QUESTIONS ABOUT THE ULTIMATE QUESTION:

CONCEPTUAL CONSIDERATIONS IN EVALUATING

REICHHELD’S NET PROMOTER SCORE (NPS)

Douglas B. Grisaffe
University of Texas at Arlington

ABSTRACT potential threats present in various elements

of the NPS formulation.
Approximately four years ago in a
Harvard Business Review article, Frederick INTRODUCTION
Reichheld (2003) – noted Harvard Business
School Press author, speaker, loyalty expert, In the influential Harvard Business
and Director Emeritus of Bain & Company Review, Frederick Reichheld (2003)
Consulting – introduced a concept called the introduced the idea of a Net Promoter Score
Net Promoter Score (NPS). Reichheld’s (NPS). He claimed this single summary
claim was straightforward: of all the customer number from one customer survey question is
survey metrics an organization can track, one a sufficient basis for profitably measuring and
stands out above all others in terms of its managing customer loyalty. On a 0-to-10
relationship with company financial scale, customers answer the question: "How
performance – an aggregate-level measure likely is it that you would recommend
derived from scores on a “likely to [company X] to a friend or colleague?"
recommend” survey item. In his article, Anyone rating 0 to 6 is labeled a "detractor",
Reichheld presented the case for his premise. 7 or 8 "passively satisfied", and 9 or 10 a
While most scholars would agree that positive "promoter." The Net Promoter Score (NPS)
word of mouth is highly beneficial and that is the percent “promoters” minus the percent
negative word of mouth is detrimental, less “detractors.” According to Reichheld (2003),
tenable is Reichheld’s claim that a single this single number has more relationship with
word of mouth metric is the ‘one thing’ a company financial performance than all
company needs to track and manage. A others he tested, leading to the following
recently published Journal of Marketing statement:
paper challenges the validity of Reichheld’s
claims on empirical grounds (Keiningham, "This number is the one
Cooil, Andreassen, and Aksoy 2007). number you need to grow. It's
However, in addition to empirical scrutiny, that simple and that profound."
evaluation of Reichheld’s NPS should include p. 54.
detailed conceptual scrutiny. If there are
threats to validity in the conceptual elements, Following the original article,
these must be factored into evaluations of any Reichheld continued to spread that message in
empirically-based claims. This paper adds to additional published material (e.g., Reichheld
the assessment of NPS by going back to 2004; 2006a), conference presentations (e.g.,
Reichheld’s original work and suggests that Reichheld, 2006c), and a Harvard Business
rethinking on conceptual grounds will reveal School Press book exclusively devoted to the
Volume 20, 2007 37
topic (Reichheld 2006b). His NPS concept With this collection of differing
has also gained considerable momentum opinions and viewpoints, what is to be made
because of its appealing simplicity; of NPS? Minimally, thorough evaluation of
compelling claims of links to profitability NPS must be made from at least two
(Reichheld 2003; 2004; 2006a; 2006b); perspectives, a) on empirical grounds, and b)
apparent independent replication of those on conceptual grounds. To confirm or fail to
links by other researchers (Marsden, Samson, confirm the claims that surround NPS as the
and Upton 2005); reported adoption by “ultimate” question, rigorous empirical testing
prominent companies such as GE, American must be done, as indeed some have been
Express and Microsoft (Creamer, 2006; undertaking (e.g., Keiningham et al. 2007;
Keiningham et al. 2007); and Reichheld’s Marsden et. al 2005; Morgan and Rego 2006).
own strong consulting credibility and stature. Second, there must be rigorous scrutiny of the
Another force augmenting NPS attention is its conceptual foundations underpinning
natural fit with the modern “revolution” Reichheld’s work and his message. These
taking place regarding consumer-to-consumer issues also are critical in evaluating the
communication and consumer generated foundation upon which NPS has been built
media, blogs, and viral marketing (c.f., Kirby because quantification and associated
and Marsden 2005). Evidence of some of the empirical analyses strictly hinge on the
breadth and volume of attention received in quality of conceptualization and
just four short years can be seen with a simple operationalization.
Google search on the term “Net Promoter,” The core purpose of this current article
and a visit to the ‘What They’re Saying’ is to examine the conceptual foundations of
section of the official NPS website NPS. The evaluation and arguments
(www.netpromoters.com). presented here stem largely from a
Despite the impressive momentum of practitioner’s perspective (see Endnote).
the net promoter concept, not everyone has Additional points of critique are drawn from
been willing to so quickly accept and adopt logic and principles of social science and
Reichheld’s NPS. Immediate and subsequent marketing methodology. The claims under
challenges to Reichheld’s claims have arisen. scrutiny come from Reichheld’s original
A number of points of critique emerged from paper on the NPS topic (Reichheld 2003),
practitioners and academics shortly after the several of which are reiterated here through
2003 article (e.g., Grisaffe 2004a; Grisaffe the use of a liberal set of direct quotes.
2004b; Morgan and Rego 2004; Kristensen Finally despite the concerns that
and Westlund 2004). And while the years follow regarding NPS, two important notes of
following have seen many enthusiastically clarification and intent should be added. First,
embracing Reichheld’s prescription, others one can raise points of critique regarding NPS
have continued to raise notes of caution that while still being an absolute advocate of
the simple claims about NPS may not reflect earning positive word of mouth
the “ultimate” in customer measurement and communication from customers, and
management. Again, these concerns have strategically avoiding negative word of mouth
been raised by both academic and practitioner communication. Marketers accept that word
authors (Brandt 2007; Crosby and Johnson of mouth is a critical behavioral outcome of
2007; Keiningham et al. 2007; Morgan and strategic customer experience management.
Rego 2006; Nicks 2006; Pingitore, Morgan, Word of mouth in that light is a consequence
Rego, Gigliotti, and Meyers 2007). resulting from customer perceptions and
38 Evaluating Reichheld’s NPS

