How to Use Big Data to

Drive Your Supply Chain

Nada R. Sanders

Big data analytics has become an imperative for business leaders across every industry sector. Analytics appli-
cations that can deliver a competitive advantage appear all along the supply chain decision spectrum—from
targeted location-based marketing to optimizing supply chain inventories to enabling supplier risk assessment.
While many companies have used it to extract new insights and create new forms of value, other companies
have yet to leverage big data to transform their supply chain operations. This article examines how leading
companies use big data analytics to drive their supply chains and offers a framework for implementation
based on lessons learned. (Keywords: Decision Making, Supply Chain, Technology)

O
ver the past few years “big data” has emerged as the new frontier
of IT-enabled innovation and is a subject in the forefront of busi-
ness, science, and society as a whole.1 It is virtually impossible to
open a publication without seeing some reference to big data,
new analytics breakthroughs, and novel insights. A recent article even refers to
data scientists as the “sexiest job of the 21st century.”2 While the academic and sci-
entific sectors are looking to big data for unprecedented opportunities to better
understand the world, businesses are looking for a technology-based competitive
advantage. As a result, big data analytics has become an imperative for business
leaders across every industry sector—from healthcare to manufacturing to aero-
space.3
The ability to capture, store, aggregate, and analyze data—and then extract
intelligence—is now rapidly becoming a mandate for virtually all organizations.
Leading-edge companies across the globe have reported benefits in their use of
big data. Consider Walmart, eBay, Progressive Insurance, and Target.4 These com-
panies are succeeding in this game-changing environment by embracing and lead-
ing the change. They have used big data analytics to extract new insights and
create new forms of value. After all, it is through big data that Walmart learned
that customers prefer to stock up on the sugary treat Pop Tarts during a hurri-
cane;5 eBay identified which web designs generate the highest sales;6 Progressive

26 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

Insurance learned how to optimize insurance pre-

Nada R. Sanders is the Distinguished
miums by risk category,7 and Target learned how Professor of Supply Chain Management at
to identify a pregnant customer.8 the D’Amore-McKim School of Business at
Northeastern University. <n.sanders@neu.
Analytics applications that can deliver a
edu>
competitive advantage appear all along the supply
chain decision spectrum—from targeted location-based marketing to optimizing
supply chain inventories to enabling robust supplier risk assessment. Supply
chains generate huge amounts of data that companies can turn into intelligence
through analytics, from POS, GPS, and RFID to social media feeds. Success stories
of companies that have harnessed the power of big data analytics abound, such as
Amazon and Walmart. However, most companies have yet to leverage big data
analytics to transform their supply chain operations.9 The vast majority of compa-
nies are not the Amazons and the Walmarts. For them, acquiring new technolo-
gies and software is costly with an unclear ROI. Many are awash in data but are
unsure how to use it to drive their supply chains.10 Furthermore, many are
engaging in fragmented utilization or implementation rather than a systematic
and coordinated effort. The results are unclear benefits, including lack of real
insight and competitiveness. The truth is that except for a few large companies,
particularly those that are information-rich such as Linkedin, Facebook, and Google,
most are struggling with deciding what to do and how to proceed.11
In order to extract the lessons of how leading companies use big data analytics
to drive their supply chains, we conducted a multistage study over a period of
two years. We first conducted exploratory interviews with executives and managers
across a wide range of industries to identify critical challenges.12 This was followed by
a survey of senior managers and executives to identify specific implementation strat-
egies and barriers.13 Finally, we conducted in-depth case studies of four leading
organizations that have successfully implemented big data analytics. Collectively, this
process served to create an understanding as to the extent of use of big data across
a wide range of organizations and identify common barriers to implementation.
Further, we learned how leading companies use big data, commonalities of success-
ful implementation, and strategies for overcoming barriers.
Our research resulted in three key findings. First, we present a prescriptive
framework showing how companies can implement big data analytics to drive
supply chains. Through the use of case studies and interviews the framework
identifies hurdles of implementation and shows how exemplar companies do this.
Analytics driven companies—Amazon, Walmart, Dell, and UPS—do all these
things in a coordinated way as part of an overarching strategy championed by
top leadership and pushed down to decision makers at every level. Although deep
analytics capabilities are being developed for sub-functions along the supply
chain, and are receiving much hype, they are often hyper-specialized. Unless
the insights offered by these functional applications can be linked to the rest of
the supply chain, their output may not provide tangible benefits. Second, we pres-
ent stages of implementation through a maturity map. Proceeding through these
stages ensures data quality, refinement of key metrics, and timely access to infor-
mation by decision makers. We caution against following the hype and jumping
into large-scale implementation without proceeding through the stages, given

