork, loyalty and accountability. Zelenskyy speaks of his “duty to
protect” Ukrainians and to “signal” with his own conduct how others should act. By
remaining in Ukraine, he becomes a role model of this virtue while simultaneously
demonstrating the virtue of courage.

Zelenskyy’s own courage has been widely noted, but we observed that he also repeatedly
acknowledges that of his fellow citizens, thereby encouraging them to act with virtue.

3
Humanity as virtue: Volodymyr Zelenskyy visits a hospital in Kyiv in late March. GettyImages
A formidable opponent

By expressing the seemingly unshakeable hope that Ukrainians will secure victory because
of their courage, Zelenskyy demonstrates the virtue of transcendence – the optimism and
faith that a cause is meaningful, noble and will prevail.

Zelenskyy’s views about what motivates other countries display his wisdom. In the interview
he demonstrates a broad strategic perspective and insight into the varying interests that
shape other nations’ responses to the war. This helps him craft his appeals to allies, and to
Russia, which then have a greater chance of resonating.

The final virtue, prudence, complements that wisdom. It involves an ability to gauge what is
the right thing to do and is something of a meta-virtue, guiding the choice of which other
virtues are needed from moment to moment. We found repeated instances of Zelenskyy
demonstrating just that, weaving together multiple virtues in his responses to questions.
Our analysis of his leadership indicates Zelenskyy possesses strength of character and
emotional, intellectual and moral clarity about what is at stake. This explains his effective
crisis leadership to date. Despite the clear military mismatch between Russia and Ukraine,
Putin has taken on a formidable opponent.

Additional Resources: Watch The Economist interview with Zelenskyy:

https://youtu.be/LNjiRmIWtss

Reference: Wilson, S. & Newstead, T. (2022, April 29). The value of virtue: 7 reasons why
Volodymyr Zelenskyy’s crisis leadership has been so effective. The Conversation. Accessed
2 January 2023 from: https://theconversation.com/the-value-of-virtue-7-reasons-why-
volodymyr-zelenskyys-crisis-leadership-has-been-so-effective-182041

4
Journal of Business Ethics (2023) 183:653–670
https://doi.org/10.1007/s10551-021-05032-7

ORIGINAL PAPER

Are Algorithmic Decisions Legitimate? The Effect of Process

and Outcomes on Perceptions of Legitimacy of AI Decisions
Kirsten Martin1 · Ari Waldman2

Received: 7 May 2021 / Accepted: 26 December 2021 / Published online: 10 February 2022
© The Author(s) 2022

Abstract
Firms use algorithms to make important business decisions. To date, the algorithmic accountability literature has elided
a fundamentally empirical question important to business ethics and management: Under what circumstances, if any, are
algorithmic decision-making systems considered legitimate? The present study begins to answer this question. Using facto-
rial vignette survey methodology, we explore the impact of decision importance, governance, outcomes, and data inputs
on perceptions of the legitimacy of algorithmic decisions made by firms. We find that many of the procedural governance
mechanisms in practice today, such as notices and impact statements, do not lead to algorithmic decisions being perceived
as more legitimate in general, and, consistent with legitimacy theory, that algorithmic decisions with good outcomes are
perceived as more legitimate than bad outcomes. Yet, robust governance, such as offering an appeal process, can create a
legitimacy dividend for decisions with bad outcomes. However, when arbitrary or morally dubious factors are used to make
decisions, most legitimacy dividends are erased. In other words, companies cannot overcome the legitimacy penalty of
using arbitrary or morally dubious factors, such as race or the day of the week, with a good outcome or an appeal process for
individuals. These findings add new perspectives to both the literature on legitimacy and policy discussions on algorithmic
decision-making in firms.

Keywords AI · Legitimacy · AI ethics · Algorithms · Technology ethics

Introduction public is skeptical. Roughly six in ten are concerned that

ADM will be biased and unfair; two-thirds think algorith-
There is a new challenge to firm legitimacy—namely, the mic financial scoring and hiring decisions are unacceptable
increasing use of algorithms and computers to make deci- (Smith, 2018). Studies show that people are largely con-
sions that directly impact people’s lives (Citron & Pasquale, cerned about privacy risks, biases, fairness, and usefulness
2014). Businesses use algorithmic decision-making (ADM) of ADM (Araujo et al., 2020). They think ADM systems
systems to make hiring, firing, and promotion decisions are less “authentic” and, therefore, less ethical than iden-
(Ajunwa, 2020). Banks assess loan eligibility and credit tical human decisions (Jago, 2019). And Newman et al.
risk algorithmically (Citron & Pasquale, 2014). Social media (2020) have shown that individuals perceive ADM used by
platforms rely on ADM to moderate and curate content (Vin- human resources departments as reductionist and unfair.
cent, 2020; Gillespie, 2019). Recommendation algorithms This suggests that ADM may materially impact corporate
nudge us toward products, news, entertainment, and infor- legitimacy because a firm’s legitimacy is a product of and
mation (Zuboff, 2019). positively correlated with its actions (Chung et al., 2014;
And yet, despite the increasingly frequency with which Santana, 2012).1 Within business ethics, the perceived
management decisions are being made algorithmically, the legitimacy of stakeholders—consumers, users, employ-
ees, etc.—are important to firms as key audiences to their

* Kirsten Martin
[email protected] 1
We recognize, following Eesley and Lenox (2006), that deci-
1 sion legitimacy merits separate evaluation and study. And, the link
University of Notre Dame, Notre Dame, USA
between decision legitimacy and overall firm legitimacy is of critical
2
Northeastern University, Boston, USA importance to business and business ethics scholars.

13
Vol.:(0123456789)
654 K. Martin, A. Waldman

actions, and legitimacy is seen as a fundamental concept in Second, we contribute to the business ethics scholarship on
business ethics (Chen et al., 2020). In their reassessment of legitimacy by considering it through the lens of independ-
legitimacy as a stakeholder attribute, Ali (2017, p. 157) has ent variables—decision importance, procedural governance,
called identifying stakeholder perceptions of legitimacy one and outcomes—that have not been studied in the business
of three “pivotal” questions for legitimacy. In other words, ethics literature (Badas, 2019; Gibson & Caldeiera, 2009;
understanding what is or is not legitimate is a “process that Tyler, 2006/1990). Finally, we contribute to public policy as
involves perceptions” (p. 160) because stakeholder and firm regulators and firms struggle to understand how to improve
legitimacy is “based on the perceptions of the society, stake- ADM systems for all stakeholders. This research suggests
holders, and management” (p. 164). that testing for arbitrary or morally dubious factors is more
Given the risks that ADM systems pose to corporate legit- important for perceptions of legitimacy of the decision and
imacy, this study seeks to determine whether and under what the company than offering a better notice or impact state-
circumstances, if any, the use of ADM to make commer- ments, two particularly common recommendations in the
cial decisions impacts perceived legitimacy of the decision. literature (Kaminski, 2019a, 2019b, 2019c; Katyal, 2019;
To be clear, this study does not seek to normatively defend Reisman et al., 2018). Current law and new regulatory pro-
the practice of using ADM to making social, economic, or posals prioritize procedural rules for ADM that, our study
policy decisions about people. Nor does it suggest that adop- shows, will not increase perceived legitimacy.
tion of any particular form of governance makes the use of
ADM legitimate in a democratic society. Rather, we build
on the empirical work of Elsbach (1994), Finch et al. (2012), Theoretical Context
and Jahn et al. (2020) in that we seek to empirically measure
perceptions of legitimacy of commercial actors. We do so Algorithmic Decision‑Making in Firms
in a novel context of ADM and also add to the literature by
considering the relative importance of factors in determining This study focuses on algorithmic-decision-making (ADM)
legitimacy. We use the terms legitimacy dividend and legiti- systems, which, following Calo (2017), we define generally
macy penalty to describe the positive and negative effects as processes involving algorithms, or sequences of logical,
that the presence or absence of a condition has on perception mathematical operations, to implement policies by software.
of the legitimacy of an algorithmic decision. Some ADM tools are powered by artificial intelligence of
Using factorial vignette survey methodology to survey varying maturity and types; our study focuses on algorithms
individuals’ normative judgments about algorithmic deci- developed from training data generally. We also recognize
sions, we ran nine surveys to measure the relative impor- that there exists a range of automation, with some deci-
tance of governance, outcomes, and inputs on perceived sions almost fully computerized while others are merely
ADM decision legitimacy. We find that although decision augmented with technology (Araujo et al., 2020; Martin,
importance is negatively associated with perceptions of 2018). Defining all ADM systems is not our goal. As this
legitimacy, and outcomes are positively associated with study focuses on perceptions of legitimacy of algorithmic
perceptions of legitimacy, only particularly robust proce- decisions of private, commercial firms, we assume a defini-
dural governance increases perceptions of legitimacy when tion that recognizes the role of data inputs, computers, and
algorithmic decisions lead to negative outcomes or outcomes automation of decisions.2
with which individuals disagree. That said, neither positive Unfortunately, researchers have shown that such algo-
outcomes nor procedural governance can correct for the sig- rithmic systems are risky. Because ADM systems make
nificant legitimacy penalties associated with ADM systems probabilistic predictions about the future, these programs
that use arbitrary and morally dubious factors such as race. may make mistakes about the ambiguous situations they are
Given that predictive ADM systems reify biases in society, trying to predict; probabilities are necessarily generalized,
our results have significant impact on firm decisions and with individual cases falling through the cracks (Eubanks,
public policy. 2018; Hu, 2016). Studies have also shown that ADM’s
We contribute to legitimacy theory by explicitly test-
ing the perceived legitimacy of ADM. Extending this
type of analysis to algorithmic and computer-driven deci- 2
In conceptualizing ADM in this way, we follow a long line of
sions, considering the degree to which the public is aware scholars studying popular perceptions of artificial intelligence (AI)
and skeptical of such decisions (Araujo et al., 2020), we generally (Araujo et al., 2020; Dodge & Kitchin, 2007). This paper
extend the scope of legitimacy scholarship. First, we build investigates the development of an algorithm from training data and
the use of that algorithm to make a decision about individuals. We
on work by Elsbach (1994), Finch et al. (2012), and Jahn
do not study the underlying technique used to develop the algorithm
et al. (2020), which has explored positive drivers of decision (neural networks, supervised or unsupervised learning, etc.). We
legitimacy by demonstrating factors’ relative importance. leave that for future study.

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 655

predictive capabilities may be exaggerated (Dressel & Farid, create value for stakeholders (Freeman et al., 2020). As an
2018; Jung et al., 2017; Salganik et al., 2020). ADM’s opac- actor within a larger industry or marketplace, firms have an
ity makes algorithmic systems difficult to interrogate and obligation to be perceived as legitimate and act in a man-
hold accountable (Cheng, 2013; Loi et al., 2020). Algorith- ner that is “desirable, proper, or appropriate within some
mic systems also provide an incentive for surveillance and socially constructed system of norms, values, beliefs, and
data collection because they need large information sets for definitions” (Suchman, 1995, p. 574) because a firm’s legiti-
model training and analysis (Martin, 2019). This creates the macy impacts the perceived legitimacy of the larger industry
circumstances for invasions of privacy and erosion of pri- or marketplace. In other words, a firm acting in a manner
vacy norms (Zwitter, 2014). that is not “desirable, proper, or appropriate” within the
Finally, ADM systems are biased (Johnson, 2020, n.d.; system of values has harmful spillover effects on similarly
Winner, 1980). They rely on the corpus of data on which situated firms and markets, which then harms others.
they are based and, therefore, data that is biased along race, In addition, firms should care about whether their actions
gender, sex, and socioeconomic lines will lead to discrimi- are perceived as legitimate for the survival of their own firm
natory results (Caliskan et al., 2017; Katyal, 2019; Noble, since stakeholders, including employees, suppliers, and cus-
2018; O’Neil, 2016). Developers make value-laden assump- tomers, among others, rely on the firm to survive and thrive.4
tions about how to treat missing data and outliers that also A firm that is perceived as more credible and, thus, per-
impacts the bias of the developed algorithm (Martin, 2022). ceived as a more legitimate company has greater customer
Recently, scholars have begun investigating not only the loyalty, political leeway, is perceived to make decisions for
harms and discrimination created by ADM but also how appropriate reasons, and enjoys voluntary acceptance and
individuals perceive decisions that rely on algorithms. This compliance (Chung et al., 2014; Deephouse et al., 2017, p.
work has paralleled existing theory on the perceptions of 32; Finch et al., 2012; Suchman, 1995).5
fairness, trust, and legitimacy of firms and decisions. When a firm is perceived to be legitimate, it can generate
public support and loyalty (Boyd, 2000; Coombs, 1992). In
Legitimacy addition, organizations with high perceptions of legitimacy
have more flexibility in decision-making and less public
At the organizational level, entities are perceived to be and political interference. In other words, if an organization
legitimate when their actions are “are desirable, proper, wants to maintain customer loyalty, stability within a regula-
or appropriate within some socially constructed system of tory environment, and achieve market success, it needs to be
norms, values, beliefs, and definitions” (Suchman, 1995, p. perceived as legitimate, both with respect to the organization
574). Legitimate firms pursue “socially acceptable goals in as a whole and its individual decisions (Suchman, 1995).
a socially acceptable manner” (Ashforth & Gibbs, 1990, p. Those firms that make illegitimate decisions are considered
177). Legitimacy is an “attitude” (Jahn et al., 2020, p. 546) illegitimate themselves, and those firms cannot survive in
or “perception” (Suddaby et al., 2017, p. 451) that is closely the market (Ruef & Scott, 1998). Therefore, businesses have
related to institutional credibility. powerful ethical, economic, political, and social incentives
Firms seek legitimacy3 for ethical and strategic reasons to ensure that both their individual decisions and entire
in a manner that is similar to those seeking to be trustwor- organizations are perceived by the public to be legitimate.
thy and fair. For most strategy and business ethicists, being A number of factors are known to influence perceptions
fair, trustworthy, and legitimate is both strategically sound of organizational legitimacy. Organizational legitimacy has
as well as a moral imperative and an end in itself (Freeman
et al., 2020; Phillips, 1997; Pirson et al., 2017). The norma-
tive argument for firms caring about legitimacy parallels the
4
argument for why firms should care about being perceived While these two arguments—an overtly normative or ethical argu-
ment in one paragraph is separated from the more ‘strategic’ argu-
as trustworthy and fair. Firms should seek to understand ment in a following paragraph, the authors do not see these as being
the concerns of their stakeholders—employees, users, cus- so separate (Freeman, 1994; Freeman et al., 2007). However, we
tomers, communities—based on obligations of care (Wicks would like to thank a reviewer for reminding us that some readers
et al., 1994), fairness (Phillips, 1997), and an obligation to need to see the argument made separately. We see acting ethically as
good for the firm and that actions that create value for stakeholders
also create value for the firm.
5
Tyler (2006/1990) and Tyler and Huo (2002, p. 102) define legiti-
3
The phrase “legitimacy of the firm” references the perceived legiti- macy more narrowly, as “perceived obligation to comply with the
macy by others as we build on legitimacy as an “attitude” (Jahn et al., directives of an authority, irrespective of the personal gains” or “a
2020, p. 546) or “perception” (Suddaby et al., 2017, p. 451). This is quality possessed by an authority, a law, or an institution that leads
parallel to referring to the trustworthiness of a firm when trustworthi- others to feel obligated to obey its decisions and directives voluntar-
ness is defined as the belief of others as to the trust factors of the firm ily.” Both Suchman and Tyler recognize that there is a moral valence
(Pirson et al., 2017). to legitimacy.