evaluations of a company’s total offering among the points of concern is the

(e.g., excellence in products, services, value overarching core claim made by Reichheld,
for the money, reputation, etc.). Managing to namely, that tracking one number based on
excellence on those causally-driving one customer-survey question (likelihood to
dimensions is required to generate positive recommend to others) is a sufficient approach
word of mouth from customers, and to avoid to the measurement and management of
negative word of mouth. Thus, it is noted up customer loyalty. When viewed through a
front that a critique of NPS is not in any way customer measurement practitioner lens, this
an indictment of the value of understanding claim and its supporting arguments and
and trying to manage customer word of mouth implications lead to a number of practical and
behaviors. logical concerns spelled out in the sections
Second, in evaluating NPS as a that follow.
concept, the focus is not on Reichheld as a
person or his past work. He is a prominent 1. Recommendations Alone
figure and has earned a strong favorable are not Enough
reputation in industry and academic circles.
Many of his ideas are widely cited and Obviously, customer recommend-
certainly deserve respect. This paper is ations are important, particularly in certain
strictly limited to the formulation of the NPS sectors and markets. Earning positive word
method, particularly questioning whether the of mouth communication from customers can
conceptual logic supports the nature and force be a powerful force augmenting a company’s
of the claims Reichheld has made about it. marketing efforts, especially in today’s
The paper is thus about NPS, not about “connected customer” contexts (Kirby and
Reichheld or his past scholarship. Marsden 2005). It is in fact a very noble aim
to provide the kind of excellence,
CONCEPTUAL CONSIDERATIONS differentiation, and value for the money that
IN EVALUATING NPS leads customers of their own volition to
recommend patronage of the firm – definitely
Frederick Reichheld is an established a testimony to the organization’s ability to
expert, a noted author and speaker, and effectively meet customer wants and needs.
clearly cares about advancing the topic of Thus few would argue with the premise that
customer loyalty. His frequently cited book recommendations are a good thing. But, that
on the positive effects of earning loyalty really is not Reichheld’s basic assertion. His
(Reichheld 1996a) continues to be extremely claim is that recommendations are the main
influential. However, regardless of the thing, truly the one thing that companies need
eminence of the originating source, ultimately to attain to manage and drive business success
ideas and claims should be accepted or and growth. That singular claim raises a set
rejected based on their defensibility, of logical questions.
particularly in business where large dollar Will increasing recommendations
amounts are at risk if wrongly invested. Thus really be the single best method of driving
while Reichheld in the past has been a business success? Will it have more business
powerful voice in the area of loyalty, his latest impact than reducing customer loss? If I lose
ideas about NPS (Reichheld 2003) seem less 35 percent of my customer base per year, but
tenable on a number of fronts. This paper most of those who stay would recommend,
raises questions about several specific am I really in good shape? Will
elements of Reichheld’s perspective. Primary recommendation be more powerful than
Volume 20, 2007 39
increasing current customers’ volume, cross has a high fever. The “one number,” his or
sales, or share of purchase? Will it be more her temperature, clearly is not where it should
powerful than company controllable be. A doctor having that one number may
marketing actions aimed at acquisition of now know there is a problem, but still does
targeted, profitable new customers? Will it not know what the specific problem is, and by
extend the lifetime or lifetime value of the implication, what the most appropriate
existing customer base? The core NPS treatment should be. The one number tells
premise leaves out such examples of more him something, but not everything. In fact, it
traditional thinking about customer loyalty, would be in-appropriate to rely on the one
paradoxically some of which have been raised number alone. Imagine the doctor saying,
and discussed previously by Reichheld “Your child has a fever, we must make the
himself (Reichheld 1996a). one number improve.” He has made no real
diagnosis. He has not charted a specific
2. Reichheld’s Message has Changed course for curative action based on knowing
or describing the one number.
At least on the surface, Reichheld’s The doctor must go deeper than that to
one-number claim seems to contradict his make a specific diagnosis. He must go
own past writing. He previously argued that through a number of more detailed lines of
reducing customer loss by even five percent investigation to understand the root cause of
radically multiplies profitability (e.g., the problem and to determine what best-
Reichheld 1996a). Loyalty and customer fitting course of treatment is required to move
retention were the primary focus, not the temperature number to a better place. He
recommendations. Further, he did not present has to know details about the cause of the
word of mouth as a measure of loyalty, but fever to know a fitting treatment. The
rather as an outcome of loyalty. In fact, temperature number alone gives little if any
customer recommendations were just one such actionable guidance. Certainly the
among several important outcomes springing temperature metric is an appropriate and
from loyalty. Other powerful dynamics useful indicator of health. No one would deny
discussed that seem logically more connected that. But it does not and cannot by itself tell
to revenues and profitability included: the whole story.
sustained base profit across time through The same is true with Reichheld’s
retention; increased volume; in-creased share single-question approach. However,
of purchase; additional prod-ucts and services Reichheld claims that the one number is
cross-sold; and other loyalty dynamics. sufficient in itself to drive motivating,
Somehow, those other powerful outcomes of curative organizational action.
loyalty are now supplanted by this current
emphasis on recommendations alone. The
new picture, while parsimoniously appealing, “Most customer
appears to leave out important ideas from satisfaction surveys aren’t very
prior conceptualizations. useful. They tend to be long
and complicated, yielding low
3. One Number Tells You Something, response rates and ambiguous
but not Everything implications that are difficult
for operating managers to act
A single diagnostic measure can be on.” p. 47
vitally important but not comprehensive.
Consider an analogy. Imagine that your child
40 Evaluating Reichheld’s NPS

“By substituting a knowable based on the score itself.