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 27

 How to Use Big Data to Drive Your Supply Chain

the need for organizational learning and continuous improvement. Third, we iden-
tify key issues companies should consider in outsourcing analytics. Big data analyt-
ics applications increasingly require deep technical knowledge that is not the core
competency of most companies and we identify considerations in outsourcing this
capability.

What is Different?
Big data without analytics is just a massive amount of data. Analytics without
big data are simply mathematical and statistical tools and applications. Many of these
tools have been around for decades, such as correlation and regression analysis. It is
the combination of big data and analytics, fueled by today’s computing power, which
creates the ability to extract meaningful insights and turn information into intelli-
gence. As Google’s director of research, Peter Norvig, noted: “We don’t have better
algorithms. We just have more data.”14 However, this is only partially true. Actually,
the availability of big data and the advancements in machine intelligence have
created significant new opportunities for both the data and the development of
algorithms. Based on an extensive review of scholarly research we identify what
is different today.15
The first difference is the unprecedented opportunity for inquiry. The rapid
pace at which all types of transactions, both economic and social, are moving
online has enabled real-time digital capture of data. Certainly the notion of using
analytical techniques to derive insights from data is as old as the field of statistics.
However, the ability to capture and understand the content of human dialogue
has expanded the types of data sets that can be analyzed. Large and complex data
sets related to any type of phenomenon researchers want to study are now readily
available. This can range from deconstructing the human genome to learning how
the Internet affects financial markets16 to learning how to optimize the timing of
push-notifications to targeted customers with mobile analytics.17
The second difference is that the nature of inquiry has changed. In the past,
data was collected with the aim of testing a human-generated hypothesis. Today,
data are being collected for the possibility of testing hypotheses that have not yet
been conceived and many of which are generated via computer. Machines are
becoming smarter through self-teaching algorithms, continuously being fed data
on a large scale. As such, the algorithm is able to detect every type of relationship
between the variables, such as behaviors and demographics, asking interesting
questions and refining them without active human intervention. As a result, the
computer has now become an active participant in the inquiry process and is
becoming capable of creating new knowledge and making discoveries on its
own.18 In medicine, for example, we are seeing computer learning used to iden-
tify unknown genes from prior knowledge of gene functions.19 A large body of
this research is being devoted solely to developing deep technical capabilities for
machine learning.20
The third difference is that the nature of experimentation has changed. The Inter-
net has created the ability to conduct large-scale experiments on many economic and
social phenomena. One example is the ability of LinkedIn to turn granular structured