13
656 K. Martin, A. Waldman

been positively associated with conformity to norms (Sud- popular perceptions of legitimacy hinges, at least in part,
daby et al., 2017), negatively correlated with extrinsic or on the existence of procedural safeguards and the opportu-
self-interested motives (Jahn et al., 2020), and dependent nity to be heard. In Tyler’s work, the legitimacy dividend
upon transparency and communication (Elsbach, 1994). of fair processes overcome any lingering distrust, opposi-
Corporate social responsibility actions and rhetoric impact tion, or negative reaction associated with an adverse result
perceptions of legitimacy of firms (Bronn & Vidaver-Cohen, (Tyler, 1994). That is, even those individuals who came
2009; Castello & Lozano, 2011; Du & Vieira, 2012), as do out worse off due to the actions of authorities, institutions,
transparent executive responses to crises (Beelitz & Merkl- or law were willing to comply with the law if the process
Davies, 2012). was fair (Easton, 1965).
Although there are different types of legitimacy in the Similar findings have been made in organizational stud-
organizational literature (Palazzo & Scherer, 2006; Sud- ies. Elsbach (1994) used a within-subject experimental
daby et al., 2017), this study is based on the theory that design to test the cattle’s industry’s legitimizing strategies,
individual firm decisions impact organizational legitimacy— finding that the public expects companies to communicate
namely, the perception that a firm is operating according to about how they are conforming to accepted norms and
shared norms, values, and rules (Chung et al., 2014). That values. Finch et al. (2012) examined perceptions of the
perception is based, at least in part, on what Boyd (2000) legitimacy of the environmentally unfriendly oil sands
and Coombs (1992) call issue legitimacy—namely, the per- industry in California, finding that participants who favor
ception that a particular corporate activity is appropriate, environmental values judged the industry as less legiti-
understandable, and done for the right reasons. mate than those who supported economic development
Within this line of scholarship, organizational legitimacy and outcomes. Like similar studies in the legal literature
is driven by the perceived legitimacy of an organization’s described above, these studies suggest that process—in
specific claims, decisions, and behavior (Dowling & Pfeffer, this case, notice—and outcomes may be associated with
1975; Eesley & Lenox, 2006; Santana, 2012). To the extent popular perceptions with legitimacy.
that a firm uses ADM to make salient decisions about its Both perceptions of fairness and legitimacy have been
stakeholders, the legitimacy of those decisions will have a applied to ADM. Scholars have investigated respondent-
significant impact on the legitimacy of the organization as level attributes and decision factors that contribute to
a whole. This parallels findings that the perceived fairness perceptions of fairness with ADM systems (Araujo et al.,
of an algorithmic decisions impacts someone organizational 2020; Nagtegaal, 2021; Newman et al., 2020). However,
commitment to the firm (Newman et al., 2020). according to Kaina (2008), legitimacy is a distinct con-
cept from fairness and trust. Within work on judgements
Decision Legitimacy about algorithmic decisions, nascent work on legitimacy
of algorithmic decisions has suggested that governance
At the level of individual decisions, motives are also impor- is important to the legitimacy of ADM. For example,
tant drivers of perceptions of legitimacy in the legal studies Danaher et al. (2017) examine the legitimacy of ADM
literature on legitimacy, which mostly focuses on popular governance mechanisms (rather than the legitimacy of the
perceptions of legitimacy of institutions like the Supreme decision), and Lünich and Kieslich (2021) studied attrib-
Court or the police. For instance, Badas (2019) showed utes of respondents (i.e. general trust) in their perception
that when a court hands down a decision with which one of legitimacy of a vaccine distribution algorithm. Starke
disagrees, individuals think the decision was motivated by and Lünich (2020) empirically examine input, process,
extrinsic factors like politics and ideology and is, therefore, and output legitimacy of EU decisions. What Starke and
illegitimate. Badas (2019), Bartels and Johnston (2013), Lünich (2020) did not do, and what we attempt to do here,
and Christenson and Glick (2015) also suggest that policy is empirically assess the relative importance of inputs, pro-
disagreements and ideology influence perceptions of the cess, and outcomes to the perceived legitimacy of algo-
legitimacy of a given court decision and the institution of rithmic decisions.
the judiciary as a whole; when a court makes a decision
with which individuals disagree or negatively affects those
individuals, they tend to see the court as less legitimate.
At the same time, Gibson and Caldeira (2009) and Gibson Hypotheses
et al. (2005) have shown that repeated exposure to symbols
of authority create a positivity bias that insulates institutions We next develop hypotheses as to the drivers of perceived
like the Supreme Court from legitimacy penalties. legitimacy of algorithmic decisions. Based on legitimacy
And in canonical studies of legal legitimacy, Tyler and algorithmic decision scholarship, we expect the type
(2006/1990) and Sunshine and Tyler (2003) showed that

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 657

of decision, inputs, process, and outcomes to impact the reasons or basing decisions on factors outside the context of
perceived legitimacy of algorithmic decisions. the decision would render the decision unjust (Nozick, 1974;
Walzer, 2008). For algorithmic decisions, using race-based
Decision Type factors in ADM contributes to illegal, and seemingly not
legitimate, decisions (Barocas & Selbst, 2016). And theo-
Social scientists studying algorithms argue that the type retically, using race as a factor for decisions that have noth-
of decision being made algorithmically may drive norms ing to do with race can delegitimize decisions in law (Ellis
around appropriate governance. Where Nagtegaal (2021) & Diamond, 2003). The extent to which the use of arbitrary
finds that the complexity of the decision impacts the percep- factors delegitimizes ADM has been theorized but has not
tions of fair ADM, others have theorized that the degree the been empirically tested. This leads to our third hypothesis:
decision is important in the life of the individual will drive
perceptions of algorithmic decisions. For example, Tufekci H3 The use of arbitrary or race-based factors has a negative
(2015) highlights the importance of governing “gatekeep- impact on perceived legitimacy.
ing” algorithms; Calo (2017), Burrell (2016), O’Neil (2016),
and Martin (2019) argue that “consequential” or “pivotal” Governance
decisions should be given the most attention. In other words,
scholars have theorized that individuals’ perception of the One of the most dominant theories in legitimacy studies
legitimacy of commercial use of ADM varies with their suggests that people perceive even adverse decisions as
assessment of the decision’s importance in their lives. This legitimate as long as fair procedural governance mechanisms
leads to our first hypothesis: are in place (Tyler, 2006/1990). Danaher et al. (2017) also
focuses on the legitimacy of the governance mechanism of
H1 As decision importance increases, individuals’ percep- algorithmic decisions. This is the approach taken by many
tion of the legitimacy of using an algorithm to make the current and proposed data protection laws, which require
decision decreases. organizations to communicate to customers the “logic
behind” its algorithms and/or complete impact assessments
In other words, we predict a larger legitimacy penalty for (Kaminski, 2019a, 2019b, 2019c). Indeed, current policy
algorithmic hiring and firing decisions or access to health debates center not on whether ADM requires some form
care decisions than for decisions about music playlists or of accountability mechanism, but on what that governance
social media curation. regime should look like (Pasquale, 2019). The theoretical
argument within legal legitimacy scholarship is that robust
Outcomes governance—impact assessments (Kaminski, 2019b), audit
trails, detailed explanations, publicly accessible code, sys-
Sociolegal scholarship on legitimacy also suggests that deci- tems testing (Citron, 2007, pp. 1305–1313), human-in-the-
sion outcomes matter for popular perceptions of the legiti- loop (Froomkin et al., 2019; Jones, 2017; Rahwan, 2018),
macy of the institutions and processes led to those outcomes. impact assessments (Katyal, 2019; Reisman et al., 2018),
For example, Badas (2019) theorizes that disagreement with contestability (Mulligan et al., 2020), and codes of conduct,
the outcome of institutional decisions negatively impact per- impact assessments, and whistleblower protections (Katyal,
ceptions of the legitimacy of those institutions, and Gibson 2019) —will legitimize decisions regardless of the outcome.
and Caldiera’s (2009) “positivity theory” posits that agree- Within ethics scholarship, the theorizing around governance
ment with the outcome an institutional decision contributes and legitimacy is mixed. For example, de Fine Licht and de
to a larger legitimacy dividend than the legitimacy penalty Fine Licht (2020) theorize that providing justifications for
associated with disagreement with outcomes. algorithmic decisions could provide grounds for perceived
legitimacy, whereas, Leicht-Deobald et al. (2019) suggest
H2 With all other factors held constant, a good, positive that algorithm-based decisions may be perceived as more
outcome is associated with a legitimacy dividend, or an legitimate because it is difficult for individuals to question a
increase in the perceived legitimacy of the decision. complex, computerized system (Fig. 1).
In other words, the literature suggests that perceptions of
Arbitrary Rationales legitimacy should increase as the robustness of procedural
governance mechanisms increases. However, endogeneity
As the algorithmic accountability literature makes clear theory (Edelman, 2016) suggests that many legal rules are
(Noble, 2018; O’Neil, 2016), predictive algorithms based developed through an endogenous process involving regu-
on discriminatory data and discriminatory modeling produce lated entities themselves. If individuals perceive those rules
unjust results. For justice scholars, the idea of using arbitrary unfair, unjust, or the product of corporate self-interest (Jahn

13
658 K. Martin, A. Waldman

Fig. 1  Figure of hypotheses

Type of Decision

Pivotal Decision

Outcome
–
+ H2
Favorable H1

Substance – H3
+
Perceived Legitimacy
Race
H4c H4b
+ +
H4a
Governance +

Appeal

Table 1  Types of decisions Decisions As operationalized in the vignettes

Ads Which ads a person sees online ….their predicted preference and interests
Music What songs are suggested in a person’s playlist… their predicted preferences and interests
Video Whether someone's posted videos are taken down online & their past and predicted behavior
Hired Who is hired at a company….the applicant’s predicted likelihood to be successful at that job
Insurance Which insurance claims are paid …. the likelihood the policy holder will fight the claim

et al., 2020), procedural governance may not be able to pro- H4c There is a legitimacy penalty associated with algo-
vide the legitimacy dividends that corporations hope. This rithms that use arbitrary or race-based factors, regardless of
leads to two additional hypotheses: the form of procedural governance used.

H4 Any governance regime brings a legitimacy dividend to

algorithmic decision-making relative to no governance, but Study Design
there is a larger legitimacy dividend as procedural govern-
ance becomes more robust. We used the factorial vignette survey methodology to
explore the relative importance of governance, procedure,
H4b Robust procedural governance mechanisms moderate and outcomes on the perceived legitimacy of decisions
the legitimacy penalties associated with bad outcomes. based on a firm’s ADM system. A factorial vignette sur-
vey presents respondents with a series of scenarios where
Scholars have also assumed that good governance several factors are systematically varied in the vignette;
through procedural guardrails could legitimize decisions respondents then judge the scenario using a single rating
made on arbitrary or morally dubious ground (Froomkin task (Jasso, 2006; Wallander, 2009). The general narrative
et al., 2019; Jones, 2017; Kaminski, 2019a, 2019b, 2019c; remains consistent across all vignettes with theoretically
Rahwan, 2018). However, ongoing concerns about algo- important factors (based on the hypotheses) randomly gen-
rithms’ racial biases may call that into question. Using erated with replacement. The vignette factors (explained
racist or arbitrary factors may be so delegitimizing that in Tables 1, 2, 3 and 4) are derived from scholarship and
no amount of procedure could fix it. This leads to our systematically varied in the vignette, thereby offering
final hypothesis:

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 659

Table 2  Outcome factor for vignettes

Decision As Operationalized in the Vignettes: Bad v good outcome

Ads a user sees ads for well-paying jobs

a user sees primarily ads for poorly paying jobs
Music a user hears songs that are totally different from those that they like.
a user hears songs that are similar to those that they like
Hired an applicant is not hired at the grocery store
an applicant is hired at the grocery store
Insurance a policy holder is then told that the claim is fully paid.
a policy holder is then told that the claim will not be paid
Video a user who posted a video then sees that the video is removed.
a user who posted a video then sees that the video is not removed

Table 3  Additional factors included in algorithm which the decision described was legitimate. Each respond-
ent was presented with 30 short vignettes describing an algo-
Substance As operationalized in the vignettes
rithmic decision made by a firm. In general, the vignettes’
Activity their online activity, searches, and purchases narrative had four elements: a decision-maker, a decision
Day the day of the week type, the outcome of the decision, and procedural govern-
Race their race and ethnicity ance associated with the decision-making process, if any.
The elements are illustrated in Table 1 and described in more

Table 4  Governance factor for Governance As operationalized in the vignettes

vignettes
Human only entire vignette is about human decision
appeals that automated decisions can be appealed to a review board
Human governance the organization hired a professional to oversee the automated decision
Audited an independent audit by an external firm is conducted annually to ensure automated
decisions are not biased
Impact assessment the organization did an impact assessment on the automated system
Notification that the organization notifies individuals that a computer program made the decision

a mechanism to test the theoretical relationships in the

detail below. A general outline of the vignettes and samples
hypotheses (Dickel & Graeff, 2018).
of how they were presented to survey respondents is pro-
This methodology allows researchers to measure the rela-
vided in Online Appendix A.
tive importance of the factors of the vignettes affect partici-
The use of the single rating task in the factorial vignette
pants’ attitudes, judgments, or views. The results are theo-
survey methodology supports the inductive measurement of
retically generalizable; the vignettes combine the benefits of
the concept—here, the legitimacy of a company’s decision—
laboratory experiments with those associated with experi-
through the analysis of the factors in the vignette. Through
ments in the field (Oll et al., 2018).6 As noted by Dickel and
the analysis, we measured the relative importance of the
Graeff (2018), factorial vignette surveys provide a more real-
vignette factors in driving the perception of legitimacy of
istic setting and triggers more honest answers than typical
the respondent. This methodology, with a single rating task,
survey instruments (see also Auspurg et al., 2014; Weinberg
has been used to measure the relative importance of vignette
et al., 2014).
factors on just wages (Jasso, 2007), just punishments (Hagan
In this study, decision factors were independently varied
et al., 2008), the trustworthiness of an organization (Pirson
with replacement and the respondents judged the degree to
et al., 2017), and privacy expectations (Martin & Nissen-
baum, 2020). The vignette factors constitute the theoretically
6
This design avoids two types of weaknesses in typical surveys. important constructs that may drive the perception of rating
First, the respondents are forced to use a single rating task while tak- task (trust, fairness, privacy, legitimacy). The methodology
ing into consideration multiple factors at the same time. Second, the is useful to allow the researcher to identify how each factor
design avoids respondent bias where respondents attempt to answer
survey questions to appear more ethical or to please the researcher. drives the perception of the dependent variable.

13
660 K. Martin, A. Waldman

Vignette Template either positive for the individual in the vignette (someone
was hired) or negative for the individual (someone was not
The vignette template supported including different factors hired). The good/bad outcome was contextualized to the
(below) in the vignette. The template was dynamically cre- decision type as in Table 2.
ated as the respondent took the survey. For each survey run,
a different combination of factors was included. The fac- Arbitrariness
tors—decision type, outcome, governance, and arbitrary—
and the levels of each factor are described below in Tables 1, To test hypothesis 3, we included the types of factors both
2, 3 and 4. The vignette template is as follows—the labels arbitrary (day of the week) and discriminatory (their race
are added only for clarity as well as the number of levels or and ethnicity) as well as a null (their online activities).
options within each factor:
Decision: A computer program determines {Deci- Governance
sion—5 levels}. To make the decision, the program
uses {Decision_2}. To test whether the type of governance over an algorithmic
Outcome: Based on the program, {Outcome—2 levels}. decision impacted the degree to which the decision is judged
Arbitrary: It turns out, the decision was partly based legitimate, we included five options for the vignettes: trans-
on {Arbitrary—3 levels} parency, in which the organization notifies individuals that a
Governance: The individual is told {Governance—6 computer program made the decision (Diakopoulos, 2020);
levels} impact assessment, in which the organization completed an
assessment of the impact of the algorithmic process on fair-
ness and privacy (Yam & Skorburg, 2021); audit governance,
Factors in which an external entity completed an annual independent
audit to ensure algorithmic decisions are not biased (Mittel-
Decision Type stadt, 2016); human governance, or so-called “human in the
loop” of the algorithmic process (Elish, 2019); and appeals,
The first set of hypotheses center on the type of decision and its in which the decision can be appealed by the individual to
importance in people’s lives. We included five decision types in an internal review board (Mulligan et al., 2020). Perceptions
our study: which advertisements people see online, which songs of the legitimacy of these decisions were compared to a null
are suggested by music app, which applicants are hired for a in which a human made the decision without the help of an
job, which insurance claims are filled, and which video content algorithm. There are, of course, other possible governance
is taken down by an online platform. These decision types were options. We chose these options because they are the most
chosen based on how pivotal they are in society based on cur- commonly proposed governance mechanisms in the legal
rent literature (O’Neil, 2016; Tufekci, 2015). The degree the studies literature and because they sit on a range from more
decision was deemed pivotal was verified by a pre-test survey robust to lax, providing a proxy for the effect on legitimacy
of 1,024 respondents on Amazon Turk. The respondents were of different types of procedural governance (Citron, 2007;
assigned one of two conditions—either the importance to an Froomkin et al., 2019; Jones, 2017; Katyal, 2019; Loi et al.,
individual or the importance to society—and asked: “Please 2020; Martin, 2018; Rahwan, 2018; Reisman et al., 2018).
rate the degree to which the following decision represents a
critical decision affecting someone’s life” or “Please rate to Vignette Creation
which the following decision represents a critical decision in
society.” The results are in Table A1 in Online Appendix. When the factors in Tables 1–4 are inserted in the vignette
Ads and music were deemed the least pivotal and hiring and template, the respondent then is able to view and rate the
insurance decisions were the most pivotal for both the indi- dynamically created vignettes as shown in the examples
vidual and society based on the pre-test. In the results below, below. The labels (Decision, Outcome, Arbitrary Govern-
highly pivotal decisions were operationalized as vignettes with ance) are only included to explain the methodology; the
hiring or insurance decisions and low pivotal decisions were respondents did not see these labels.
operationalized as vignettes with ads and music decisions.
Example 1:
Outcomes Decision: A computer program determines what songs
are suggested in a playlist (e.g., Spotify). To make the
To test the role of the outcome in the perceived legitimacy decision, the program uses their predicted preferences
of ADM decisions, we varied the outcome from being and interests.