single question…for the Obviously more information is required.
complex black box of the Reichheld seems to acknowledge this himself
typical customer satisfaction at one point, contradicting the premise of his
survey, companies can actually one-question NPS approach.
put consumer survey results to
use and focus employees on the “Follow-up questions
task of stimulating growth.” p. can help unearth the reasons
48 for customers’ feelings and
point to profitable remedies.
It is counter-intuitive that a single But such questions should be
overall question is a sufficient basis to put tailored to the three categories
results to use, but that acting upon of customers. Learning how to
information from multiple more specific turn a passively satisfied
questions is difficult. Certainly Reichheld’s customer into a promoter
one NPS number can reveal something about requires a very different line of
a company’s overall health. However, that questioning from learning how
single score cannot provide all the to resolve the problems of a
information needed to guide targeted strategic detractor.” p. 53
improvement actions. To move the one
number upward, what specifically shall we 4. The Nature of the One Question
do? We must diagnose the underlying causal
factors that truly drive it. Reichheld offers no Reichheld’s argument is that effective
prescription for that kind of diagnosis. measurement of loyalty can center on one
Indeed, he seems to indicate that knowing the question, “How likely is it that you would
one number is sufficient in itself: recommend [company X] to a friend or
colleague?” p. 50. What seems to be missing
“The most basic surveys..can is a critical scientific measurement
allow companies to report clarification. Is that item an outcome of
timely data that are easy to act loyalty, a measure of loyalty, or a cause of
on.” p. 53 loyalty? In measurement science, ante-
cedents (causes), consequences (effects), and
“…the managerial charge, indicators (items that help to measure some
‘We need more promoters and underlying construct) are clearly
fewer detractors in order to distinguished. The distinction is vital
grow.’ The goal is clear-cut, scientifically as well as from an applied
actionable, and mot-ivating.” perspective since it shapes what should be
p. 53-54 done organizationally. Different courses of
action will be required, depending on how the
The goal may be clear-cut, but it does question is “conceptualized.” If it is an
not seem actionable with NPS alone in hand. antecedent or indicator of loyalty, we may try
Organizational change agents will be left to to drive the measure itself. If it is a
speculate about what specifically needs to be consequence of loyalty, we will try to drive
done, among all possible things that could be loyalty to make the outcome increase.
done, to really make the number go up. In Scientifically and pragmatically, the
fact, how to make the score move up is not appropriate distinction about the nature of this
Volume 20, 2007 41

item must be made clear. Yet, Reichheld’s 5. How is Loyalty Defined?

own language does not offer a clear
conceptual distinction, as evidenced by the Interestingly, despite the confusion
following quotes: about customer recommendation as an
indicator, antecedent, or consequence of
A. “…the ‘would rec-ommend’ loyalty, Reichheld does at one point put a
question generally proved to stake in the ground on a conceptual definition
be the most eff-ective in of loyalty itself. That too is a critical part of
determining loyalty and good science – providing strong conceptual
predicting growth…” p. 48 definitions of constructs under study.
However, merely being clear in stating a
B1. “…the percentage of definition does not ensure its validity.
customers who were Therefore, construct definitions need to be
enthusiastic enough to refer a scrutinized for their soundness. Reichheld ties
friend or colleague – perhaps NPS to a particular definition of loyalty as
the strongest sign of customer follows:
loyalty…” p. 48
“Loyalty is the willingness of
B2. “…such a recom- someone – a customer, an
mendation is one of the best employee, a friend – to make
indicators of loyalty…” p. 48 an investment or personal
sacrifice in order to strength-
C1. “…loyal customers talk up en a relationship.” p. 48
a company to their friends,
family, and colleagues.” p. 48 Reichheld views recommendation as
fitting that definition – as a form of sacrifice,
C2. “…what may be the ult- since the recommender’s personal reputation
imate act of loyalty, a is at stake when a referral is made.
recommendation to a friend” Recommendation certainly can fit that
p. 50 definition when considered that way. But just
because recommendation fits the chosen
Quotes A, B, and C respectively make definition, does not mean that definition really
it sound like the one question determines fits the idea of loyalty. Again, in scientific
loyalty, is an indicator of loyalty itself, and is measurement terms, it is a question of
an outcome of loyalty. Which is the case? validity. Not only does this definition differ
Does recommendation cause, indicate, or from more well-accepted conceptualizations
result from loyalty? It makes a big difference of loyalty (e.g., Jacoby and Chestnut 1978;
in terms of diagnosing how best to drive Dick and Basu 1994; Oliver 1999), but also
desired customer behaviors, and therefore logical consideration calls into question the
ultimately in terms of business action. Clear degree to which it fulfills Reichheld’s
definitions of concepts, and correct intended purpose.
specification of causal relations, are vital. Can we think of an example that also
Reichheld’s NPS approach and his discussion fits the definition, but which does not clearly
of it leave those distinctions unresolved. constitute loyalty? For instance, consider a
42 Evaluating Reichheld’s NPS

bachelor who is a “player,” dating many are primarily what lead to enhanced
women at once but committing to none. He is profitability. How then can it be that
willing to make substantial sacrifices on fancy recommendation alone can comprise the
dinners, presents, his time and effort, etc., to entirety of the loyalty picture – even if
build his relationship with each of his many someone is not continuing to purchase from
dates. That seems to meet the definition of the company at all?
sacrifice to strengthen relationships. Prevailing theory is that true loyalty is
However, it does not sound like loyalty. So both attitudinal and behavioral, and that the
from the start, there are some concerns about behavioral component is repeat purchase
Reichheld’s definition of loyalty. But it gets (e.g., Jacoby and Chestnut 1978; Dick and
more problematic as we dig even further into Basu 1994; Oliver 1999). Attitudinal loyalty
his explanation. without behavioral loyalty should not be
Reichheld reasserts, with many considered “true” loyalty (Salegna and
previous loyalty theorists, that mere repeat Goodwin 2005). Reichheld does not embrace
purchase is not the same as loyalty. this view in his case for NPS, believing
Repurchase could stem from inertia or exit someone who is attitudinally loyal but not
barriers or other reasons not really fitting our behaviorally so is just as legitimately called
natural sense of the word loyalty. However, truly loyal.
he steps completely out of more orthodox
thinking about loyalty when he argues that 6. Information in Real Time
true loyalty does not require repeat purchase.
Reichheld argues that complex survey
“…loyalty may have approaches offered by applied customer
little to do with repeat measurement firms somehow cannot offer the
purchases. As someone’s kind of real-time, technologically facilitated
income increases, she may customer feed-back that can be achieved
move up the automotive ladder through adoption of the NPS approach.
from the Hondas she has
bought for years. But if she is “The most basic sur-veys…can
loyal to the company, she will allow companies to report
enthusiastically recommend a timely data that are easy to act
Honda to, say, a nephew who on. Too many of today’s
is buying his first car.” p. 48 satisfaction survey processes
yield complex information
While repeat purchase doesn’t that’s months out of date by
constitute loyalty, it is very atypical to find the time it reaches frontline
loyalty defined without repurchase. But managers.” p. 53
according to Reichheld, as long as someone
refers the company they validly can be This claim unnecessarily ties the
labeled a loyal “customer” whether they choice of measurement approach to
purchase or not. That is fascinating given his technological sophistication. In reality, apart
previous writings (e.g., Reichheld 1996a) from NPS, widely available CRM
where he argued that the bulk of financial technologies and the proprietary “portal”
benefits of loyalty come through sustained platforms offered by most major
repeat purchase. He argued the byproducts of customer/marketing research firms offer real-
repeat purchase across the customer lifecycle time record/sample management, contact,
Volume 20, 2007 43