28 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

and unstructured data into a quantification of member attitudes just by observing

texts.21 The company uses a home grown tool to “listen” to what members are talk-
ing about and is able to quantify the full range of member reactions including “pain
points” and “moments of delights,” enabling LinkedIn to turn data into insights. Sim-
ilarly, Facebook researchers conducted a massive study determining that they could
manipulate the mood of users.22 These are essentially controlled experiments con-
ducted on large numbers of people enabling the understanding of an unprecedented
number of variables. Certainly there are privacy and ethical questions with conduct-
ing these experiments. However, the bigger point is that the ability for massive
experimentation has changed the manner in which it can be conducted and the
insights that can be gained.
These three distinct differences of big data analytics point to it being one of
the most significant “tech” disruptions in history. Big data is unique because of the
volume, variety, and velocity of the data, which today is widely available and
much less expensive to access and store.23 However, our analysis of big data
research developments shows that the majority of these advancements are highly
functionally specific. A recent study identified five areas as having a “big impact”
for big data research.24 They are: e-commerce and market intelligence; e-govern-
ment and politics; science and technology; health and medicine; and security and
public safety. Of these, market intelligence is the only area directly related to sup-
ply chain management. Further, while providing deep insights, these analytics
applications are highly fragmented. This fragmented nature of application devel-
opment and inquiry leaves little opportunity for systematic implementation. This
may be less problematic in other fields such as medicine where, for example, ana-
lytics can be used to answer targeted questions such as identifying hospital read-
mission for patients with congestive heart failure. However, it creates special
challenges for executives who seek to improve entire organizational systems
rather than merely one function.

Supply Chain Analytics Applications

Analytics applications are available along the entire supply chain spectrum—
from source to sell (Exhibit 1). The largest growth has been seen on the marketing
side, with applications constantly evolving to gain better market intelligence. Logis-
tics has used analytics applications for routing and vehicle scheduling for years, while
operations has been using applications that optimize operational requirements, from
inventory and capacity to labor scheduling. Although lagging behind marketing,
applications for both logistics and operations are increasingly growing in complexity
and detail, with next generation algorithms adding unprecedented capabilities.
Lastly, sourcing is increasingly seeing applications that segment suppliers, measure
risk, and inform supplier negotiations. Although sourcing has lagged in application
development compared to marketing and other supply chain functions it is fore-
casted to experience the largest growth over the next few years.25
These analytics applications offer deep insights and tend to be functionally
specific. An example is a new capability from GE that provides enhanced asset reli-
ability for clients by optimizing asset inspection, maintenance, and repair.26 Typical

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 29

 How to Use Big Data to Drive Your Supply Chain

EXHIBIT 1. Analytics Applications Across the Supply Chain

Inventory Optimization Location-Based Marketing

Capacity Constraints In-Store Behavior Analysis
Facility Location Customer Micro-Segmentation
Facility Layout Multichannel Marketing
Workforce Analytics Assortment Optimization

SOURCE MAKE MOVE SELL

Supplier Risk Distribution & Logistics Optimization

Product Characteristics Transportation Alternatives

Sourcing Channel Options Routing

Supplier Integration Level Scheduling

Supplier Negotiation Vehicle Maintenance

of these applications the knowledge and insights provided are deep and the focus
hyper-specialized.

Marketing Applications
Marketing analytics applications are customer oriented and are on the sell side
of the supply chain. The nature of marketing has driven the development of big data
applications that focus on capturing customer demand, enabling micro-segmentation,
and predicting consumer behavior. In fact, micro-segmentation has become a highly
important application of big data analytics. Although market segmentation has long
been a marketing capability, the coupling of big data with sophisticated analytic tools
has enabled micro-segmentation at increasingly granular levels.27 Companies can
now use technology to gather and track data on the behavior of individual customers,
and then combine these with traditional market research tools to gain greater
insight. The collected data is increasingly tracked in real time, enabling companies
to quickly readjust their customer strategies. This is seen with retailers such as Neiman
Marcus where behavioral segmentation is matched with a multi-tier membership
rewards program.28 The company uses sophisticated analytics to identify key custom-
ers and then creates targeted purchase incentives resulting in higher margin pur-
chases from the company’s higher-margin customers.