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 661

Table 5  Conditions and sample

Survey 1 2 3 4 5 6 7 8 9

Vignette factors included Decision Decision Decision Decision Decision Decision Decision Decision Decision
Outcome Outcome Outcome Outcome Outcome
Arbitrary Arbitrary Arbitrary Arbitrary Arbitrary
Human Govern Govern Govern Govern
N (respondents) 412 384 417 517 520 469 538 470 500
Vignettes 10,300 9600 10,425 12,925 13,000 11,725 16,140 14,100 15,000
Female 42.3% 41.2% 41.5% 45.7% 45.7% 40.3% 43.9% 41.3% 45.4%
Age over 35 53.8% 53.6% 49.4% 53.9% 57.1% 55.9% 54.8% 60.6% 59.0%

Outcome: Based on the program, a user hears songs that respondents. However, we were concerned that some factors,
are totally different from those that they like. such as the arbitrary factors, could prove to be so important
Arbitrary: It turns out, the decision was partly based as to overwhelm the respondent and cloud how they per-
on their race and ethnicity. ceived the other vignettes. We, therefore, spent more time
Governance: The individual is told an independent audit running nine surveys over four months in order to better
by an external firm is conducted annually to ensure auto- isolate the importance of each type of factor.
mated decisions are not biased. The surveys were run on Amazon Mechanical Turk, a
Example 2: crowdsourcing marketplace where researchers publish a job
Decision: A computer program determines which ads a (“HIT”) for respondents to take a survey. Each respondent
person sees online. To make the decision, the program rated 20–30 vignettes (depending on the condition) taking
uses their predicted preferences and interests. approximately 10 min; U.S. respondents were paid $1.60-
Outcome: Based on the program, a user sees ads for well- $1.80 and were screened for over 95% HIT approval rate.
paying jobs. The survey implementation was designed to minimize a
Arbitrary: It turns out, the decision was partly based number of concerns with samples from Amazon Mechanical
on the day of the week. Turk. First, the factorial vignette survey methodology was
Governance: The individual is told that the organization created to avoid respondent bias in normative judgments—
notifies individuals that a computer program made the namely, where respondents might try to game the system to
decision. appear more ethical or socially desirable. Second, the struc-
ture of the data—in two levels with individuals at the first
Rating Task level and vignette ratings at the second level—supports the
researcher calculating whether respondents ‘clicked through’
For each vignette, respondents were instructed to indicate on a without actually judging the vignette (Coppock, 2018; Daly
slider the degree to which they agreed with the statement: “This & Natarajan, 2015; Martin, 2019; Tucker, 2014).7 Finally,
decision is legitimate.” The left side of the slider indicated the design of the survey is to identify theoretically generaliz-
“Strongly Disagree” and the right side of the sider indicated able results as to the relative importance of factors in driving
“Strongly Agree.” The slider was on a scale of − 100 to + 100 perceptions of legitimacy of AI decisions.8
with the scale not visible to the respondents. The slider option
allows the respondent more freedom to rate the vignette. 7
A few tests allow the researcher to identify whether the respondent
‘clicked through’ including whether the range of responses was small
Conditions and Sample (clustered around − 100, 0, or + 100) by analyzing either the ‘range’
of responses or the standard deviation. These respondents were not
included in the analysis. Previous studies of respondent quality com-
The vignette survey was run under nine conditions described
paring Mechanical Turk to a (more expensive) nationally representa-
in Table 5. This allowed us to isolate the importance of each tive survey through Knowledge Networks showed a significantly
factor by comparing legitimacy averages as the factors were greater proportion of the sample was discarded for clicking through
included. For example, we tested the impact of including for Knowledge Networks (16%) compared to Mechanical Turk (2%)
(Martin & Nissenbaum, 2020).
the outcome by comparing the averages of Survey 2 (only 8
Turk has been used for theoretical generalizability quite success-
decision included) versus Survey 3 (decision and outcome
fully, as in the examination of the relationship between concepts or
included). Normally, a factorial vignette survey analysis ideas (Kang et al., 2014; Martin & Nissenbaum, 2017a; Redmiles
would include a block analysis to isolate which factors, out et al., 2017). In critiques of Turk samples, the Turk results are com-
of many, were dominant in driving the judgement of the pared to phone surveys (Kang et al., 2014) as well as online nation-

13
662 K. Martin, A. Waldman

Table 6  Regression results of legitimacy rating task on vignette factors for each survey run
(a)
Survey 1 Survey 2 Survey 3 Survey 4 Survey 5
β p β p β p β p β p

HighPivotal − 25.68 0.00 − 47.20 0.00 − 10.62 0.00 − 46.59 0.00 − 28.25 0.00
Null = Low Pivotal
BadOutcome − 12.92 0.00 − 43.90 0.00
Null = GoodOutcome
DaySubstance − 31.28 0.00 − 35.31 0.00
RaceSubstance − 40.26 0.00 − 46.57 0.00
Null = activity
AppealGovern 4.54 0.00
AuditGovern − 3.01 0.04
HumanGovern − 6.49 0.00
ImpactAssessGovern − 6.58 0.00
Null = Notice Govern
_cons 27.30 0.00 51.35 0.00 42.10 0.00 38.71 0.00 36.23 0.00
Average − 17.11 25.53 14.14 − 14.05 18.58
N (respondents) 412 384 417 517 520
(b)
Survey 6 Survey 7 Survey 8 Survey 9
β p β p β p β p

HighPivotal − 39.34 0.00 − 23.30 0.00 − 19.06 0.00 − 6.51 0.00

Null = Low Pivotal
BadOutcome 0.00 − 15.19 0.00 − 16.68 0.00 − 27.43 0.00
Null = GoodOutcome
DaySubstance − 27.29 0.00 − 32.22 0.00 − 22.86 0.00
RaceSubstance − 43.69 0.00 − 42.39 0.00 − 39.62 0.00
Null = activity
AppealGovern 1.10 0.47 1.65 0.22 8.33 0.00
AuditGovern 1.40 0.36 2.15 0.11 2.96 0.03
HumanGovern − 2.78 0.07 0.28 0.83 − 0.44 0.74
ImpactAssessGovern − 1.83 0.23 − 2.02 0.13 − 4.61 0.00
Null = Notice Govern
_cons 34.92 0.00 23.62 0.00 29.03 0.00 34.34 0.00
Average − 11.38 − 20.65 − 9.84 19.26
N (respondents) 469 538 470 500

Results

Type of Decision

Footnote 8 (continued) Hypothesis 1 states that as decision importance increases,

ally representative samples; as such the critiques focus on questions individuals’ perception of the legitimacy of using an algo-
of statistical generalizability (Kang et al., 2014; Sharpe Wessling rithm to make the decision decreases. To test hypothesis
et al., 2017). We offer a theoretical examination, where the findings 1, we regressed the rating task, the degree the decision
will support or not support the hypothesized relationships between
vignette factors. Such research seeks the generalizability of ideas
was judged legitimate, on the vignette factors. The results
rather than the generalizability of data patterns within a specific pop- are in Tables 6a and b. The coefficients (β) measure the
ulation (Lynch, 1982). Our results focus on theoretical generalizabil- relative importance of the given vignette factor to the rat-
ity, e.g., whether cause-effect relationships hold or whether concepts ing task, and p is the significance of the finding. Pivotal
are related (Lynch, 1982).

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 663

Impact of Outcome Included Arbitrary Factors

LowPivot alDecision HighPivotal Decision

60.00
Hypothesis 3 states that the use of arbitrary or race-based
factors has a negative impact on perceived legitimacy. The
50.00 50.21
inclusion of arbitrary factors in Survey 4 lowers the legiti-
40.00
macy rating, all else being equal, from 25.53 (Survey 2)
ave Legitimacy Rating

30.00
23.19 to − 14.05 (Survey 4) (t = − 46.06; p < 0.001). In general,
20.00 the lowest average legitimacy rating is found in Surveys
10.00 that include arbitrary factors (Survey 6 − 11.38; Survey 7
0.00 -3.49 − 20.65; Survey 8 − 9.84).
Good Outcome Bad Outcome
-10.00
-9.31 Governance and Legitimacy
-20.00

Hypothesis 4a stated that the inclusion of any governance

Fig. 2  Impact of good and bad outcome on pivotal decisions
regime has a positive impact on perceived legitimacy of
algorithmic decision-making relative to no governance,
decisions are those decisions deemed more important in but there is a larger legitimacy dividend as procedural
the pre-survey test. For each survey run, more pivotal governance becomes more robust. To test Hypothesis 4a,
decisions lowered the legitimacy of the decision with we examine the regression results in Tables 6a and 6b
β < 0 and p < 0.001. For example, in Survey 2, which only for the surveys with governance included (Surveys 5, 6,
included the decision type in the vignette, including high 8, and 9). Here we see that offering an appeal process
pivotal decisions lowered the legitimacy rating by − 47.20 provides a legitimacy dividend for decisions with no out-
(p < 0.001). This finding held across each survey and com- come or arbitrary factors included (Survey 5). The more
bination of factors. Less pivotal decisions made by AI are robust governance mechanism, to allow an appeal, posi-
seen as more legitimate than more pivotal decisions. tively impacts perceived legitimacy ( β = 4.54, p < 0.01).
However, alternative governance mechanisms such as an
Outcomes impact assessment, including a human in the loop, and
having the program audited, lower the legitimacy of the
Hypothesis 2 states a good outcome is associated, an decision compared to mere notice. This was a surprise and
increase in the perceived legitimacy of the decision. To counter to the hypothesis.
test Hypothesis 2, we regressed the dependent variable,
the degree to which the decision was judged legitimate, on Outcomes and Governance
the vignette factors for Survey 3 (with only the decision
and outcome included). We find a bad outcome creates a Hypothesis 4b states that robust procedural governance
legitimate penalty of − 43.90 (p < 0.001) compared to a mechanisms moderate the legitimacy penalties associated
good outcome. In other words, a good outcome has a posi- with bad outcomes. In other words, offering an appeal to
tive impact on perceived legitimacy. This result held even the algorithmic decision legitimizes even a bad outcome.
when additional factors are added in Surveys 7, 8, and 9 We tested hypothesis 4b by comparing the legitimacy
in Table 6b. rating for decisions with an outcome across governance
We also tested if the legitimacy dividend associated with mechanisms. As shown in Fig. 3, we found that for bad
a good outcome increases as decision importance increases. outcomes, the use of an appeal process does provide a
Figure 2 illustrates the legitimacy ratings for both good and legitimacy dividend (t = − 12.53; p < 0.0001); however,
bad outcomes. The benefit of a good outcome over a bad the use of an appeal process had no effect on the perceived
outcome is larger for a low pivotal decision (Good = 50.21; legitimacy of algorithmic decisions with good outcomes
Bad = − 9.31; (t = 35.9793; p < 0.001)). compared to a high (t = − 1.74; p = 0.04). Further, the dividend did not ‘make
pivotal decision (good = 23.19; bad = − 3.49; (t = 14.53; up’ for the penalty of the bad outcome. The average legiti-
p < 0.001)). In Fig. 2, the steeper slope represents a greater macy rating for a bad outcome even with an appeal process
legitimacy penalty for bad outcomes for low pivotal deci- is still lower than that of a good outcome (with or without
sions as compared to high pivotal decisions (χ2 = 196.02; an appeal process included) —however the appeal process
p < 0.001). This could be because respondents expect a good does close the gap between good and bad outcome legiti-
outcome with low pivotal decisions and then penalize the macy as in Fig. 3.
surprise bad outcome more.

13
664 K. Martin, A. Waldman

Fig. 3  Impact of adding a robust Impact of Appeal on Good and Bad Outcomes
governance mechanism (appeal)
to AI decisions with good and Good Outcome Bad Outcome
bad outcomes
60

50

40
35.83

Ave Legitimacy Rating

30 33.23

20
17.89

10

0
Decision Only Appeal
-10
-4.07

-20

Fig. 4  Impact of adding a robust Impact of Appeal on Race and Day Arbitrary Factors
governance mechanism (appeal)
to algorithmic decisions with Day Race
race and day arbitrary factors
60.00

50.00

40.00

30.00
Ave Legitimacy Rating

20.00

10.00

0.00
Decision Only Appeal
-10.00
-13.12
-20.00
-21.87
-30.00
-31.94
-34.09
-40.00

Arbitrary Factors and Governance shown to have the greatest legitimacy dividend, on deci-
sions with either race or day of the week. The impact of
Hypothesis 4c states that there should be a legitimacy including an appeal on decisions that used the day of the
penalty associated with algorithms that use arbitrary or week was modest but significant (t = 0.3.481; p < 0.001).
race-based factors, regardless of the form of procedural The impact of including an appeal for decisions utilizing
governance used. To test Hypothesis 4c, we compared race was not significant (t = 0.863; p = 0.19). The legiti-
the average legitimacy rating for the inclusion of race and macy dividend of including an appeal process improves
arbitrary factors both with and without governance mecha- the perceived legitimacy of decisions that include arbitrary
nisms. Figure 4 shows the impact of including the appeal factors such as the day of the week but not when including
of the decision as a governance mechanism, previously unjust factors such as race.

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 665

Fig. 5  Impact of adding a race Impact of Race on Good/Bad Outcomes

and day arbitrary factors to
algorithmic decisions with good Bad Outcome Good Outcome
and bad outcomes 60.00

40.00
33.23

20.00

Ave Legitimacy Rating

0.00
Outcome Outcome w/Race
-4.07
-20.00

-34.18
-40.00
-43.59

-60.00

Table 7  Summary of findings

Hypothesis Finding

H1 As decision importance increases, individuals’ perception of the Less pivotal decisions made by AI are seen as more legitimate than
legitimacy of using an algorithm to make the decision decreases more pivotal decisions
H2 A good outcome is associated with an increase in the perceived We find a bad outcome is a legitimate penalty compared to a good
legitimacy of the decision outcome. In other words, a good outcome has a positive impact on
perceived legitimacy
H3 The use of arbitrary or race-based factors has a negative impact on The inclusion of arbitrary factors in Survey 4 lowers the legitimacy
perceived legitimacy rating, all else being equal, from 25.53 (Survey 2) to − 14.05 (Survey
4) (t = − 46.06; p < 0.001)
H4a Any governance regime brings a legitimacy dividend to algorith- The more robust governance mechanism, to allow an appeal, positively
mic decision-making relative to no governance, but there is a larger impacts perceived legitimacy. However, alternative governance
legitimacy dividend as procedural governance becomes more robust mechanisms such as an impact assessment, including a human in the
loop, and having the program audited, lower the legitimacy of the
decision compared to mere notice
This was a surprise and counter to the hypothesis
H4b robust procedural governance mechanisms moderate the legiti- For bad outcomes, the use of an appeal process does provide a legiti-
macy penalties associated with bad outcomes macy dividend. However, the use of an appeal process had no effect
on the legitimacy of algorithmic decisions with good outcomes. The
average legitimacy rating for a bad outcome even with an appeal
process is still lower than that of a good outcome
H4c There is a legitimacy penalty associated with algorithms that use The legitimacy dividend of including an appeal process improves the
arbitrary or race-based factors, regardless of the form of procedural legitimacy of decisions including arbitrary factors such as the day of
governance used the week but not when including unjust factors such as race

To better understand the legitimacy penalty for hav- only slightly for good outcomes and race (from − 34.18 to
ing arbitrary rationales (race or day of the week) with − 24.59 with an appeal; t = 3.14; p < 0.001) and slightly
algorithmic decisions, we compared the impact of includ- for bad outcomes and race (from − 43.59 to − 33.17 with
ing the arbitrary factor of race on decisions with both an appeal; t = 3.32; p < 0.001). Including race as an arbi-
good and bad outcomes. The results, in Fig. 5, show that trary and unjust factor in an algorithmic decision stub-
including race as a factor is a legitimacy penalty for bad bornly delegitimizes the decision regardless of a good
outcomes (t = 26.40; p < 0.001). In addition, any legiti- or bad outcome and regardless of including an appeal
macy dividends of a good outcome are erased when the (Table 7).
decision included race as a factor (t = 50.52; p < 0.001).
Adding an appeal for governance improves the legitimacy