collection, analysis, and distribution tools That kind of interpretation makes

through sophisticated technological sense for a question about the likelihood to
applications. These tools add significant continue doing business. A lack of choice on
value by accelerating collection, analysis, the part of a customer certainly would
distribution, organizational access and use of influence how they answer such a question.
customer information. Many companies in But when it comes to positive word of mouth
partnership with marketing research firms or behavior – i.e., would recommend – there is
through their own information technology no restriction on doing so, even if there is a
solutions now have real-time customer restriction on choice. For example, one may
information, at any level of customer not be able to choose his or her electric
breakdown – by total population, segment, company but that doesn’t restrict in any way
account, and individual customer levels – positive or negative word of mouth to friends
with organization-wide distribution and and peers. Likewise, if one is using a
access to such data. Sophisticated technology system chosen by someone else
technological tools have nothing to do with (e.g., an IT manager), it in no way prevents
survey length or format and therefore should one from speaking positively or negatively
not be used as justification for one-item about the system to others. Thus even when
surveys. While such tools can be used with choice is not in customers’ direct control,
NPS, companies also may leverage these what they say to others about their
powerful technological benefits completely experiences is in their control.
independent of adopting Reichheld’s NPS Some other dynamics likely are
approach. happening in those “exception” cases, yet
Reichheld does not offer much more about
7. Interpretation of Exceptions what those other dynamics might be. Indeed,
despite observing exceptions, he still offers a
Another issue to consider, by blanket prescription for the one-item
Reichheld’s own admission (Reichheld 2003), approach. However, the exceptions show that
is that NPS was not the one thing that best the approach doesn’t work in some industries.
related to growth rates in some cases. For In the exceptional cases, we are told explicitly
such cases, he interprets the mixed pattern of other questions appeared to work much better,
findings as being due to a lack of choice in according to Reichheld himself.
those situations.
“The ’would recom-
“Asking users of the mend‘question wasn’t the best
system whether they would predictor of growth in every
recommend the system to a case. In a few situations, it
friend or colleague seemed a was simply irrelevant” p. 51
little abstract, as they had no
choice in the matter.” p. 51-52 He mentions several example
industries where the question did not seem to
“… ‘would recom-mend’ also work as well: (e.g., database software,
didn’t predict relative growth computer systems, local phone, and cable
in industries dominated by TV). Since his research is based on “more
monopolies and near than a dozen industries,” apparently, some
monopolies, where consum-ers significant percent of the time, his single item
have little choice.” p. 52
44 Evaluating Reichheld’s NPS

approach was not the best way to go. By company/product (one that logically could
Reichheld’s admission, other items provided produce recommendations), simultaneously
better information. What implication does accompanied with a lack of repurchase
that have for other industries not included in behavior.
his “more than a dozen” sample? There is at
least the possibility that his one-item 8. Manage the Cause or the Effect?
approach doesn’t work in many of those
either. Based on his mixed results, it seems In his fundamental premise, Reichheld
risky to generalize in a broad blanket argues for managing the NPS formulation
statement that NPS is the one and only because it correlates with business
number needed to grow. Yet that is what performance.
Reichheld does:

“This number is the “… the percentage of

one number you need to grow. customers who were enthus-
It’s that simple and that iastic enough to refer a friend
profound.” p. 54 or colleague…correlated dir-
ectly with the differences in
The fact that his own data reveals growth rates among com-
differences in loyalty dynamics across petitors.” p. 48
different industry sectors should imply
something more is happening than can be “…a strong correl-
captured in any single item. Even if we can ation existed between net
measure loyalty itself with a fairly simple promoter figures and a
approach, the dynamics of what causally company’s average growth
drives that loyalty clearly differs by industry. rate…Remarkably, this one
Many other academic and practitioner simple statistic seemed to
theorists have spelled out that position quite explain the relative growth
clearly. Pricing may carry different weight rates…” p. 51
depending on degree of differentiation in
market offerings. Service quality may rule “…in most industries,
the day in service-oriented industries, whereas there is a strong correlation
product-quality may rule in tangible goods. It between a company’s growth
seems risky to presume that one number can rate and the percentage of its
tell the full story and provide a course for customers who are pro-
enterprise action across the many varied moters’” p. 52
business contexts that exist.
Interestingly, if loyalty instead is a
pre-cursor to recommendation as Reichheld Based on these correlations, Reichheld
originally believed (Reichheld 1996a), implicitly concludes that a causal relationship
available theory could explain cases where is present – manage to higher positive
customers are recommending but business recommendations, and a company will
results are not indicating growth. Specifically, achieve growth. That causality is implied is
in the matrix formulation of Dick and Basu clear from his language.
(1994) some customers can have a highly
positive attitudinal state toward the
Volume 20, 2007 45