30 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

Another important marketing application is in price optimization.29 Price

optimization has moved to a new level, permitting analysis at a high granularity
of data on pricing and sales. Pricing decisions can now be made in near real time
using a variety of data sources, examples of which have become common in hos-
pitality. Marriott International, for example, uses a sophisticated analytics system
to optimize prices for guest rooms considering variables such as customer type and
even the weather.30
Marketing applications are not restricted to traditional business-to-consumer
(B2C) relationships. Although B2B offers special complexities, such as fewer cus-
tomers characterized by infrequent but large orders, algorithms have been devel-
oped to track, segment, and better understand the spend of business customers.
Further, these applications are becoming ever more sophisticated, such as provid-
ing “affinity analysis” that identifies which products a business customer is willing
to buy from the company rather than its competitors.

Logistics Applications
Logistics applications help move goods through the supply chain and are
some of the oldest. They have been used for optimizing inventory, identifying
optimal distribution center locations and supply routes, and minimizing transpor-
tation costs. One highly utilized application is in transportation and routing.
GPS-enabled big data telematics and route optimization are being used to optimize
transportation.31 Further, analytics applications can improve productivity by opti-
mizing fuel efficiency, preventive maintenance, driver behavior, and vehicle rout-
ing. Tracking of disruptive events, such as weather, can be continually updated
and routes re-optimized in real time. An example is UPS, which started gathering
such data more than 20 years ago. The company uses an analytics tool called
ORION (On-Road Integrated Optimization and Navigation) that enables drivers
to find the most efficient route in their delivery areas.32
Applications can provide granular information such as segmenting transpor-
tation routes and including transit factors for different types of products.33 An excel-
lent example is in inventory management where RFID technology has been useful in
tracking inventory in motion, capturing both location and quantity, as well as moni-
toring security. The technology is particularly important in the “cold-chain” where it
is used to track ambient temperature and transit duration. These are especially
important when transporting perishable items where the algorithms optimize order
quantities as well as service levels with transit times, and can create alerts as soon
as a problem is detected.

Operations Applications
Analytics applications are used in operations for a range of decisions—from
inventory management and optimization of stock levels, to maintenance optimiza-
tion and facility location. Measuring productivity and quality are finding large
applications where companies can now run daily analyses of performance.34 These
statistics can be aggregated and reported on a more granular level, such as by store
sales, SKU sales, and sales per employee. Although such applications have been
available for years, what is now different is scale. These systems are moving ever

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 31

 How to Use Big Data to Drive Your Supply Chain

closer to real time where they can alert companies of problems, including real-time
changes in productivity or quality. Current applications enable granular reporting
of statistics at high frequency, allowing managers to make more targeted real-time
adjustments.
Workforce analytics has become an especially important area in operations.35
Current technologies can reduce costs while maintaining service levels by optimizing
labor, automating and tracking attendance, and improving labor scheduling.
Retailers can analyze cashier performance at granular levels such as transactions
per hour. At call centers, managers can analyze quality of customer service based
on customer complaints, satisfaction surveys, or the percentage of customer
issues solved with a single call. Analytics can also optimize staffing needs by
matching forecasts with labor optimization, which is especially beneficial during
peak demand periods.

Sourcing Applications
Analytics applications are becoming increasingly important in sourcing and
are the fastest growing. Consider that in most manufacturing organizations sourc-
ing represents the largest single category of spend for the company, ranging from
50 to 90 percent of revenue.36 Therefore applying analytics here can yield great
savings. A number of companies report using analytics to optimize sourcing chan-
nel options and integrate suppliers into their own operations. Some applications
segment suppliers based on key characteristics, helping with sourcing strategy
and balancing cost versus risk. Amazon, for example, uses analytics to determine
the optimal sourcing strategy and manage all the logistics to get a product from
manufacturer to customer. Analytics are used to determine the right mix of joint
replenishment, coordinated replenishment, and single sourcing. In fact, Amazon
applies advanced analytics across its fulfillment, capacity expansion, inventory
management, procurement, and logistics functions in order to coordinate all sup-
ply chain management processes.37
Analytics can also be used to support supplier negotiations by providing a
thorough analysis of customer preferences and buying behaviors. This informa-
tion can be used to inform negotiations with suppliers by providing factual lever-
age. For example, companies can use information on prices and transactions to
negotiate concessions on key products. Lastly, analytics is being used for tail-end
spend, automating the process of managing this category as companies increas-
ingly recognize its sizable monetary value.