13
666 K. Martin, A. Waldman

Discussion and Conclusion Future work could similarly examine how specific design
decisions of ADM (including the use of AI and different
This paper explores the conceptual antecedents to the types of machine learning) impact business ethics outcomes
perceived legitimacy of algorithmic decisions. Building such as justice, fairness, trust, and legitimacy. For example,
on cross-disciplinary empirical measurements of deci- scholarship could investigate how data, governance, out-
sion, organizational, and institutional legitimacy (Els- comes, and the use of AI in general impact employee per-
bach, 1994; Finch et al., 2012; Jahn et al., 2020; Tyler, ceptions of fairness. Trust in a firm, and trust factors such as
2006/1990; Badas, 2019) and theoretical accounts of ability, benevolence, and integrity, may be impacted by the
legitimacy (Palazzo & Scherer, 2006; Suchman, 1995), use of particular data, the degree to which the firm allows
we studied how factors of ADM—the type of decision, the decisions to be contested, and the outcome of the algorith-
outcomes, the rationale, and the governance—impact the mic decision. Scholars could leverage work on the purpose
perception of decisional legitimacy. of the firm to help AI ethics scholars understand how ADM
Empirically, we used factorial vignette methodology outcomes, designed by computer scientists, may undermine
and made several hypotheses. We show that perceived legitimate purposes of the firm. Rather than mistakenly see-
legitimacy varies inversely with decision importance. We ing the introduction of algorithmic decision making as fun-
hypothesized that the robustness of procedural govern- damentally changing how we assess ethical decisions, busi-
ance over algorithmic decisions made by firms would vary ness ethics should leverage well researched ethical concepts
positively and directly with legitimacy. The data suggest to illustrate how AI does not change the nature of corporate
something more nuanced. The only form of procedural responsibility. Work in AI Ethics within business ethics can
governance that carries a legitimacy dividend is the most (and should) start to connect the known moral implications
robust that we studied: a formal appeal to a human deci- of AI with our existing frameworks in business ethics in
sion maker. Those more commonly in practice or in legis- order to have concrete implications for practice as well as
lative proposals today—notice and impact assessments— contribute to business ethics theory.
either have no impact or carry a legitimacy penalty.
We hypothesized that outcomes, defined as whether an AI Ethics Research
individual agrees with a decision or the decision is good
for the individual, would matter more for legitimacy than Although work on fairness, accountability, transparency, and
procedural governance. This was mostly true, although explainability (FATE) has flourished in the past few years,
the data suggested that there was no legitimacy dividend more work on the perceptions of ADM—parallel to the work
for positive outcomes for decisions of low importance. on perceptions of human-focused decisions—should con-
We also hypothesized that the use of morally question- tinue. For example, scholars have studied whether ADM is
able or arbitrary factors in making decisions would carry perceived as authentic (Jago, 2019) or fair (Araujo et al.,
legitimacy penalties. Our study suggested that this was 2020; Nagtegaal, 2021; Newman et al., 2020). More work
the case, erasing all legitimacy dividends associated with should be done taking existing work on ethical decision
robust governance or positive outcomes. making in business ethics and management to investigate
whether the factors important to human decisions maintain
the same relative importance to judgements about ADM.
Algorithms and Business Ethics This could then help guide AI ethics research on the design
of algorithmic decision systems.
Scholars in business ethics have critically examined ADM This paper contributes to AI ethics scholarship by extend-
within specific applications such as accounting (Gunz & ing the moral implications that should be a concern to firms.
Thorne, 2020; Munoko et al., 2020), financial services While firms have been told to address transparency and
(Arthur & Owen, 2019) and HR (Leicht-Deobald, et al., accountability as well as discrimination, this study suggests
2019), and have identified specific moral implications such that firms should also be concerned about legitimacy of the
as social-media addiction (Bhargava & Velasquez, 2020), decision and the organization as being impacted by how the
personalized pricing (Seele et al., 2019; Steinberg, 2020), ADM is designed. Future work could continue to examine
gamification (Kim, 2018), and accountability (Martin, design decisions and their impact on different dependent
2019). Previous work has also critically examined the details variables including legitimacy, fairness, and trust in addition
of algorithms in ride sharing can create gender discrimi- to how the design decisions impact users and subjects of the
nation in ratings (Greenwood et al., 2020). We extend this ADM. Extending the research to include how the firm itself
line of AI Ethics scholarship within business ethics to the is impacted by the value-laden decisions may resonate with
specific examination of how ADM will impact legitimacy. managers and business scholarship.

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 667

Legitimacy (2014) suggest that public exposure to performative sym-

bols of authority insulates authoritative institutions from
The results of this study add to our understanding of issue, the worst legitimacy penalties associated with ideological
organizational, and institutional legitimacy. Business, man- or policy disagreement. Our study suggests that this may
agement, and sociolegal scholars studying legitimacy have only be true for decisions of high importance, but the extent,
shown that factors like communication (Elsbach, 1994), if any, of the mediation of decision importance would have
motives (Jahn et al., 2020), procedural fairness (Tyler, to be studied empirically.
2006/1990), ideology (Badas, 2019; Finch et al., 2012), Finally, this research highlights the dangers to legitimacy
and agreement or disagreement with outcomes (Gibson & for any firm that makes decisions via ADM systems that rely
Caldeira, 2009) impact legitimacy. Our study not only adds on race, proxies for race, or seemingly arbitrary factors that
additional factors—decision importance (H1), outcomes individuals do not see as related to the decision itself. Given
(H2a) and the use of arbitrary or morally dubious variables how difficult it is to remove racial discrimination from AI
(H3)—but also tests the interaction among the factors (H4a, systems trained on data that is itself the product of systemic
H4b, and H4c). In so doing, we add nuance to the exist- and institutional discrimination, the algorithmic account-
ing literature. For example, Tyler (2006/1990) suggests that ability literature is right to increasingly focus on issues of
people tend to think that authorities’ decisions are legitimate structural fairness rather than mere remediation through
when the decision-making process is fair and when individu- governance or better data (Pasquale, 2019).
als have the opportunity to be heard through procedural due
process. Our findings are partly in line with this research, but Public Policy
we show that not all procedural governance mechanisms are
the same. Only an appeal to a human authority is generally Unfortunately, current policy proposals lag behind this the
capable of legitimizing algorithmic systems, all other factors “second wave” of the algorithmic accountability literature
held constant. It is possible that other procedural mecha- (Pasquale, 2019). Article 22 of the European Union’s Gen-
nisms do not provide the kind of robust guarantees of fair- eral Data Protection Regulation places restrictions on private
ness that individuals may associate with the right to appeal. firms that employ algorithms without human intervention to
In work studying popular perceptions of the legitimacy of make decisions that have significant effects on individuals.
the United States Supreme Court, Badas (2019) and Gibson Firms can do so, but only if they adopt “suitable measures to
et al. (2005) suggest that ideological agreement or disagree- safeguard the data subject’s rights and freedoms and legiti-
ment with an authority’s decision is a significant driver of mate interests,” which might include “at least the right to
legitimacy, regardless of fair process. Finch et al. (2012) obtain human intervention …, to express his or her point
showed that ideological views about environmental protec- of view and to contest the decision (General Data Protec-
tion and economic interests helped determine perceptions tion Regulation, 2018). Other provisions of the GDPR
of legitimacy of a high-pollution industry. Our study tests may also impact algorithmic decision-making systems, as
this relationship in the context of private firm decisions well (Edwards & Veale, 2017), but they all share one pri-
and finds, with additional nuance, that outcomes do affect mary common thread: they rely on procedural due process
perceptions of legitimacy. Positive outcomes confer legiti- mechanisms to protect individual rights and ensure fairness
macy dividends for decisions of high importance, but not for (Waldman, 2021). Our study calls into question the capac-
decisions of low importance, and negative outcomes create ity of such procedural governance to legitimize algorithmic
larger legitimacy penalties for less important decisions than decisions in the eyes of the public, lending credibility to
highly pivotal decisions. This is in line with current litera- arguments from critical scholars that current law needs to
ture, which suggests that individuals credit institutions that be more robust or, perhaps, disused in certain circumstances
make the “right” decisions as they see them without seeking and for certain purposes all together (Pasquale, 2015, 2018,
explanations (Lodge & Taber, 2013). Although our study did 2021). Although our study did not explore perceptions of
not examine the rationales for legitimacy penalties for out- the legitimacy of firms that stop using discriminatory algo-
comes mediates by decision importance, the variance may rithms, the strong legitimacy penalties associated with use
be because individuals expect that such decisions should be implies the possibility of legitimacy dividends for changing
easy to get right and, as such, tend to react negatively when course.
computer get them wrong. Of course, even the procedural guardrails of the GDPR
These findings contribute to our understanding of posi- do not exist in the United States. Firms are stepping into
tivity theory (Gibson & Caldeira, 2009). Positivity theory uncharted waters in the US, which lacks a federal agency
attempts to explain why individuals punish institutions less for algorithms and neither requires notice nor impact assess-
for decisions with which they disagree than they reward ments for algorithmic decisions (Crawford & Schultz, 2013).
them for decisions with which they agree. Gibson et al. Our study suggests that US policymakers should be careful

13
668 K. Martin, A. Waldman

following the GDPR’s procedural governance model when, Ashforth, B., & Gibbs, B. (1990). The double-edge of organizational
if ever, they start regulating private firm use of algorithmic legitimization. Organizational Science, 1, 177–194.
Badas, A. (2019). Policy disagreement and judicial legitimacy: Evi-
decision-making systems. If, as we show, that only the most dence from the 1937 Court-Packing plan. Journal of Legal Stud-
robust procedures, such as offering an appeal akin to Mul- ies, 48, 377–408.
ligan et al., (2020), confer legitimacy benefits, that outcomes Barocas, S., & Selbst, A. D. (2016). Big data’s disparate impact. Cali-
are far more powerful drivers of perceptions of legitimacy, fornia Law Review, 104, 671–732.
Bartels, B., & Johnston, C. (2013). On the ideological foundations of
and that almost all kinds of algorithmic decisions are viewed Supreme Court legitimacy in the American public. American
as illegitimate when they use race-based or arbitrary fac- Journal of Political Science, 57, 184–199.
tors, then policymakers should consider more substantive Beelitz, A., & Merkl-Davies, D. (2012). Using discourse to restore
limits on algorithmic inputs and uses rather than procedural organizational legitimacy: ‘CEO-speak’ after an incident in a
German nuclear power plant. Journal of Business Ethics, 108(1),
safeguards. 101–120.
Bhargava, V. R., & Velasquez, M. (2020). Ethics of the attention econ-
Supplementary Information The online version contains supplemen- omy: The problem of social media addiction. Business Ethics
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 10551-0​ 21-0​ 5032-7. Quarterly, 31(3), 1–39.
Boyd, J. (2000). Actional legitimation: No crisis necessary. Journal of
Author Contributions All authors contributed to this manuscript. Public Relations Research, 4, 341–353.
Bronn, P., & Vidaver-Cohen, D. (2009). Corporate motives for social
initiative: Legitimacy, sustainability, or the Bottom Line. Journal
Declarations of Business Ethics, 87, 91–109.
Burrell, J. (2016). How the machine ‘thinks’: Understanding opacity
Conflict of interest The authors received funding from ND-Technolo- in machine learning algorithms. Big Data & Society, 3(1), 1–12.
gy Ethics Center for this paper and have no known conflicts of inter- Caliskan, A., Bryson, J., & Narayanan, A. (2017). Semantics derived
est. The authors have no relevant financial or non-financial interests automatically from language corpora contain human-like biases.
to disclose. Science, 356, 183–186.
Calo, R. (2017). Artificial intelligence policy: A primer and roadmap.
Informed Consent Informed consent was obtained from all individual University of California, Davis Law Review, 51, 399–435.
participants included in the study. Castello, I., & Lozano, J. (2011). Searching for new forms of legiti-
macy through corporate responsibility rhetoric. Journal of Busi-
Research Involving Human and Animal Rights This paper does include ness Ethics, 100(1), 11–29.
research involving humans and/or animals. The authors received IRB Chen, S., Zhang, J. A., Gao, H., Yang, Z., & Mather, D. (2020). Trust
approval at the University of Notre Dame (#20-09-6227). erosion during industry-wide crises: the central role of consumer
legitimacy judgement. Journal of Business Ethics, 175(4), 1–22.
Cheng, E. (2013). Being pragmatic about forensic linguistics. Journal
Open Access This article is licensed under a Creative Commons Attri- of Law and Policy, 21, 541–550.
bution 4.0 International License, which permits use, sharing, adapta- Christenson, D., & Glick, D. (2015). Chief Justice Roberts’ health care
tion, distribution and reproduction in any medium or format, as long decision disrobed: The microfoundations of the Supreme Court’s
as you give appropriate credit to the original author(s) and the source, legitimacy. American Journal of Political Science, 59, 403–418.
provide a link to the Creative Commons licence, and indicate if changes Chung, J., Berger, B., & DeCoster, J. (2014). Developing measurement
were made. The images or other third party material in this article are scales of organizational and issue legitimacy: A case of direct-
included in the article's Creative Commons licence, unless indicated to-consumer advertising in the pharmaceutical industry. Journal
otherwise in a credit line to the material. If material is not included in of Business Ethics, 137, 405–413.
the article's Creative Commons licence and your intended use is not Citron, D. (2007). Technological due process. Washington University
permitted by statutory regulation or exceeds the permitted use, you will Law Review, 85(6), 1249–1313.
need to obtain permission directly from the copyright holder. To view a Citron, D., & Pasquale, F. (2014). The scored society: Due process
copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. for automated predictions. Washington Law Review, 89, 1–33.
Coombs, W. (1992). The failure of the task force on food assistance: A
case study of the role of legitimacy in issue management. Journal
of Public Relations Research, 4(2), 101–122.
References Coppock, A. (2018). Generalizing from survey experiments conducted
on mechanical turk: A replication approach. Political Science
Ajunwa, I. (2020). The paradox of automation as anti-bias intervention. Research and Methods, 7(3), 613–628.
Cardozo Law Review, 41, 1671–1742. Crawford, K., & Schultz, J. (2013). Big data and due process: Toward
Ali, M. (2017). Stakeholder salience for stakeholder firms: An attempt a framework to redress predictive privacy harms. Boston College
to reframe an important heuristic device. Journal of Business Law Review, 55, 93–128.
Ethics, 144(1), 153–168. Daly, T., & Natarajan, R. (2015). Swapping bricks for clicks: Crowd-
Araujo, T., Helberger, N., Kruikemeier, S., & de Vreese, C. (2020). In sourcing longitudinal data on Amazon Turk. Journal of Business
AI we trust? Perceptions about automated decision-making by Research, 68(12), 2603–2609.
artificial intelligence. AI & Society, 35, 611–623. Danaher, J., Hogan, M. J., Noone, C., Kennedy, R., Behan, A., De Paor,
Arthur, K. N. A., & Owen, R. (2019). A Micro-ethnographic study of A., et al. (2017). Algorithmic governance: Developing a research
big data-based innovation in the financial services sector: Gov- agenda through the power of collective intelligence. Big Data &
ernance, ethics and organisational practices. Journal of Business Society, 4(2), 2053951717726554.
Ethics, 160(2), 363–375. Deephouse, D., Bundy, J., Tost, L., & Suchman, M. (2017). Organiza-
tional legitimacy: Six key questions. In R. Greenwood, C. Oliver,