HDL cholesterol won’t drive the healthy

“… the percentage of benefits implied by a causal interpretation of
customers who are promoters the observed correlation. While they are
of a brand or company minus correlated, HDL may not be the cause.
the percentage who are Exercise for example could be the real
detractors – offers organ- underlying causal agent producing the
izations a powerful way to observed correlations. When senior adults
measure and manage customer exercise, their HDL levels increase. They
loyalty. Firms with the highest also have less heart trouble, better muscle
net promoter scores tone, better bone density, and better positive
consistently garner the lion’s emotion. It is exercise, not HDL itself that is
share of industry growth.” p. producing all the positive benefits. Trying to
53 manage the HDL number could miss the
efficacious root cause. Rather, what should
be managed is exercise itself. Then, HDL
But, scientific logic delineates the fact will go up, and so will all the other benefits.
that correlation does not necessarily imply In the customer context, what if true
causation. Certainly, if A causes B, we will loyalty is the underlying root cause of
see correlation between A and B. But if A recommendation? What if true loyalty also
and B are correlated, it doesn’t necessarily underlies increased shares of purchase,
mean that A causes B. Yet in Reichheld’s purchase of additional products and services,
discussion, he appears several times to extend resistance to competitive offers – things that
from the existence of correlation to the lead to business success? That common
interpretation of causation. underlying root cause, loyalty, thereby would
An analogy reveals why the leap to cause recommendation and business growth
causation in this case could be dangerous. factors to correlate.
Assume for example the desire to see greater Understanding the distinction as to
levels of physical health in senior adults. why things are correlated could not be more
Studying a number of factors reveals a important in its management implications.
biometric that correlates with better physical Teasing out true cause and effect, and making
health in older age: HDL (high-density sure to avoid spurious conclusions, is an
lipoprotein) cholesterol. Higher HDL is accepted fundamental of the scientific
associated with better health - less heart method. The focus of our efforts should be
trouble, better muscle tone, better bone management of the causal factor itself, not
density, better positive emotion, etc. So, if management of an outcome of the ultimate
we simply find a way to make HDL go up, causal factor. Rather than managing
will all of those positive health benefits be recommendation directly, we should be
realized? Not necessarily. managing its root cause, true loyalty. Getting
There is a plausible alternative true loyalty to increase will cause
hypothesis as to why that one number recommendations to go up and will cause
correlates with better health. HDL might be other positive indicators to go up too. It is a
the effect of some other true underlying cause very important technical distinction.
that drives both HDL and the other positive Reichheld seems to have missed that
benefits. If so, drug-based management of distinction in his article.
46 Evaluating Reichheld’s NPS

What is interesting is that in the sake of managerial control. If managers

Reichheld’s previous work (e.g., Reichheld invest in some presumed causal antecedent,
1996a) he seemed not to miss the distinction. but the causal link to the desired outcome has
His argument was that positive referrals did not been rigorously established, the
help a business grow, but it was one of investments may not produce the desired
several positive outcomes of managing and returns. In this case, investing to manage
realizing loyalty itself. The emphasis was NPS upward could possibly not lead to
not “get referrals,” it was “get loyalty and you growth. Temporal precedence is a necessary
will get referrals and a host of other economic condition in establishing a cause-and-effect
benefits.” Now his focus, and apparently his system, but it is clear that this condition has
conceptual logic, has changed. Perhaps his not been fully established in the empirical
old assertions were more plausible than his case for NPS. Yet it is evident that causality
new assertions: to grow and prosper, we is being inferred based on the language used
should manage the cause – loyalty – not the in interpreting the relationships found. For
effect – recommendations. example, in describing the airline industry,
this causality inference is implicit in the stated
9. Temporal Precedence conclusion.

A threat to scientific interpretation of “Remarkably, this one

true causality emerges in another place in the simple statistic seemed to
paper. Reichheld (2003) presents research explain the relative growth
done in collaboration with a customer rates across the entire
measurement and technology firm and draws industry; that is, no airline has
conclusions based on data that do not meet found a way to increase
one of the fundamental conditions required to growth without improving its
infer causality. Namely, if X causes Y, then ratio of promoters to
X must occur before Y. However in the detractors.” p. 51
article, that condition is not adhered to, and
yet a causal explanation is still offered. Not only is the claim incorrect
Specifically, Reichheld builds his case because of the technical issue, it is logically
for the causal connection between corporate false. It implies several carriers in the airline
growth and NPS using data that does not fully industry have already been managing by the
meet required conditions of temporal net promoter number and any and all who
precedence. The measure of corporate growth have realized growth, have done it only by
spanned a window from 1999-2002. The increasing net promoter scores.
survey-based measure of customer Net promoter scores from 2001
recommendation intention did not start until predicting growth partially drawn from 1999
2001. Thus answers to a forward-looking and 2000 is a threat to validity in the
2001 “likelihood to recommend” measure are conclusion of causality, and therefore a risk
predicting growth observed in part in 1999 for the applied manager wishing to implement
and 2000. That means something that actions that drive growth.
happened in the future is being used as a
cause of something that happened in past.
This again is a technical point, but it is
an important one in assessing the validity of
research that claims to tease out causality for
Volume 20, 2007 47

10. Questions about “promoters:” those who scored 9 and 10;

Dynamic Interpretation “passively satisfied:” those who scored 7 and
8; and “detractors:” those who scored 0-6.
Another problematic element of the That categorization is claimed to be the best
empirical research used to justify the claim way to use the 0-10 ratings.
that “one number is all you need” is the
matching of a cross-sectionally measured