13
Are Algorithmic Decisions Legitimate? The Effect of Process and Outcomes on Perceptions of… 669

T. Lawrence, & R. Meyer (Eds.), The SAGE handbook of organi- Gunz, S., & Thorne, L. (2020). Thematic symposium: The impact of
zational institutionalism (2nd ed.). Oxford: Sage. technology on ethics, professionalism and judgement in account-
Diakopoulos, N. (2020). Transparency. Oxford University Press. ing. Journal of Business Ethics, 167, 153–155.
Dickel, P., & Graeff, P. (2018). Entrepreneurs’ propensity for cor- Hagan, J., Ferrales, G., & Jasso, G. (2008). How law rules: Torture, ter-
ruption: A vignette-based factorial survey. Journal of Business ror, and the normative judgments of Iraqi judges. Law & Society
Research, 89, 77–86. Review, 42(3), 605–644.
Dodge, M., & Kitchin, R. (2007). The automatic management of driv- Hu, M. (2016). Big data blacklisting. Florida Law Review, 67(5), 1735.
ers and driving spaces. Geoforum, 38(2), 264–275. Jago, A. (2019). Algorithms and authenticity. Academy of Management
Dowling, J., & Pfeffer, J. (1975). Organizational legitimacy: Social Discoveries, 5, 38–56.
values and organizational behavior. Pacific Sociological Review, Jahn, J., Eichhorn, M., & Bruhl, R. (2020). How do individuals judge
18(1), 122–136. organizational legitimacy? Effects of attributed motives and cred-
Dressel, J., & Farid, H. (2018). The accuracy, fairness, and limits of ibility on organizational legitimacy. Business & Society, 59(3),
predicting recidivism. Science Advances, 4(1), 1–5. 545–576.
Du, S., & Vieira, E. (2012). Striving for legitimacy through corporate Jasso, G. (2006). Factorial survey methods for studying beliefs and
social responsibility: Insights from oil companies. Journal of judgments. Sociological Methods & Research, 34(3), 334–423.
Business Ethics, 110(4), 413–427. Jasso, G. (2007). Studying justice: Measurement, estimation, and
Easton, D. (1965). A systems analysis of political life. Wiley. analysis of the actual reward and the just reward. IZA Discus-
Edwards, L., & Veale, M. (2017). Slave to the algorithm? Why a ‘right sion Papers.
to an explanation’ is probably not the remedy you are looking for. Johnson, G. (n.d.). Are algorithms value-free? Feminist theoretical vir-
Duke Law and Technology Review, 16, 18–84. tues in machine learning. Journal Moral Philosophy.
Eesley, C., & Lenox, M. J. (2006). Firm responses to secondary stake- Johnson, G. (2020). Algorithmic bias: On the implicit biases of social
holder action. Strategic Management Journal, 27(8), 765–781. technology. Synthese, 198(10), 9941–9961.
Elish, M. C. (2019). Moral crumple zones: Cautionary tales in human– Jones, M. L. (2017). The right to a human in the loop: Political con-
robot interaction (pre-print). Engaging Science, Technology, and structions of computer automation and personhood. Social Stud-
Society. https://​doi.​org/​10.​2139/​ssrn.​27572​36. ies of Science, 47, 216–239.
Ellis, L., & Diamond, S. S. (2003). Race, diversity, and jury composi- Kaina, V. (2008). Legitimacy, trust and procedural fairness: Remarks
tion: Battering and bolstering legitimacy. The Chicago-Kent Law on Marcia Grimes’ study. European Journal of Political
Review, 78, 1033. Research, 47(4), 510–521.
Elsbach, K. (1994). Managing organizational legitimacy in the Califor- Kaminski, M. E. (2019b). Binary governance: Lessons from the
nia cattle industry: The construction and effectiveness of verbal GDPR’s approach to algorithmic accountability. Southern Cali-
accounts. Administrative Science Quarterly, 39(1), 57–88. fornia Law Review, 92(6), 1529–1616.
Eubanks, V. (2018). Automating inequality: How high-tech tools pro- Kaminski, M. (2019a). The right to explanation, explained. Berkeley
file, police, and punish the poor. Martin’s Press. Technology Law Journal, 34, 189–218.
Finch, D., Deephouse, D., & Varella, P. (2012). Examining an indi- Kaminski, M. E. (2019c). The right to explanation, explained. Berkeley
vidual’s legitimacy judgment using the value-attitude system: Technology Law Journal, 34, 189.
The role of environmental and economic values and source cred- Kang, R., Brown, S., Dabbish, L., & Kiesler, S. B. (2014). Privacy
ibility. Journal of Business Ethics, 127(2), 265–281. Attitudes of Mechanical Turk Workers and the US Public. In
de Fine Licht, K., & de Fine Licht, J. (2020). Artificial intelligence, SOUPS (pp. 37–49). Presented at the SOUPS.
transparency, and public decision-making: Why explanations are Katyal, S. (2019). Private accountability in the age of artificial intel-
key when trying to produce perceived legitimacy. AI & Society, ligence. University of California, Los Angeles Law Review, 66,
35, 917–926. 54–141.
Freeman, R. E. (1994). The politics of stakeholder theory: Some future Kim, T. W. (2018). Gamification of labor and the charge of exploita-
directions. Business Ethics Quarterly, 4(4), 409–421. tion. Journal of Business Ethics, 152(1), 27–39.
Freeman, R. E., Martin, K., & Parmar, B. (2007). Stakeholder capital- Leicht-Deobald, U., Busch, T., Schank, C., Weibel, A., Schafheitle,
ism. Journal of Business Ethics, 74(4), 303–314. S., Wildhaber, I., & Kasper, G. (2019). The challenges of algo-
Freeman, R. E., Parmar, B. L., & Martin, K. (2020). The power of and: rithm-based HR decision-making for personal integrity. Journal
Responsible business without trade-offs. Columbia University of Business Ethics, 160(2), 377–392.
Press. Lodge, M., & Taber, C. (2013). The rationalizing voter. Cambridge
Froomkin, A. M., Kerr, I., & Pineau, J. (2019). When AIs outperform University Press.
doctors: Confronting the challenges of a tort-Induced over-reli- Loi, M., Ferrario, A., & Vigano, E. (2020). Transparency as design
ance on machine learning. Arizona Law Review, 61, 33–99. publicity: Explaining and justifying inscrutable algorithms.
General Data Protection Regulation. (2018). Regulation 2016/679, of Ethics and Information Technology. https://​doi.​org/​10.​1007/​
the European Parliament and of the Council of 27 April 2016 on s10676-​020-​09564-w.
the Protection of Natural Persons with Regard to the Processing Lünich, M., & Kieslich, K. (2021). Using automated decision-making
of Personal Data and on the Free Movement of Such Data, and (ADM) to allocate Covid-19 vaccinations? Exploring the roles
Repealing Directive 95/46/EC, May 4, 2016, 2016 O.J. (L 119). of trust and social group preference on the legitimacy of ADM
Gibson, J., & Caldeira, G. (2009). Citizens, courts, and confirmation: vs. human decision-making. arXiv preprint. arXiv:​2107.​08946.
Positivity theory and the judgments of the American people. Lynch, J. G., Jr. (1982). On the external validity of experiments in con-
Princeton University Press. sumer research. Journal of Consumer Research, 9(3), 225–239.
Gibson, J., Caldeira, G., & Spence, L. K. (2005). Why do people Martin, K. (2018). Ethical implications and accountability of algo-
accept public policies they oppose? Testing legitimacy theory rithms. Journal of Business Ethics, 160, 1–16.
with a survey-based experiment. Political Research Quarterly, Martin, K. (2019). Designing ethical algorithms. MIS Quarterly Execu-
58, 187–201. tive, 18(2), 129–142.
Gibson, J., Lodge, M., & Woodson, B. (2014). Losing but accepting: Martin, K. (2022). Algorithmic bias and corporate responsibility:
Legitimacy, positivity theory, and the symbols of judicial author- How companies hide behind the false veil of the technological
ity. Law & Society Review, 48(4), 837–866.

13
670 K. Martin, A. Waldman

imperative. In K. Martin (Ed.), Ethics of data and analytics. Tay- Salganik, M., et al. (2020). Measuring the predictability of life outcomes
lor & Francis. with a scientific mass collaboration. PNAS, 117(15), 8398–8403.
Martin, K., & Nissenbaum, H. (2017a). Privacy interests in public Santana, A. (2012). Three elements of stakeholder legitimacy. Journal of
records: An empirical investigation. Harvard Journal of Law and Business Ethics, 105(2), 257–265.
Technology, 31(1), 111–143. Seele, P., Dierksmeier, C., Hofstetter, R., & Schultz, M. D. (2019). Map-
Martin, K., & Nissenbaum, H. (2020). What is it about location? Berkeley ping the ethicality of algorithmic pricing: A review of dynamic and
Technology Law Journal, 35(1), 251. personalized pricing. Journal of Business Ethics, 170, 697–719.
Mittelstadt, B. (2016). Automation, algorithms, and politics| auditing Sharpe Wessling, K., Huber, J., & Netzer, O. (2017). MTurk character
for transparency in content personalization systems. International misrepresentation: Assessment and solutions. Journal of Consumer
Journal of Communication, 10, 12. Research, 44(1), 211–230.
Mulligan, D. K., Kluttz, D., & Kohli, N. (2020). Shaping our tools: Con- Smith, A. (2018). Attitudes toward algorithmic decision-making. Pew
testability as a means to promote responsible algorithmic decision Research Center.
making in the professions. In K. Werbach (Ed.), After the digital Steinberg, E. (2020). Big data and personalized pricing. Business Ethics
Tornado. Cambridge: Cambridge University Press. Quarterly, 30(1), 97–117.
Munoko, I., Brown-Liburd, H. L., & Vasarhelyi, M. (2020). The ethical Suchman, M. (1995). Managing legitimacy: Strategic and institutional
implications of using artificial intelligence in auditing. Journal of approaches. Academy of Management Review, 20(3), 571–610.
Business Ethics, 167(2), 209–234. Suddaby, R., Bitektine, A., & Haack, P. (2017). Legitimacy. Academic of
Nagtegaal, R. (2021). The impact of using algorithms for managerial Management Annals, 11(1), 451–478.
decisions on public employees’ procedural justice. Government Sunshine, J., & Tyler, T. (2003). The role of procedural justice and legiti-
Information Quarterly, 38(1), 101536. macy in shaping public support for policing. Law & Society Review,
Newman, D., Fast, N., & Harmon, D. (2020). When eliminating bias isn’t 37, 513–547.
fair: Algorithmic reductionism and procedural justice in human Tucker, C. (2014). The reach and persuasiveness of viral video ads. Mar-
resource decisions. Organizational Behavior & Human Decision keting Science, 34(2), 281–296.
Processes, 160, 149–167. Tufekci, Z. (2015). Algorithmic harms beyond Facebook and Google:
Noble, S. (2018). Algorithms of oppression. N. Y. U. Press. Emergent challenges of computational agency. Colorado Journal
Nozick, R. (1974). Anarchy, state, and utopia (Vol. 5038). Basic Books. of Telecommunications and High Technology, 13, 203–218.
O’Neil, C. (2016). Weapons of math destruction: How big data increases Tyler, T. (2006/1990). Why people obey the law. Princeton University
inequality and threatens democracy. Broadway Books. Press.
Oll, J., Hahn, R., Reimsbach, D., & Kotzian, P. (2018). Tackling complex- Tyler, T. (1994). Governing amid diversity: The effect of fair decision-
ity in business and society research: The methodological and the- making procedures on the legitimacy of government. Law & Soci-
matic potential of factorial surveys. Business & Society, 57, 26–59. ety Review, 28(3), 809–831.
Palazzo, G., & Scherer, A. (2006). Corporate legitimacy as delibera- Tyler, T., & Huo, Y. J. (2002). Trust in the Law: Encouraging Public
tion: A communicative framework. Journal of Business Ethics, Cooperation with the Police and Courts. New York: Russell Sage
66, 71–88. Foundation.
Pasquale, F. (2015). Black box society. Harvard University Press. Vincent, J. (2020). Facebook is now using AI to sort content for quicker
Pasquale, F. (2018). When machine learning is facially invalid. Com- moderation. The Verge. https://​www.​theve​rge.​com/​2020/​11/​13/​
munications of the ACM, 61(9), 25–27. 21562​596/​faceb​ook-​ai-​moder​ation.
Pasquale, F. (2019). The second wave of algorithmic accountability. Law Waldman, A. (2021). The new privacy law. U.C. Davis Law Review
and Political Economy Blog. https://​lpebl​og.​org/​2019/​11/​25/​the-​ Online, 55, 19–42.
second-​wave-​of-​algor​ithmic-​accou​ntabi​lity/. Wallander, L. (2009). 25 years of factorial surveys in sociology: A review.
Phillips, R. A. (1997). Stakeholder theory and a principle of fairness. Social Science Research, 38(3), 505–520.
Business Ethics Quarterly, 7(1), 51–66. Walzer, M. (2008). Spheres of justice: A defense of pluralism and equal-
Pirson, M., Martin, K., & Parmar, B. (2017). Formation of stakeholder ity. Basic Books.
trust in business and the role of personal values. Journal of Business Wicks, A. C., Gilbert, D. R., Jr., & Freeman, R. E. (1994). A feminist
Ethics, 145(1), 1–20. https://​doi.​org/​10.​1007/​s10551-​015-​2839-2. reinterpretation of the stakeholder concept. Business Ethics Quar-
Rahwan, I. (2018). Society-in-the-loop: Programming the algorithmic terly, 4(4), 475–497.
social contract. Ethics and Information Technology, 21, 5–14. Winner, L. (1980). Do artifacts have politics? Daedalus, 109(1), 121–136.
Redmiles, E. M., Kross, S., Pradhan, A., & Mazurek, M. L. (2017). How Yam, J., & Skorburg, J. A. (2021). From human resources to human
well do my results generalize? Comparing security and privacy rights: Impact assessments for hiring algorithms. Ethics and
survey results from MTurk and web panels to the US. University Information Technology, 23, 611–623.
of Maryland. Zuboff, S. (2019). The age of surveillance capitalism. Profile Books.
Reisman, D., Schultz, J., Crawford, K., & Whittaker, M. (2018). Algorith- Zwitter, A. (2014). Big data ethics. Big Data & Society. https://​doi.​org/​
mic impact assessments: A practical framework for public agency 10.​1177/​20539​51714​559253.
accountability. AI Now Institute. https://​ainow​insti​tute.​org/​aiare​
port2​018.​pdf. Publisher's Note Springer Nature remains neutral with regard to
Ruef, M., & Scott, W. (1998). A multidimensional model of organiza- jurisdictional claims in published maps and institutional affiliations.
tional legitimacy: Hospital survival in changing institutional envi-
ronments. Administrative Science Quarterly, 43, 877–904.

13
 TUTORIAL 7: INTERNAL AND EXTERNAL ENVIRONMENTS

Huawei and the Ren Zhengfei Way

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. Huawei should respond to changes in and outside of China by adapting to the

general environment.

B. Huawei should respond to changes in and outside of China by influencing the task
environment.

C. Huawei should respond to changes in and outside of China by loosening the

alignment between its invisible and visible corporate culture.

D. Huawei should respond to changes in and outside of China by building a corporate

culture that is responsive to the external environment.

1
 Huawei and the Ren Zhengfei Way
Huawei Technologies is the world’s largest manufacturer of telecommunications technology
and is based in Shenzhen, Guangdong, the centre for high technology in China. Huawei was
founded by civil engineer, Ren Zhengfei in 1987 after his retirement from the army at 43
years old i.

Huawei’s founder, Ren Zhengfei. Photo licensed under CC BY.

The firm began by reselling private brand exchange (PBX) switches from Hong Kong to
China. After identifying demand for locally made PBX switches, Ren invested in research and
development and hired 50 employees. Huawei started selling its own switches in 1992 ii.
From the start, the culture at Huawei was gruelling. To produce their first PBX switches
managers and employees worked through the night, relying on power naps when they
became too exhausted to continue iii.
Huawei grew rapidly through the 1990s and 2000s. By 1995 the company had expanded
beyond China. By 2005 its earnings from international markets had outstripped its domestic
earnings. In 2015 Huawei overtook IBM to rank in the top 75 Fortune 500 companies in
terms of income and was number 1 in terms of growth iv.
In 2020 Huawei overtook Apple and Samsung as the top smartphone provider globally. v As
of 2021 they employ more than 194 000 people around the world, operate in 170 counties
and service three billion customers vi.
The development of Huawei’s corporate culture
Ren’s military service influenced his leadership style at Huawei from the beginning. He runs
the company like an army with a focus on strict rules, clear goals, brave spirit and high
efficiency vii.

2
In 1998 Ren outlined his management approach in a document known as the Basic Law. This
document included formal rules related to everything from the company’s mission to its
objectives, growth, distribution of value, as well as business policies related to the
organisation, human resources and control viii.
Ren’s leadership helped to develop the corporate culture that has been key to Huawei’s
success.
Customer-first attitude
Huawei is known for putting the interests of customers first and being serious about
implementing their commitment. One commonly told story is about Huawei’s commitment
to serving the rural areas of China where rats would bite through wires and disrupt services.
While other telecommunications companies typically saw such issues as outside of the
scope of their operations, Huawei committed to solving the problem and developed a
wealth of experience producing chew-proof cables ix.
Employee dedication
Huawei also fosters and rewards employee dedication. Huawei is well known for its
renumeration system, which includes a salary, bonuses, and shares in the company. New
employees receive a rank which they can improve through performance evaluations and
technical exams. It typically takes about two years to move up a rank to receive a higher
salary x.
Employees are allocated shares based on indicators including competence, scope of
responsibility, contributions, attitude and commitment to taking risks. The earnings from
shares mean that Huawei employees earn far more than the highest earners at the same
age in other industries. A typical salary is higher than the wages of employees at Amazon,
Microsoft and IBM. Employee ownership of stocks also creates a commitment to the
company as employee’s fortunes are aligned with those of Huawei xi.
Employees can choose to sign the controversial Dedication Agreement to forfeit leave
entitlements in return for positive performance evaluations and access to company shares
xii.