antecedent with a longitudinally measured “…three categories…turn out

outcome. Changes in company growth were to provide the simplest, most
measured longitudinally, yet the NPS scores intuitive, and best predictor of
were not. This raises questions about the customer behavior…” p. 51
appropriateness of claiming that upward
changes in the net promoter score explain Several concerns and questions arise
upward changes in growth. That really is not regarding this formulation. First, only two of
something Reichheld studied, and therefore three of the collapsed categories are used
should not be something about which such subsequently to compute the “net promoter
definitive conclusions are stated. Yet score” (promoter percent minus detractor
Reichheld generalizes just such an percent). Second, one could ask why 6s are
interpretation of the data across most of the included in the “detractors” group. A rating
industries studied. of 5 is the “midpoint” of the scale, so 6s are
on the positive side of “neutral.” Why
“...no airline has interpret them as negative? Third, how is it
found a way to increase that the collapsed categories are the “best
growth without improving its predictor of customer behavior” when one
ratio of promoters to normally would expect the increased
detractors. That result was variability of more scale points to allow for
reflected, to a greater or lesser better prediction? Fourth, it is not clear
degree, in most of the whether these particular break points to create
industries we examined…” p. the three groups were chosen arbitrarily,
51 based on some conceptual logic, or
empirically. Finally, this particular approach
11. Scaling the One Question to categorization and the subsequent use of
two collapsed categories to create the NPS
Reichheld makes an argument for an measure involves computations that allow
11-point scale ranging from 0 to 10, with several very different scenarios to produce the
verbal labels at the ends of the scale and at the same net promoter score.
midpoint. Typically, scales like that are used
to try to ensure variability in responses and to 12. Examining the Math
more confidently presume the ratings are at a
sufficient level of measurement to warrant As described, customers are divided
certain statistical analyses (technically, to into one of three categories, and then two of
assume “interval-level” measurement). those categories are used to calculate the net
However, what Reichheld does with ratings promoter score. NPS is the percent
from this scale is to collapse them into a “promoters” – those most likely to
categorical measure with three levels:
48 Evaluating Reichheld’s NPS

recommend – less the percent “detractors” – The one NPS number in isolation
those less likely to recommend. Bigger requires more information to appropriately
numbers obviously are better. In the interpret and act upon – even the very
extremes, 100 percent promoters and 0 numbers that went into the single score itself.
percent detractors would yield a net promoter Without that minimal extra information, there
score of 100. Zero percent promoters and 100 is no hope of knowing the rest of the story. In
the example, Company B relative to A has ten
percent detractors would yield a minus 100 times as many promoters, and simultaneously
net promoter score (-100). ten times as many detractors. These are
A concern with this math is that very totally different situations. The implication of
different scenarios can produce precisely the Reichheld’s assertions, however, is that both
same result. Thus, while completely different patterns should lead to the same basic growth
management actions are likely to be called for in the market. It is even more dangerous to
under different scenarios, the net promoter think about Reichheld’s suggestion that we
score in itself – the one number Reichheld should be comparing regions, branches,
says you need for management – will not customer segments, even against competitors’
expose those differences. scores on this one number. That seems risky
For example, consider two different given that vastly different scenarios will
contrived, but possible, company scenarios to produce the same net promoter numbers.
demonstrate the point. Imagine Company A Likely, even knowing the component scores
scenario with 5 percent promoters, 90 percent for NPS will not be enough. Beyond the
passively satisfied, and 5 percent detractors. potential problems posed by the NPS math,
We compute the net promoter score to be zero likely, more will be required to truly
(5% - 5%). Now imagine Company B understand which underlying forces might be
scenario with 50 percent promoters, 0 percent changed or improved to drive the net
passively satisfied, and 50 percent detractors. promoter components in the desired direction.
Again the net promoter score is zero (50% -
50%). Two completely different situations 13. What is New at the Core?
produce the same net promoter score, but
logically require very different managerial Reichheld (2003) does not provide all
actions. the details of his empirical work, but the
In scenario A, a very small minority is reader is informed that measures of actual
divided in the extreme ends of the three- customer repeat purchases and actual
category distribution, while the company is customer recommendations were used.
doing a mediocre job for most of its
customers. In scenario B, the company is “…my colleagues and I
doing great with half of its customers, and not looked for a correlation
great with the other half, essentially between survey responses and
producing a two-group distribution divided actual behavior – repeat
exclusively into the extreme end categories of purchases or recom-
the NPS components. Completely different mendations to friends and
scenarios, exactly the same net promoter peers…”p. 50
scores. Shall we expect the same market
performance for the two companies? Will the It is unclear from the description
same management action be required in both whether they looked for correlation with
cases? It seems unlikely on both accounts. actual repeat purchases in isolation, or actual
Volume 20, 2007 49

recommendations in isolation, or some measurement and management. For example,

combination of the two metrics. However it Reichheld states:
was done, behavioral referrals and
continuation served as dependent / criterion “Companies won’t
variables. It sounds as if he and his realize the fruits of loyalty
colleagues tracked actual purchases and until usable measurement
referrals, then searched for survey questions
that were most highly correlated with those
measurements. He reports that two of the systems enable firms to
three top predicting items were survey-stated measure their performance
likelihood to recommend and survey-stated against clear loyalty goals –
likelihood to continue purchasing. In other just as they now do in the case
words, survey reported intentions to exhibit of profitability and quality
certain behaviors (recommendation and goals.” p. 49
continuation) were what correlated most with
observance of those behaviors (actual Clearly companies do need usable
recommendation and continuation). Is this measurement systems to monitor progress in
really a new contribution to our understanding their efforts to achieve loyalty goals. Other
of customer behavior, or is the fundamental points raised by Reichheld in the past and in
empirical basis of NPS more along the lines this article also fit well with current thinking
of something social scientists and market about customer loyalty. Some examples
researcher already widely accept, namely that include: available technological systems
behavioral intention questions correlate with greatly facilitate measurement and
actual behavior? And if the empirical management of loyalty; the measurement of
evidence for NPS really boils down to a customer satisfaction is not enough;
simple replication of intention-behavior satisfaction is not the best predictor of repeat
correlation, aren’t we back to conceptual purchase and loyalty; customer loyalty is
questions about behavior intentions not being about more than just repeat purchase; repeat
a sufficient basis for measuring and managing purchase can be simply from inertia,
customer loyalty (e.g., Jacoby and Chestnut
1978; Dick and Basu 1994; Oliver 1999; sub-segment of the customer base typically
Grisaffe 2001)? can be called truly loyal, and other points
from various instantiations of Reichheld’s
thinking (e.g., Reichheld 1996a, 1996b, 2001,
AGREEMENTS AND DISAGREEMENTS 2003).
REGARDING NPS However, other claims and premises
Reichheld has advocated in his work on NPS
After all the previous questions and do raise some questions and concerns: is
issues have been raised about Reichheld’s one tracking one number all that is needed to
number, it should be noted that those points measure and manage customer loyalty; is
do not constitute a blanket rejection of “would recommend” the best indicator of
everything Reichheld (2003) stated in his loyalty; can a single survey item provide
article. Without embracing all his points enough information to be actionable; does
about NPS, other points clearly do fit within real-time customer feedback in tech-
widely accepted views about customer loyalty nologically-driven systems depend on one-
50 Evaluating Reichheld’s NPS