Employees typically work late into the night and sleep in their offices. The company even
places mattresses under employees’ desks to encourage overtime. One Huawei employee
suggests, “The pads to us were a representation of hard work in the old days and this idea
has now been translated into the spirit of being dedicated to do the best in anything we do”
xiii.

Ren typifies the ethics and self-discipline he expects from his employees. Ren visited Huawei
employees in Afghanistan while the country was at war as well as Libya days before conflict
started. In a visit to Nepal, he told staff “I promised that as long as I could still fly I would
come to tough regions affected by war and disease to accompany you. If I fear death, how
can I call on you to tough it out. If I spur you on to strive for, I do so as well” xiv.

3
Long term vision
Employee ownership helps Huawei to plan for the long term. Unlike publicly listed
companies, Huawei is not subject to fluctuations on the capital market. Unlike most other
telecommunications companies who make yearly plans, Huawei is able to set plans for the
next ten years xv.
Huawei’s three deputy chairmen rotate through the role of CEO every 6 months, with Ren
providing oversight. This unusual arrangement makes the company less vulnerable in the
long term if something happens to one of the CEOs xvi.

Gradual decision-making
Rather than rushing to quick decisions, Ren is careful to take time to reflect. He encourages
intellectual exchange throughout the company and encourages managers to read outside of
their area of expertise xvii. In 1998 he visited IBM and was able to learn how they were able
to improve the efficiency and outcomes of the research and development department. Ren
successfully implemented similar measures at Huawei xviii.
Is Huawei’s corporate culture sustainable?
Although the strong corporate culture at Huawei is widely credited for the company’s rapid
growth, it has come under scrutiny in the past few years.
Former employees have questioned whether the gruelling workload required to succeed at
Huawei is sustainable. One former employee suggested, “I had no time for recreation.
Basically, all I did was keep working mindlessly. Often we were still in meetings until
11pm” xix.
He suggested that it was impossible to survive at Huawei unless you signed the “voluntary”
dedication agreement accepting overtime without pay and forgoing leave xx.
The intense work culture is not exclusive to Huawei. In 2019 some Chinese programmers
coined the phrase “996” to describe the expectations that tech employees in Shenzhen
work from 9am to 9pm for 6 days a week. The phrase “996 working, ICU waiting” gained
popularity and sparked discussion about the negative impacts of long-hours on employees’
health and personal lives xxi.
Although the 996 culture encourages technology workers to regularly work up to 72 hours a
week, the standard legal working hours in China is 40 xxii.
In the midst of these debates, the city of Shenzhen recently passed new regulations that
require employees who regularly work long hours to be granted extra leave to mitigate
burnout. It is unclear whether these measures were in response to the 996 work-culture in
the city’s technology companies xxiii.

4
 Huawei’s headquarters in Shenzhen. Photo licensed under CC BY-SA

Huawei’s future
Huawei’s corporate culture has not only had negative impacts on its own employees, but
beyond the company as well.
Former Huawei employees have spoken out against the so-called “wolf culture” at Huawei
which they suggest encourages employees to bend local laws so long as the outcome
delivers benefits for the company xxiv.
In 2018 Huawei executive and Ren’s daughter, Meng Wanzhou, was arrested in Canada on
the accusation that she committed bank fraud to help the company progress in Iran. Other
employees have been accused of bribing government officials to win business in Africa, and
of copying the source code of an American competitor xxv.
Alongside these issues, Huawei is facing a hostile political environment in the Anglo-sphere.
The so-called Five Eyes security partnership between Australia, the US, Canada, New
Zealand and the UK have raised security concerns about Huawei xxvi.
While Ren is quick to rebut their concerns, the Five Eyes nations are worried that Huawei
could use 5G infrastructure to conduct surveillance of their citizens on behalf on the Chinese
Government xxvii.

5
The Australian Government went so far as to ban Huawei from building the country’s 5G
network, while the other four countries decided to use different providers for the rollout of
5G xxviii.
While it is unclear whether these concerns are related to broader geopolitical tensions
outside of Huawei’s influence, they nevertheless impact on the company’s future in the
Anglo-sphere. As a small part of the company’s global operations, however, these five
countries are not a priority for Huawei xxix.
The new antitrust rules proposed by the Chinese Government are likely to be of bigger
concern for Huawei. These rules, like those proposed in other countries, seek to curve
monopolistic behaviour among technology companies and reduce their power xxx.
After thirty years of growth, Ren Zhengfei reflects on what is next for Huawei. Should recent
developments in the general and task environment prompt a change in direction? Or should
Huawei focus internally on shifting their corporate culture to one that can meet the challenges
of the next decade?
Additional Resources
Watch this video to see Ren Zhengfei discuss Huawei’s future, relations with the US, concerns
about spying, Meng’s arrest and 5G technology:
https://www.youtube.com/watch?v=Nl2jCWDwE8w

i
Jiing-Lih, F. & Xin, P. (2018). Huawei the Ren Zhengfei Way: the "Tough Guy" and His Corporate Philosophy.
Case Study. China Europe International Business School.
ii
Jiing-Lih & Xin. (2018).
iii
Jiing-Lih & Xin. (2018).
iv
Jiing-Lih & Xin. (2018).
v
Sherisse P. (2020) Samsung slump makes Huawei the world's biggest smartphone brand for the first time,
report says". CNN. Retrieved 30 July 2020 from: https://edition.cnn.com/2020/07/30/tech/huawei-samsung-
q2-hnk-intl/index.html
vi
Who is Huawei? (2020) Retrieved 4 February 2021 from: https://www.huawei.com/en/corporate-
information
vii
Jiing-Lih & Xin. (2018).
viii
Jiing-Lih & Xin. (2018).
ix
Jiing-Lih & Xin. (2018).
x
Jiing-Lih & Xin. (2018).
xi
Jiing-Lih & Xin. (2018).
xii
Jiing-Lih & Xin. (2018).
xiii
De Cremer, D. & Tao, T. (2015). Huawei’s culture in key to its success. Harvard Business Review. Retrieved 4
Februrary 2021 from: https://hbr.org/2015/06/huaweis-culture-is-the-key-to-its-success
xiv
Jiing-Lih & Xin. (2018).
xv
De Cremer & Tao (2015).
xvi
De Cremer & Tao (2015).
xvii
De Cremer & Tao (2015).
xviii
Jiing-Lih & Xin. (2018).
xix
Xiao, B. (2021). Former Huawei employee speaks out on Shenzhen's '996' culture as Chinese city enforces
paid leave. ABC News. Retrieved 4 February 2021 from: abc.net.au/news/2021-01-02/china-shenzhen-996-
working-culture-paid-annual-leave/13023218.

6
xx
Xiao, B. (2021). Former Huawei employee speaks out on Shenzhen's '996' culture as Chinese city enforces
paid leave. ABC News. Retrieved 4 February 2021 from: abc.net.au/news/2021-01-02/china-shenzhen-996-
working-culture-paid-annual-leave/13023218.
xxi
Xiao, B. (2021). Former Huawei employee speaks out on Shenzhen's '996' culture as Chinese city enforces
paid leave. ABC News. Retrieved 4 February 2021 from: abc.net.au/news/2021-01-02/china-shenzhen-996-
working-culture-paid-annual-leave/13023218.
xxii
Xiao, B. (2021). Former Huawei employee speaks out on Shenzhen's '996' culture as Chinese city enforces
paid leave. ABC News. Retrieved 4 February 2021 from: abc.net.au/news/2021-01-02/china-shenzhen-996-
working-culture-paid-annual-leave/13023218.
xxiii
Xiao, B. (2021). Former Huawei employee speaks out on Shenzhen's '996' culture as Chinese city enforces
paid leave. ABC News. Retrieved 4 February 2021 from: abc.net.au/news/2021-01-02/china-shenzhen-996-
working-culture-paid-annual-leave/13023218.
xxiv
Zhong, R. (2018). Huawei’s ‘Wolf Culture’ Helped It Grow, and Got It Into Trouble. The New York Times.
Retrieved 4 February 2021 from: https://www.nytimes.com/2018/12/18/technology/huawei-workers-iran-
sanctions.html
xxv
Zhong (2018).
xxvi
Bagshaw, E. (2020). How the US steamrolled Chinese tech giant out of Five Eyes. Sydney Morning Herald.
Retrieved 4 February 2021 from: https://www.smh.com.au/world/asia/how-the-us-steamrolled-chinese-tech-
giant-out-of-five-eyes-20200706-p559fa.html
xxvii
Bagshaw, E. (2020). How the US steamrolled Chinese tech giant out of Five Eyes. Sydney Morning Herald.
Retrieved 4 February 2021 from: https://www.smh.com.au/world/asia/how-the-us-steamrolled-chinese-tech-
giant-out-of-five-eyes-20200706-p559fa.html
xxviii
Bagshaw, E. (2020). How the US steamrolled Chinese tech giant out of Five Eyes. Sydney Morning Herald.
Retrieved 4 February 2021 from: https://www.smh.com.au/world/asia/how-the-us-steamrolled-chinese-tech-
giant-out-of-five-eyes-20200706-p559fa.html
xxix
Bagshaw, E. (2020). How the US steamrolled Chinese tech giant out of Five Eyes. Sydney Morning Herald.
Retrieved 4 February 2021 from: https://www.smh.com.au/world/asia/how-the-us-steamrolled-chinese-tech-
giant-out-of-five-eyes-20200706-p559fa.html
xxx
Bahree, M. (2020) What’s behind China’s Big Tech crackdown and what does it mean? Aljazeera. Retrieved 4
February 2021 from: https://www.aljazeera.com/economy/2020/11/13/china-is-clamping-down-on-its-tech-
companies-heres-what-we-know

7
 TUTORIAL 8: STRATEGIC MANAGEMENT

Sobah Non-Alcoholic Beverages

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. Sobah is successful because its core competencies are relevant to the five
competitive forces in the craft beer industry.

B. Sobah is successful because it has chosen to pursue an appropriate business-level

strategy for its line of non-alcoholic beers.

C. Sobah is successful because it has chosen to pursue an appropriate strategy for

growth at the corporate-level.

D. Sobah is successful because it pays more attention to implementing rather than

formulating strategy.

1
 Sobah Non-Alcoholic Beverages

Sobah founders and owners, Lozen and Clinton Schultz. Photo licensed under CC BY-ND.

The beginnings of Sobah

Sobah, Australia’s first non-alcoholic craft beer, is the brainchild of psychologist and
Gamilaroi man, Clinton Schultz and his wife Lozen. After giving up drinking, Clinton became
frustrated at the lack of non-alcoholic options available on the market.

“The only thing I’d really get offered was a ginger beer and you can only have a couple
before you feel bloated and sugared out. I’d get pissed off having to pay five bucks for a
soda water with a piece of lime in it” i.

Lozen also struggled to find a non-alcoholic craft beer to drink while she was pregnant ii.

At the time, the Schultz’s were running a food truck selling native and Aboriginal inspired
foods. They experimented with brewing a non-alcoholic beer and began selling it out of a
keg on their food truck.

“We also wanted something unique to be able to sell from our food truck that nobody else
would have. So, having our native-infused, non-alcoholic beers just seemed like a perfect
match” iii.

2
Initially, Clinton and Lozen struggled to find a contract brewer who was willing to brew a
beer with traditional Aboriginal ingredients and no preservatives. However, Pickled Pig
Brewery in Tweed Heads agreed to give it a go iv.
The non-alcoholic beer proved popular, and the couple decided to launch a crowdfunding
campaign to produce a line of Sobah beers that customers could take away.

Breaking into the craft beer market proved to be a challenge, as Clinton describes, "It's not
just the individual retailers, there's a real monopoly of the market, and it's damaging, it's
damaging for the art of brewing and entrepreneurship because it can quickly discourage
people from wanting to take a part of it, knowing that there's just that big brick wall facing
you when you've got a product and you're ready to get it out there. It doesn't matter if it's
individual bottle shops, through to the chains, through to pubs and clubs and other
establishments” v.

Four years later, however, and Sobah is sold in around 200 bottle shops and venues across
Australia including through liquor giants, BWS and Dan Murphy’s vi.

Sobah’s line of beers

Sobah’s first beers featuring artwork by Gamilaroi artist Jason Passfield Photo licensed under CC BY-ND

When Clinton and Lozen started brewing in 2017, they were the first to offer a line of craft
non-alcoholic beers in Australia.

3
“The non-alcoholic beers that were available here in Australia for a long time haven’t done
us any favours. When something is selling for $7 a six-pack at the supermarket you know it’s
not going to be any good” vii.
To address this challenge Clinton and Lozen experimented with native ingredients,
developing a Lemon Aspen Pilsner, a Pepperberry IPA and a Finger Lime Cerveza to create a
non-alcoholic option for fans of craft beer.
While using native ingredients was always a priority, Clinton explains that it is not
straightforward: “There’s a big difference between grabbing a handful of something I need
from the bush and finding 100 kilos of it on a regular basis” viii.
Lozen adds: “It can be a struggle to find a lot of Australian native stuff in commercial
quantities. Not every idea we have we can do. To find someone to go out there and grow for
us would take a lifetime.” ix
Another challenge involved efforts to improve the shelf stability of the beer so that it could
be sold in cans and more easily distributed interstate and internationally.
Lozen suggests, “It was difficult, it has been a rocky journey. Going from having a beer you
serve on tap in the keg and putting that in a can or bottle, there are so many new things
you’ve got to look at. We don’t use preservatives or chemicals, the beer is vegan-friendly,
low intervention and it was also a perishable product that needed to be kept cold” x.

“Cold chain distribution is hard, there’s no profit margin and it was already an expensive
product to make.” xi

“We undertook six months of R&D and needed certain equipment to pasteurise. We wanted
to pasteurise in the can because that eliminates any trace of bacteria that contributes to
secondary fermentation. Big retailers want to change the way they retail craft beer and are
urging craft brewers to pasteurise” xii.

To produce pasteurised beer in the quantities needed, Sobah engaged Tribe Brewing
partners based in Goulbourn in NSW, who currently produce most of their beer.

In addition to their original three beers, Sobah has developed special releases including a
Davidson Plum Gluten-Free Ale, a Boab and Wild Ginger Lager, a Wattleseed Gold and an
Aniseed Myrtle Stout xiii.

Working at Sobah

Clinton and Lozen and their three staff work out of their headquarters on Yugambeh
country at Burleigh Heads on the Gold Coast in Queensland.
Clinton explains that “My wife’s actually the engine room of the business and I do the
development side and the public face side of it as well” xiv.

Sobah is a proudly Aboriginal company. This foundation influences everything from their
name, the artwork on their beers, and the broader social purpose of their business.

4
As well as being a play on words, “bah” means place in the Yugambeh language, so the
company name can also mean “place of sobriety” xv.
The cans also include the Gamilarray words ‘Badha Gali’, which means ‘bitter drink as well
as artwork by Gamilaroi artist Jason Passfield which depicts a wedgetail eagle and its
songlines.

“That’s one of the aims of our business, to bring more awareness to Aboriginal and Torres
Strait Islander culture but also the diversity within cultures” Clinton says. “All our cans
explain where our native ingredients come from, what language groups they’re tied to, and
tries to give people a bit of curiosity to hopefully go out and learn a little bit more” xvi.

“I love to smash stereotypes so there’s no better way of smashing a whole group of

stereotypes than being an Aboriginal man who doesn’t drink and produces non-alcoholic
beer for majority non-Indigenous people to drink” xvii.

According to the Sobah website, the company was “born from a desire for providing a
better choice for those who are not drinking for whatever reason at whatever time…Sobah
functions from the philosophies of Gamilaraay Lore ‘dhiriya Gamil’ along with other lessons
[Gamilaroi man and psychologist, Clinton Schultz] has been fortunate to learn from Elders
from around Australia. These include acting from a position of respecting people, place and
the environment; understanding and working towards fulfilling responsibilities to that we
are connected to; and, engaging in positive reciprocity” xviii.
Lozen suggests that one of the key elements of Sobah’s success is their strong brand.
“What we’ve learnt is that you’ve got to have a brand supported by a bloody good story.
It’s not going to stand out from the crowd if you just have a good tasting beer, it’s about
building a community and getting people to be a part of that journey” xix.
In 2019 and to further grow Sobah, Clinton and Lozen sold 8% of the business to private
investors raising $80 000 and valuing the business at $1 million. In time they hope to sell
up to 20% of the business to raise further funds.
Their next plan is to develop a micro-brewery and café on the Gold Coast in collaboration
with other Indigenous businesses xx.
Sobah’s social purpose

For Clinton and Lozen the real purpose of Sobah goes beyond providing a way to drink beer
without a hangover. They want to tackle Australia’s problematic drinking culture head-on
and help reduce the stigma around sobriety.
According to their website, “Sobah is leading a conversation surrounding Australian societal
issues with alcohol consumption and breaking down the stigma of socialising sober. We
promote healthy lifestyle choices and wellness, social equity, sustainability, raise positive
awareness of Aboriginal and Torres Strait Islander culture, smash stereotypes, unite people
and of course, our beers quench thirst” xxi.

5
Clinton’s background in psychology is also a core part of the work he does at Sobah. His
passion is to support healing initiatives that are grounded in traditional, spiritual and
cultural choices rather than those dictated by business or government. Sobah supports a
number of initiatives that seek to promote mindful drinking, sobriety and healthy living
including, The Preston Campbell Foundation, The Wayne Weaver Foundation, Foundation
for Indigenous Sustainable Health (FISH), and 101 Tokens.
Since Clinton and Lozen began Sobah, the market for non-alcoholic beverages in Australia
has rapidly grown. BWS and Dan Murphy’s report that their sales of non-alcoholic drinks
doubled from 2019 to 2020 xxii.
For Clinton this is a positive sign that Australia’s heavy drinking culture is changing.
“Now there’s an overwhelming support for just non-alcoholic beers but non-alcoholic adult
beverages in general” xxiii.

Additional Resources
Watch this video to see Clinton Shultz discuss the purpose of Sobah as a company:
https://www.youtube.com/watch?v=fG8ZTOF9L4A

i
Wilkins, M. (2019). The Sobah Option. The Crafty Pint. Retrieved 2 February 2021 from:
https://craftypint.com/news/2245/the-sobah-option
ii
Burnett, C. (2019). Share sale at non-alcoholic Sobah boosts expansion plans. Brews News. Retrieved 3
February 2021 from: https://www.brewsnews.com.au/2019/06/18/share-sale-at-non-alcoholic-sobah-boosts-
expansion-plans/
iii
Masige, S. (2020). 'I love to smash stereotypes': The founder of Aboriginal-owned, non-alcoholic beer
company Sobah on the brand's social purpose. Business Insider Australia. Retrieved 2 February 2021 from:
https://www.businessinsider.com.au/sobah-non-alcoholic-beer-2020-7
iv
Burnett, Brews News.
v
Sunshine Coast Daily (2018). Raise your glass to Aussie Beer with no booze. Retrieved 2 February 2021 from:
https://www.sunshinecoastdaily.com.au/news/raise-your-glass-to-aussie-beer-with-no-booze/3609879/
vi
Masige, Business Insider Australia.
vii
Wilkins, The Crafty Pint.
viii
Wilkins, The Crafty Pint.
ix
Wilkins, The Crafty Pint.
x
Burnett, Brews News.
xi
Burnett, Brews News.
xii
Burnett, Brews News.
xiii
Sobah (2021). Special Releases. Retrieved 2 February 2021 from: https://sobah.com.au/pages/special-releases
xiv
Masige, Business Insider Australia.
xv
Masige, Business Insider Australia.
xvi
Masige, Business Insider Australia.
xvii
Masige, Business Insider Australia.
xviii
Sobah (2021). Who we are. Retrieved 2 February 2021 from: https://sobah.com.au
xix
Burnett, Brews News.
xx
Burnett, Brews News.
xxi
Sobah, Who we are.
xxii
Masige, Business Insider Australia.
xxiii
Masige, Business Insider Australia.

6
 TUTORIAL 9: ORGANISING

Corporate disfunction on Indigenous Affairs: Why heads rolled at Rio Tinto

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. To avoid causing future damage to local communities, RioTinto should retain its
product divisional structure with the addition of a “social performance” function that
reports to an executive on the corporate executive committee.

B. To avoid causing future damage to local communities, RioTinto should shift from a
product divisional structure to a functional structure to empower the corporate
relations function.

C. To avoid causing future damage to local communities, RioTinto should shift from a
product divisional structure to a matrix structure to give the corporate relations
function and product divisions joint oversight.

D. To avoid causing future damage to local communities, RioTinto should adopt a more
adaptive organisational design to empower decision-makers in local operations.

1
 Corporate disfunction on Indigenous Affairs: Why heads rolled at Rio Tinto
Outraged investors have forced the board of Rio Tinto to sack its chief executive Jean-
Sebastien Jacques along with two of the senior executives partially responsible for the
destruction of the Juukan Gorge caves in the Pilbara region of Western Australia, which
contained evidence of human habitation more than 46,000 years ago.

Why did the company commit this egregious act of cultural vandalism?

There are several layers to the answer.

First, Rio Tinto’s iron ore division was under extreme production pressure.

The iron ore in the vicinity of the caves was very high grade and the company needed as
much of it as possible to mix with other grades so as to supply the market with its
trademark Pilbara blend.

Its reputation as a reliable supplier depended on it.

Rio Tinto was siloed

It was authorised under West Australian law to mine the area, even though this would
destroy the caves, and it intended to exercise this legal right, regardless of any opposition.

Second, Rio has a segmented organisational structure.

It’s product divisions – iron ore, aluminium, copper and diamonds, minerals and energy –
operate as autonomous business units with relatively little control from the corporate
centre.

When things go well, such an organisational structure is highly profitable for the corporation
as a whole, because it leaves each product division free to take advantage of whatever
opportunities there are in its particular market.

But there is a downside. It leaves the corporation vulnerable to poor decision making by any
one of its product divisions, decisions that may have disastrous human and environmental
consequences that threaten the social license of the whole corporation.

That is what has happened in the Juukan Gorge case.

Organisational weakness

Of course, Rio was aware of this downside of its organisational structure and had taken
steps to deal with it.

It had various functional lines – chains of command – external to the product divisions,
providing services into the divisions and exercising some degree of control over their
decision making.

2
At the lower end of these chains were people working within the product divisions and at
the top, an executive sitting on the executive committee of the corporation, that is, at the
same level as the heads of the product divisions.

These functional executives were thus in a position to challenge the product division heads
about the decisions they were making.

One of these external functions was called communities, or Indigenous relations. It had staff
located within Rio Tinto Iron Ore whose job was to assess and manage the company’s
impact on indigenous people and where necessary, to negotiate with them for access to
land.

The chain of command for this function led upwards to an executive for corporate relations,
on the executive committee of corporation. Corporate relations had not always had a seat
at this table. For decades it had reported to the head of either human resources or legal,
both of whom sit on the executive committee.

Indigenous relations were ‘corporate relations’

When Jacques became chief executive in 2016, he elevated the head of corporate relations
to sit on the executive committee, “to ensure that ‘social license’ areas were reflected as an
area of specialisation around the executive committee.”

Theoretically this function should have been able to pass the word up the line about the
impending destruction of the Juukan Gorge caves, so that either the executive for corporate
relations or the chief executive himself might have intervened.

This didn’t happen. The function that should have acted as a watchdog in this situation
failed to do so.

The explanation lies in part in how the executive committee is structured. It consists of all
those who report directly to the chief executive.

It would make the job of the chief executive too unwieldy if every area of activity was
directly represented on this committee, so only the most important are.

Less important areas are grouped together and collectively represented.

Structure matters

This means the makeup of the executive committee sends a signal about relative
importance.

It consists of the heads of the four product divisions, plus the head of human resources, the
chief legal counsel, a chief commercial officer, a chief financial officer, a single head for a
grouping consisting of growth, innovation, and health safety and environment, and the
aforementioned London-based head of corporate relations, Simone Niven, who is now
departing.

3
 Departing CEO Jean-Sebastien Jacques.

This is the team of direct reports to the chief executive.

It means that Indigenous relations is not represented directly at this level. It is grouped with
several other specialisations in the portfolio managed by corporate relations.

That portfolio consists of external affairs, sustainability, communities (Indigenous relations),

brand, media, government affairs and employee communications.

This is a massive span of concerns, each of which is potentially very demanding.

Indigenous relations must compete with all these other concerns for the attention of the
head of corporate relations, and not surprisingly, it did not always get the attention it
deserved.

Indeed, Niven had not heard about the Juukan Gorge caves until just days before the
destruction, and even then, she was apparently unaware of extraordinary significance or
age of the caves.

The situation was worse.

Her direct reports did not include anyone with dedicated accountability for Indigenous
relations. The way it worked was that people with a particular responsibility for Indigenous
relations at Rio’s iron ore operations in Australia reported to an Australian head of
corporate relations, who in turn reported to the global head.

4
This Australian head had the same diverse portfolio of concerns as did his global boss, which
again meant it was difficult for Indigenous relations to get the attention it deserved.
Moreover, a look at his resume suggests that Indigenous relations was not a significant part
of his professional experience.

As well as a primary reporting line to global head of corporate affairs, the Australian head
had a secondary or dotted reporting line to the head of Rio Tinto iron ore.

As head of the Indigenous relations function it would have been his role to elevate the
Juukan Gorge matter to the senior leadership team of Iron Ore, well before preparations to
blast the area began.

However, it seems he had very little awareness of the issue and was himself briefed about it
at a meeting just three days before the blast occurred. Even then, the briefing paper “did
not identify the exceptional significance of the sites or their age”.

In short, the indigenous relations function failed dismally to alert top management because
it had lost its influence and been swallowed by a corporate relations function with a much
wider remit. This is the fourth and critical level of level of explanation.

Has Rio learnt its lesson?

The changes which the Rio board has announced suggest it has learnt this lesson.

First, it announced that a “social performance” function will be established that reports to
an executive on the corporate executive committee.

Social performance is broader than Indigenous relations, but it is far more focussed on the
rights of project-affected people than is the broad concept of corporate relations.

Second, the board specifies that the executive concerned is also be the culmination point
for heath, safety and environment, as well as technical and projects.

The inclusion of “projects” might be a serious drawback, since that role is likely to be
production-oriented. It conflicts with the thrust of the health, safety and environment and
social performance roles, which aim to ensure these issues are not sacrificed in the interests
of production.

Third, as well as being part of a specialised function that reports up to the executive
committee, social performance staff will be “embedded” within local mine management.

They will have secondary reporting lines within Rio Tinto Iron Ore which will enable them to
exercise more influence than previously.

The critical question is whether the social performance function will retain its identity to the
top of the hierarchy.

5
 Juukan Gorge caves before destruction. Puutu Kunti Kurrama And Pinikura Aboriginal Corporation

Specifically, will it be headed by a corporate social performance specialist who answers

directly to the relevant executive on Rio’s executive committee.

If not, the social performance will be swallowed, just as was Indigenous relations before it.

Companies tend to implement major organisational changes after a disaster that affects the
whole corporation.

The Juukan Gorge matter was, among other things, a corporate public relations disaster for
the whole company.

Now is the time to fix the organisational failings that contributed to it.

If Rio Tinto fails to implement the lessons effectively, the sacrifice of three of its corporate
chiefs to appease investors will have been in vain.

Additional Resource:
Watch this video to see Puutu Kunti Kurrama and Pinikura traditional owners explain what
the destruction of the Juukan Gorge site means to them. Some viewers might find the
footage of the destruction distressing:
https://www.youtube.com/watch?v=s3tVznXBkUs&t=6s
Reference:
Hopkins, A. & Kemp, D. (2020) Corporate disfunction on Indigenous Affairs: Why heads
rolled at Rio Tinto. The Conversation. Accessed 2 February 2021 from:
https://theconversation.com/corporate-dysfunction-on-indigenous-affairs-why-heads-
rolled-at-rio-tinto-146001

6
 TUTORIAL 9: SHEIN

Shein Is the World’s Most Popular Fashion Brand—at a Huge Cost to Us All How Shein's
Low Pricing is a Huge Cost to us All

Prior to attending your tutorial, read the case study printed on the next page. In your
tutorial, your group will be asked to discuss all of the statements below and select the
statement you agree with most strongly. You must be able to justify your selected structure
by drawing on theory from the online module and applying it to the case study. Your group
can choose one statement only (either A, B, C, or D) and you must be able to defend your
choice to other groups.

The statements

A. Shein should pursue an obstructionist strategy when it comes to Corporate Social

Responsibility.

B. Shein should pursue a defensive strategy when it comes to Corporate Social

Responsibility.

C. Shein should pursue an accommodative strategy when it comes to Corporate Social

Responsibility.

D. Shein should pursue a proactive strategy when it comes to Corporate Social

Responsibility.

1
 TUTORIAL 10: SUSTAINABILITY
Shein Is the World’s Most Popular Fashion Brand—at a Huge Cost to Us All How Shein's
Low Pricing is a Huge Cost to us All

Just before Christmas, Mélo—a lifestyle blogger from Paris known for promoting body
positivity—uploaded a video on Instagram showing off shiny, glittering party dresses from
Shein, a fast-fashion e-retailer from China.
”Which one do you prefer for Christmas and which one for New Years?” the French content
creator asked her 75,000 followers as she twirled, strutted, and smiled at the camera. At the
bottom of the caption, she tagged Shein, directing her viewers to a 15% discount on all their
clothing orders.
Mélo is one of the thousands of influencers on TikTok and Instagram who works with Shein
to post “haul” videos, where she tries on piles of clothes, invites followers to visit Shein’s
pop-ups, and promotes their sale events. It’s a social media strategy that has been wildly
successful for the company, pushing Shein to the top of the fashion ladder—and making it
the world’s most popular fashion brand in 2022.
According to a report compiled by Money.co.uk, Shein has taken over giants like Nike and
Adidas as the most-Googled clothing brand, and Zara and Macy’s in online sales. But several
reports over the last year reveal the company’s shocking track record of human rights
violations and an environmentally-unsustainable model—leaving many to wonder how its
popularity continues to soar among consumers.
A global juggernaut
Pronounced “she-in,” the fast fashion Chinese behemoth was founded in the city of Nanjing
in 2008 by Chris Xu, a U.S.-born entrepreneur and search engine optimization specialist.
Over the years, Shein went from being a low-cost Chinese apparel merchant to a global,
online-only fashion juggernaut, climbing in sales from $10 billion in 2020 (according to
Bloomberg) to a whopping $100 billion in 2022.
Its biggest selling point is the low pricing of clothes that are shipped to more than 150
countries and regions worldwide, catering to women in their teens and 20s. The business
model works like Amazon—a sprawling online marketplace brings together about 6,000
clothing factories in China under Shein’s label, while internal management software collects
near-instant data about which items are selling and which aren’t to visibly boost the popular
items. According to an investigation by Rest of World, Shein added anywhere between
2,000 and 10,000 individual styles to its app each day between July and December of 2021.
The clothes are also presented in polished advertising campaigns run by Shein’s head office.
The company has poured millions into Google and Facebook campaigns, advertisement
deals, and even a social media reality show co-hosted by Khloé Kardashian. But the center of
its marketing strategy is its use of influencers and their #SHEINhaul videos: the company has
partnered with countless micro-celebrities, fashion bloggers, and reality show contestants

2
who show off their Shein deliveries. The company was reportedly working with about 2,000
Indian influencers alone before the Indian government banned the app in 2020 as a
retaliatory move against China.