item surveys; are empirically-derived Anderson 1998; Luo 2007; Wangenheim

correlations enough of a solid basis to imply 2005; Ward and Ostrom 2006).
causation; is an 11-point scale, collapsed to The issue with NPS however, is not
three categories a best practice in “loyalty” the importance of word of mouth, but rather
measurement; what is the risk of adopting a Reichheld’s claim that NPS is the one and
conceptual definition or computational only thing companies need to monitor and
approach under which plausible cases can be manage to realize success. There is a
conceived which appear to contradict the difference between a concept/construct being
intent of the specification or calculation? one important thing, versus it being the one
Thus, after careful examination of the case important thing. When being claimed as the
advocated by Reichheld in his recent article one important thing, at a minimum,
(Reichheld 2003), an alternative conclusion Reichheld’s claims can and should be
can emerge which is different from his own. scrutinized for validity, both on empirical
Namely, it is unlikely that one number is all a grounds, and on conceptual grounds.
company needs to effectively measure and Whereas recent papers have begun to address
manage customer loyalty. empirical testing (e.g., Keiningham et al.
2007; Marsden et. al 2005; Morgan and Rego
2006), this article has evaluated NPS
CONCLUSION conceptually by examining elements of
Reichheld’s original description of the NPS
Ideas promoted by Frederick development. This careful evaluation of the
Reichheld have done much across the years to conceptual bases of his original work raises
advance organizational attention to a some lingering concerns.
collection of important topics around In light of all this, how then should
customer loyalty and loyalty measurement NPS be viewed? NPS ultimately may be
(e.g., Reichheld 1996a, 1996b, 2001). something more like a dashboard light. A
Recently, Reichheld has advocated measure- dashboard light is valuable and does reveal
ment and management of a single customer important information about the working of
metric – the Net Promoter Score (Reichheld the car. When the indicator gives warning,
2003; 2006a; 2006b). The rapid adoption of we know we are in trouble. However, we
his NPS concept by a number of extremely don’t then proceed to manage the light (e.g.,
prominent companies (Creamer, 2006; unscrew the bulb to make the warning go
Keiningham et al. 2007), perhaps in response away). We diagnose the root cause in the
to Reichheld’s own advocacy and some engine and address that fundamental
degree of subsequent “contagion,” underlying issue. Once the root problem is
anecdotally supports his core premise – word fixed, the indicator light then takes on healthy
of mouth can have powerful effects in status again. So with NPS, it may be a
creating growth. Then again, to claim or valuable, applied diagnostic metric. But in
demonstrate effects of positive and negative itself, on its own, it probably is not the one
components of word of mouth is really not so and only thing a company needs to manage
new or revolutionary. The importance of for success. Neither is it likely to be an ideal
positive and negative word of mouth operationalization of generally accepted
behaviors has long been emphasized in theoretical formulations of the loyalty
marketing and consumer research, both in concept.
past years (e.g., Brown and Reingen 1987;
Richins 1983) and in recent years (e.g.,
Volume 20, 2007 51

Perhaps when all the conceptual issues and while its figurehead is widely respected,
have been considered, and all the empirical in light of conceptual considerations present-
data are in, NPS will retain a revised version ed here and elsewhere, perhaps a more multi-
of its “one number” status – “one number dimensional perspective on customer loyalty
among several.” While admitting additional metrics ultimately will win the day.
complexity, perhaps a degree of additional
plausibility also would accompany a more
holistic and complete multidimensional
system of indicators of the health and strength
of a company’s relationships with its
customers (Grisaffe 2000). Conceptualization
and operationalization of such a collection of Endnote: A number of points presented in
indicators, perhaps including NPS, should be this article are drawn from an earlier version
viewed contextually in relation to a) of a “white paper” written by the author while
controllable organizational actions that can Vice President and Chief Research
causally drive various metrics upward, and b) Methodologist of the customer measurement
the consequent customer behaviors that consulting firm Walker Information of
demonstrably drive firm financial Indianapolis, Indiana. The original version
performance. Also included could be other was published online in March 2004 (Grisaffe
functional and competitive indicators of 2004b) shortly after Reichheld’s original
health, strength and growth. Helping article appeared. Updated material was added
companies succeed financially in the long run to the introduction and conclusion sections of
would thus involve understanding, measuring, this article to reflect literature that has
and managing the total system of indicators emerged on the topic. The content of the
capturing customer experiences with, attitudes current version also reflects a number of
toward, and responses to, the total offering of changes and improvements recommended by
the organization as a collective (e.g., the helpful comments of three anonymous
products, services, and other intangibles) reviewers and the Editor. The author
(Grisaffe 2000). Certainly, the specifics of expresses thanks to these individuals for their
any such system also should be subject to guidance in creating a better and more
conceptual and empirical testing, even academically-fitting paper, and to Walker
precisely against the notion of a singular NPS Information for permission to republish
metric. While there is a tempting simplicity content from the original document.
of a “1 number needed for growth” approach,
52 Evaluating Reichheld’s NPS