An Extinction Rebellion protestor holds a placard reading "SHEIN = 22% of warming gas
emitted for youths" in front of a pop-up store. Alain Pitton—NurPhoto/Getty Images

An unsustainable model
Put simply, Shein produces an astounding number of items on a daily basis—the primary
reason why the company has an unsustainable model. Shein’s CEO, Molly Miao has stated
that each item is produced only in small numbers, between 50-100 pieces a day, before it
becomes popular and is then mass-produced. But the manufacturers’ rapid use of virgin
polyester and large consumption of oil churns out the same amount of CO2 as
approximately 180 coal-fired power plants, according to Synthetics Anonymous 2.0, a report
published on fashion sustainability.
As a result, the company leaves about 6.3 million tons of carbon dioxide a year in its trail—a
number that falls well below the 45% target to reduce global carbon emissions by 2030,
which the U.N. has said is necessary for fashion companies to implement to help limit global
warming.
Across the board, the fashion industry is responsible for releasing more than 10 percent of
carbon emissions. Many brands that are signatories of the Fashion Industry Charter for
Climate Action—the industry’s flagship climate initiative spearheaded by the U.N.—have

3
committed to either halving emissions by the end of the decade, or to set science-based
reduction targets.
With almost all of its impact taking place in its supply chain, Shein also committed to
submitting its own targets for validation. In October, on the heels of the company coming
under fire for fuelling wasteful consumption, Shein announced it would spend $7.6 million
on a partnership with the nonprofit, Apparel Impact Institute, which works with
manufacturers to set and implement energy efficiency programs. It aims to reduce supply
chain emissions by 25 percent by 2030 through energy-efficient projects and a transition to
renewable power for manufacturing. The projects include the Carbon Leadership program,
which benchmarks and assesses carbon impact, and Clean by Design, which will aim to
reduce energy, water, and chemical consumption in textile production at more than 500 of
Shein’s partner facilities.
Still, the impact of these initiatives is yet to be seen, especially considering a survey by BCG
which found that just 18% of retailers who had previously set emissions targets were on
track to achieving them, while another 35% were stalled in their progress.
A damning record on workers’ rights
Despite it all, perhaps the bigger controversy regarding Shein is the treatment of its
workers, who toil away in Chinese factories in unfit conditions. A lengthy investigation by
Wired first chronicled how both laborers and consumers suffered from the production of its
clothes, while a documentary by the U.K.’s Channel 4 found that Shein employees were
working 75-hour shifts with very little time off. Then, Swiss watchdog Public Eye released
another detailed report in November which accused Shein of violating Chinese labor laws.
The group hired independent Chinese researchers to track Shein’s manufacturing and
packing process in China and Europe and found that many were running informal factories
set up in residential buildings.
Advocacy groups and journalists also uncovered evidence that Shein’s $11 bikinis and $7
crop tops were being made by people working in unsafe workshops, lacking safety protocols
like windows and emergency exits. Many also worked without contracts or minimum wage
requirements, thereby allowing the company to reportedly fail to pay its employees
properly. Channel4’s documentary, Inside The Shein Machine, sent undercover cameras to
film factory workers who were forced to pull 17-hour shifts to make hundreds of garments a
day. In one factory, they made a daily base salary of $20, which would then be docked by
$14 if any garments had mistakes.
Eventually, Shein admitted to the breaches and released a statement that read: “We know
we have a responsibility to safeguard the welfare of workers in our supply chain. In light of
the recent report in the news, we launched an investigation into the claim that 2 of our
suppliers had unacceptable working conditions at their facilities.”
But still, the brand continues to soar in popularity across the world—especially in countries
like France, Ireland, Iceland, Egypt, Sudan, South Africa, Saudi Arabia and China. When Drew
Afualo, a TikTok influencer with over 6 million followers, faced online backlash for

4
partnering with Shein, she defended the partnership by saying that “not everyone can
afford to shop sustainably.”
“Sustainable fashion is a privilege,” wrote the 26-year-old in response.
What’s next for Shein?
In December, it was reported that Shein is now exploring moving to an online marketplace
platform to enable other merchants to sell directly to customers, rather than under Shein’s
label, according to a memo to investors viewed by The Wall Street Journal. “The
marketplace platform makes available a range of additional merchandise and shipping
options, and we expect it to result in increased customer engagement and satisfaction,” the
memo said.
The move is seen as part of the company’s attempt to move its supply chain away from
China, where it has largely been rooted with over 3,000 suppliers in southern China’s
Guangdong province. With its headquarters now based in Singapore, Shein also began
manufacturing in Turkey and is leasing and operating warehouses in Poland to ship to
customers in Europe.
In the same memo, Shein addressed concerns around labor law violations, stating that all
suppliers would be required to comply with a code of conduct based on conventions by the
International Labour Organization. Shein has also built an in-house team that will monitor its
supply-chain partners and is engaging independent agencies like Intertek Group PLC to
conduct regular audits of supplier factories. Citing an Intertek study that found 96% of its
workers receive wages higher than the industry average, Shein also said it had made
“significant investments to improve the working conditions of our suppliers’ facilities.”
In the meantime, influencers are also under increasing criticism from environmental and
workers’ rights organizations for working with Shein—with some even deciding to part ways
as a result.
Georgia Portogallo, a 21-year-old British influencer with over 200,000 followers on
Instagram and TikTok, released a video stating that would no longer be working with the
brand until its working conditions had changed shortly after appearing on the Channel 4
documentary.
“I now know … that [Shein’s] staff are underpaid, they work too many long hours, they don’t
get days off. The whole working conditions are horrendous,” she told her TikTok followers.
Additional Resource:
Watch this video about the sustainability concerns levelled at Shein:
https://cdn.jwplayer.com/previews/mgWvtmzh
Reference:
Rajvanshi, A., Caldwell, J. & Johnson, A. (2023). Shein Is the World’s Most Popular Fashion
Brand—at a Huge Cost to Us All How Shein's Low Pricing is a Huge Cost to us All. Time

5
Magazine. Accessed 13 February 2023 from: https://time.com/6247732/shein-climate-
change-labor-fashion/

6
 How Nespresso’s coffee revolution got ground down
1975 to 1985: The invention of Nespresso Coffee

In 1975, a young engineer named Eric Favre had recently started working at Nestlé’s headquarters in
Vevey, Switzerland. One of his first projects was to develop a machine that would combine the
convenience of domestic coffee with the quality of an Italian espresso bar. He was inspired by a
barista in Rome who pumped a piston multiple times before releasing the coffee. This process drew
out more flavour from the beans and produced more of a crema – the layer of foam formed on top of a
good espresso. Favre returned to Switzerland and, along with a small team, set about designing a
machine that could replicate this procedure.

In 1976, Nestlé filed its first patent for a single-serve coffee system. Today, some 14bn Nespresso
capsules are sold every year, both online and from 810 brightly lit boutiques in 84 countries. More
than 400 Nespressos are drunk every second. Hundreds of rivals and imitators have emerged, some
making capsules for Nespresso machines, others pushing competitor systems. In 2013, the most recent
year it released figures, Nespresso’s revenues totalled $10.8bn.

Thirty years after its first successes, Nespresso has scale, experience and buying power that no other
premium coffee company can match. But increasingly it finds itself threatened from below by its
rivals’ cheaper capsules, and from above by fussier coffee enthusiasts. Nespresso also faces mounting
criticism over the environmental impact of its pods.

When Nespresso was in development, colleagues feared that if Favre’s invention succeeded it would
cannibalise the company’s existing coffee businesses, especially the flagship Nescafé instant brand.
At the time, Nestlé saw itself as a mass market company that sold cheap, reliable products: chocolate
and baby food and cereal. This was something different, whatever it was.

1986 to 1997: Nespresso under Gaillard’s leadership

When Nespresso was finally launched in 1986, it failed to find many takers. The first models were
designed to resemble traditional espresso machines, and only four types of capsule were available.
Pitched to businesses in Switzerland and Japan, for offices without enough space for a full-size coffee
machine, Nespresso failed to find many takers. In 1988, in a bid to rescue the product, Nestlé brought
in Jean-Paul Gaillard, a tobacco man who had created the clothing brand Marlboro Classics when he
worked at Philip Morris. Gaillard would work alongside Favre, but his brief was clear: if he couldn’t
turn the ship around, it would be sunk. “At the original launch the product was wrong, the positioning
was wrong and the targeting was wrong,” Gaillard told me. “It had cost a lot of money and brought
nothing.”

Under Gaillard, Nespresso would be transformed into a luxury brand, the look and feel of which
would be as much a part of the product as the beans themselves. “I wanted to create the Chanel of
coffee, and decided to keep it chic and bobo,” he said in a 2010 interview

Gaillard cut the price of the machines and licensed them to third parties. The first home machines had
been made with one firm, Turmix. Later, you could buy a Krupps or Alessi Nespresso machine. These
brand associations gave Nespresso familiarity in local markets, and encouraged fancy shops such as
Harrods to stock them. Gaillard also overhauled the capsules, reducing the aluminium content and
putting up the price by 50%. Most importantly, he began marketing Nespresso to individual
consumers, rather than to businesses, through the new Club Nespresso. It was no longer just a better
coffee for your office – it was a way of life.

1
When you ordered capsules, you joined the “Club”, which also meant handing over your contact
information. Over time, Nespresso gained a huge database of customers it could market to, as well as
a way of recording consumer preferences and buying habits. For customers, the club created the sense
that you were part of a sophisticated worldwide cabal of corporate espresso lovers.

After Gaillard’s reforms, Nespresso finally took off, but it is Favre who tends to take the plaudits as
the creator. The story of the tinkerer playing with pipes and valves in his workshop is more appealing
than the smooth corporate rebranding exercise. Gaillard is only too glad to correct the record. “Those
who really know the story, know it was me,” he said. “Favre was a technician. He couldn’t run the
business.” In 1990, after two years of struggle and personality clashes between the two men, Favre
resigned, the result of what he described as a “coup d’etat”. It doesn’t take much to reopen the old
wounds. “Gaillard is un diable,” Favre told me. In his version of the story, Gaillard was a brash
operator who made his position unbearable.

In 1997, Gaillard left Nespresso to run Nestlé’s ice-cream business in the US. He subsequently left the
company after falling out with the then CEO, Peter Brabeck-Letmathe.

1998 to 2012: Global growth

One crucial factor behind Nespresso’s rise, unmentioned by Gaillard, was timing. In 1998, Starbucks
arrived in the UK, and elsewhere in Europe from 2001. Previously it had been difficult to get a decent
coffee anywhere outside Italy. At Starbucks, you could enjoy Italian-style coffee, which is to say
freshly made and with frothy milk, marketed with Italian-style language. According to the historian of
consumption, Jonathan Morris, Nespresso capitalised on these new tastes: “When customers started to
ask how they could have [Starbucks-style coffee] at home, Nespresso was the best-placed product to
take advantage of that.”

In George Clooney’s first ad for Nespresso, which aired in 2006, he wanders into a Nespresso
boutique and starts making himself a cup of coffee. Clooney’s public image – sophisticated,
cosmopolitan, expensive – fitted Nespresso’s desired image. The years that followed Clooney’s first
ad were Nespresso’s happiest. In 2006, its revenues passed £500m. By 2010 they had reached 3bn
Swiss francs (£2.5bn), and the capsule market was growing five times faster than the overall coffee
market. In Switzerland, Nespresso took business from roast and ground; in China, from tea; in Britain,
from instant. Nespresso reigned supreme over an entire domain of coffee that it had effectively
created from scratch.

Although Nespresso’s rise can be told in part as a triumph of branding, it also depended on a smart
approach to patenting and design. One of Gaillard’s innovations was to rebalance the business
towards making revenue from the capsules rather than the machines. For years, that model
underpinned Nespresso’s global growth. But eventually, would-be competitors spotted an opportunity
to exploit the niche that Nespresso had created. In 2008, Gaillard launched the Ethical Coffee
Company, which sold biodegradable capsules for Nespresso machines. In 2010, the American firm
Sara Lee started to sell capsules that worked in Nespresso machines.

Nespresso furiously litigated against its rivals, arguing that its patent systems were being infringed.
Things came to a head in 2012, when a key batch of Nespresso patents from 1992 were set to expire.
That year, Nespresso lost its patent battles in Germany and England, and settled other outstanding
cases around Europe. Overnight, the company had to accept it could no longer stop third-party
capsules being sold for its machines. The court cases also made awkward PR for a company keen to
promote its ethical sourcing. To many, it seemed that the Nestlé Goliath had gone after smaller,
pluckier, seemingly more ethical Davids and been slain.

It didn’t help that in many consumers’ eyes, Nestlé was still tainted by the formula milk scandal of the
70s. A report published in 1974, titled The Baby Killer, showed how Nestlé aggressively promoted

2
formula milk over breastfeeding in poor countries, where clean water was hard to come by. Some
sales reps even wore nurses’ uniforms to gain an aura of credibility. The report led to a worldwide
boycott and reform of its sales practices. Even today, Nespresso employees I spoke to said the
memory of the scandal hampers its messaging around coffee. More recently, the company’s reputation
was further damaged when the Channel 4 documentary series Dispatches found children under 13
working 40-hour weeks on farms that supplied coffee to Nespresso and Starbucks. (Nespresso
launched an internal investigation after the programme aired. “Protecting children from exploitation
and ensuring they are able to learn is of paramount importance to us, and that is why we have zero
tolerance for child labour,” a spokesman told me.)

2013 to 2019: The pivot to sustainability?

Jean-Marc Duvoisin became CEO in 2013, and was charged with taking Nespresso to a new era,
leaving the patent disputes behind. “There are going to be rival capsules,” Duvoisin told me. “We
need to leverage our strengths: knowing our customers, and knowing our farmers.” Having worked
for 40 years to be the only coffee-pod system in town, the company had to pivot to arguing that its
capsules – made from strong, light aluminium, and filled with high-quality, responsibly farmed coffee
– were the best on the market. Eight years on from Nespresso’s annus horribilis, its biggest problem
is the aluminium itself.

In the past decade, consumers have grown increasingly concerned about the sheer amount of waste
caused by coffee pods. Halo, a firm which makes compostable pods, estimates that of the 39,000 pods
made every minute, 29,000 will end up in landfill.

Nespresso uses aluminium because it is light, strong and durable, making it the best material for a
sealed container that must be flown around the world and then subjected to extreme heat and pressure
on someone’s kitchen counter. Only a tiny amount of coffee is used in each pod, so less coffee is
wasted than in a cafetière, or with other methods, in which many grams can be used per cup. And the
pods are, in theory, 100% recyclable. But because they contain plastic as well as aluminium, they
can’t just be dropped in a regular recycling bin. Instead, used capsules must be dropped off at
Nespresso boutiques or some convenience stores; in some countries, Nespresso offers a service that
collects them from customers’ homes.

Unlike plastic, used by many of Nespresso’s rivals, aluminium is 100% recyclable, but there is a big
difference between offering recycling facilities and getting consumers to use them. Nespresso says its
global recycling rate is 30%, and that 91% of its users have access to one of its 100,000 collection
points around the world. But some experts have suggested that just 5% of Nespresso pods are
recycled. Even if Nespresso’s figure is accurate, with a conservative estimate of 14bn capsules being
sold each year, and 0.9 grams of aluminium per capsule, that means 12,600 tonnes of Nespresso
aluminium end up in landfill annually, enough for 60 Statues of Liberty.

“The business model of the future is not in grand statements about what companies will do,” says
Tima Bansal, a professor at Ivey Business school in Canada who specialises in sustainability. Bansal
was mystified as to why Nespresso didn’t provide more detailed public data, such as regional
breakdowns, about how its capsules are recycled. “With their competitive advantage, they could be a
model of sustainability, leading the circular economy,” she said. “But once you lose your way, the
competition makes it really scary.”

On top of the landfill problem, there are the environmental costs of producing aluminium in the first
place. Mining a tonne of aluminium can produce about 10-12 tonnes of waste, including 2-3 tonnes of
toxic alkaline red mud. In an attempt to go slightly more green, Nespresso is now working with the
commodities giant Rio Tinto to use only “sustainable aluminium”. You might remember Rio Tinto
from such edifying corporate stories as “accepting bribes in China”, “corruption allegations in
Guinea” and “the Norwegian government concluding that Tinto were ‘directly involved’ in ‘severe

3
environmental damage’ through a mine in Indonesia”. If a tie-in between an Anglo-Australian mining
conglomerate with a history of scandals and a secretive Nestlé-owned coffee company doesn’t calm
the doubters, what will?

As Nespresso has grown, it has come up against an awkward truth: the more popular a brand is, the
harder it is to maintain a luxury image. “Our competitor is not other coffee companies,” claimed
Duvoisin. “When you go into our boutique, you are comparing us to Dior or Louis Vuitton.” That
may have been true once, but its boutiques are now on every high street.

2020 and beyond: Le Cunff faces COVID-19

Like other high-street businesses, Nespresso has been buffeted by months of coronavirus closure. In
its late-00s incarnation, when most of its pods were sold by mailorder or on the internet, Nespresso
would have been less affected by coronavirus.

Nearly half a century after it was conceived, Nespresso finds itself in an uncomfortable new world.
Consumers who might have once craved its polished, urbane chic now look for dirty-fingered
artisanal blends to use with their pour-overs and Aeropress machines. A Nespresso machine on the
kitchen counter used to prove your membership of a convenience-loving global consumer coffee elite.
Increasingly it suggests that you are not a serious coffee person, and that your attitude to the future of
the planet is suspiciously relaxed.

In the summer of 2020, buffeted by Covid-19, Nespresso trundles on. A new CEO, Guillaume Le
Cunff, another long-term employee, who had previously worked on many of the company’s
sustainability initiatives, took over earlier this year. In late May, a new ad appeared on Nespresso’s
YouTube page, emphasising its sustainability credentials. George Clooney is nowhere to be seen.

Reference: Adapted from Cumming, E. 2020. How Nespresso’s coffee revolution got ground down.
The Guardian. Accessed 3 August 2020 from:
https://www.theguardian.com/food/2020/jul/14/nespresso-coffee-capsule-pods-branding-clooney-
nestle-recycling-environment

END OF CASE STUDY