REFERENCES Kirby, Justin, and Paul Marsden (2005),

Connected Marketing: The Viral, Buzz and
Anderson, Eugene W. (1998), “Customer Word of Mouth Revolution, London:
Satisfaction and Word of Mouth,” Journal of Butterworth-Heinemann.
Service Research, 1, 5-17. Kristensen, Kai, and Anders Westlund (2004),
Brandt, D. Randall (2007), “For Good Measure,” “The One Number You Need to Grow,”
Marketing Management, 16, 20-25. Harvard Business Review, 82, 136-137.
Brown, Jacqueline Johnson, and Reingen, Peter H. Luo, Xueming (2007), “Consumer Negative
(1987) “Social Ties and Word-of-Mouth Voice and Firm-Idiosyncratic Stock Returns,”
Referral Behavior,” Journal of Consumer Journal of Marketing, 71, 75-88.
Research, 14, 350-362. Marsden, Paul, Alain Samson, and Neville Upton
Creamer, Matthew (2006), “Do you Know Your (2005), “Advocacy Drives Growth: Customer
Score: Your Business Rides on How Advocacy Drives UK Business Growth,”
Consumers Answer the One Crucial http://www.lse.ac.uk/collections/
Question,” Advertising Age, 77, 1, 24. pressAndInformationOffice/PDF/AdvocacyDr
Crosby, Lawrence A., and Sheree L. Johnson ivesGrowth_5-9-05.pdf.
(2007), “The Endless Debate,” Marketing Morgan, Neil A., and Lopo L. Rego (2004), “The
Management, 16, 14-15. One Number You Need to Grow,” Harvard
Dick, Alan S., and Kunal Basu (1994), "Customer Business Review, 82, 134-136.
Loyalty: Toward an Integrated Conceptual Morgan, Neil A., and Lopo Leotte Rego (2006),
Framework," Journal of the Academy of “The Value of Different Customer
Marketing Science, 22, 2, 99-113. Satisfaction and Loyalty Metrics in Predicting
Grisaffe, Doug (2000), “Putting Customer Business Performance,” Marketing Science,
Satisfaction In Its Place: Broader 25, 426-439.
Organizational Research Perspectives Versus Nicks, Stephen (2006), “What Not to do with Net
Measurement Myopia,” Journal of Consumer Promoter,” Business Week Online, Aug
Satisfaction, Dissatisfaction and Complaining 1,www.businessweek.com/smallbiz/content/ju
Behavior, 13, 1-16. l2006/sb20060731_349936.htm?campaign_id
Grisaffe, Doug (2001), “Loyalty - Attitude, =rss_null.
Behavior, and Good Science: A Third Take Oliver, Richard L. (1999), “Whence Consumer
on the Neal-Brandt Debate,” Journal of Loyalty?, ‘ Journal of Marketing, 63 (Special
Consumer Satisfaction, Dissatisfaction and Issue), 33-44.
Complaining Behavior, 14, 55-59. Pingitore, Gina, Neil A. Morgan, Lopo L. Rego,
Grisaffe, Doug (2004a), “Guru Misses the Mark Adriana Gigliotti, and Jay Meyers (2007),
with ‘One Number’ Fallacy,” Creating “The Single-Question Trap,” Marketing
Loyalty,Feb.http://www.creatingloyalty.com/st Research, 19, 9-13.
ory.cfm? article_id=656. Reichheld, Frederick F. (1996a), The Loyalty
Grisaffe, Doug (2004b), “When One Equals Zero: Effect: The Hidden Force Behind Growth,
A Response to Frederick Reichheld,” A Profits, and Lasting Value, Boston, MA:
Walker Information White Paper. Harvard Business School Press.
Indianapolis, IN: Walker Information. Reichheld, Frederick F. (1996b), “Learning from
Jacoby, Jacob, and Robert W. Chestnut (1978), Customer Defections,” Harvard Business
Brand Loyalty: Measurement and Review, 74, 56-69.
Management, New York: John Wiley & Sons. Reichheld, Frederick F. (2001), Loyalty Rules:
Keiningham, Timothy L., Bruce Cooil, Tor Wallin How Today’s Leaders Build Lasting
Andreassen, and Lerzan Aksoy (2007), “A Relationships, Boston, MA: Harvard Business
Longitudinal Examination of Net Promoter School Press.
and Firm Revenue Growth,” Journal of
Marketing, 71, 39-51.
Volume 20, 2007 53

Reichheld, Frederick F. (2003), “The One

Number You Need to Grow,” Harvard
Business Review, 81, 46-54.
Reichheld, Frederick F. (2004), “The One
Number You Need to Grow,” Harvard
Business Review, 82, 133.
Reichheld, Fred (2006a), “The Microeconomics
of Customer Relationships,” MIT Sloan
Management Review, 47, 73-78.
Reichheld, Fred (2006b), The Ultimate Question:
Driving Good Profits and True Growth,
Boston, MA: Harvard Business School Press.
Reichheld, Fred (2006c) "The Ultimate Question:
Driving Good Profits and True Growth,"
Keynote presentation at Master the Art of
Word of Mouth, Viral, Buzz, and Blog
Marketing, WOMMA: Word of Mouth
Marketing Association, Orlando Florida.
Richins, Marsha L. (1983), “Negative Word-of-
Mouth by Dissatisfied Consumers: A Pilot
Study,” Journal of Marketing, 47, 68-78.
Salegna, Gary J. and Stephen A. Goodwin (2005),
“Consumer Loyalty to Service Providers: An
Integrated Conceptual Model,” Journal of
Consumer Satisfaction Dis-, satisfaction and
Complaining Behavior, 18, 51-67.
Wangenheim, Florian V. (2005), “Postswitching
Negative Word of Mouth,” Journal of Service
Research, 8, 67-78.
Ward, James and Amy Ostrom (2006),
“Complaining to the Masses: The Role of
Protest Framing in Customer-Created
Complaint Web Sites,” Journal of Consumer
Research, 33, 220-230.

Send Correspondence Regarding This Article to:

Douglas B. Grisaffe
Assistant Professor of Marketing
University of Texas at Arlington
Box 19469
Business Building, 218
Arlington, Texas 76019-0469
Telephone: (817) 272-0772
Fax: (817) 272-2854
E-Mail: [email protected]