Linking Applications Across Functions

These state-of-the art analytics applications exemplify the three distinct differ-
ences of today’s big data analytics capability. We see an unprecedented opportunity
for inquiry (e.g., consider that many retailers use facial-recognition cameras hidden
in the eyes of mannequins to track shoppers, learn their behaviors, and increase
sales, such as a “surveillance system” being utilized by Walmart, called Shoppercep-
tion38); a change in the nature of inquiry (e.g., Tesco uses a self-learning Bayesian
system for use in its online business called Tesco Direct that recommends particular
grocery items customers may want); and a change in the nature of experimentation

32 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

(e.g., retailers are using sentiment analysis to gauge the real-time response to mar-
keting campaigns and adjust business course accordingly; both Starbucks and the
Gap changed their logos using sentiment analysis in response to social media in real
time39). This array of application capability is vast and even overwhelming. The
insights they provide are deep, highly specific, and often hyper-specialized. Our case
studies show that although leading companies rely heavily on these applications,
they use them to drive their supply chains in a functionally linked manner as part
of a coordinated overarching strategy. They do not select applications to optimize sin-
gle functions or single decision areas in isolation. This is an important distinction.
One example of how analytics applications link the supply chain is illus-
trated by Walmart. Walmart is a leader among supply chain analytics competitors.
Indeed its success as the world’s largest retailer is at least in part based on the state-
of-the-art analytics applications the company uses. These applications are used to
link functions and coordinate its entire global supply network. The company col-
lects more than 1 million customer transactions every hour,40 all of which goes into
a single integrated technology platform.41 Sophisticated analytics is applied to this
massive database and used across the supply chain. The collected information is
then analyzed and used by managers to support every type of supply chain deci-
sion. At the store level, managers use the system to analyze detailed sales data
and optimize product assortment. The system also enables inclusion of qualitative
factors that help tailor assortments to local preferences. This ensures that customers
have the products they want, when and where they want them.
Using analytics, the company has learned a great deal about customer pref-
erences, some of it novel and surprising. For example, they learned that before a
hurricane consumers stock up on food items that don’t require cooking or refrig-
eration. A case in point is the algorithm uncovering that consumers purchase
Kellogg’s Pop Tarts, particularly strawberry Pop Tarts, before a hurricane. We
can only assume that Walmart can then work with Kellogg’s to stock shipments
at stores in preparation for a hurricane. This detailed level of customer tracking
gives Walmart deep insights into customer preferences and buying behavior.
Armed with this type of factual knowledge also enables Walmart to win pricing
and distribution concessions from its suppliers.
Walmart’s analytics and data availability extends to all its suppliers, cur-
rently numbering more than 17,400 suppliers in 80 countries. These suppliers are
required to use the company’s analytical platform that gives the suppliers a view
of in-store demand. This is Walmart’s “Retail Link” system that can be used to track
the movement of all products. Using the system, suppliers know in real-time when
stores need to be restocked rather than waiting for an order from Walmart. The sys-
tem allows suppliers to search for information on sales, shipments, purchase orders,
invoices, claims, and forecasts, and it allows them to run data queries. The system
also gives suppliers access to Walmart’s assortment planning, enabling suppliers
to create store specific modular layouts based on sales data and store traits.
Our case studies show that leading companies like Walmart understand that
optimizing one supply chain function is insufficient. Walmart uses big data analyt-
ics to link the entire chain from source to sell. Its applications on the sell side
capture and track demand through POS data. This information moves efficiently

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 33

 How to Use Big Data to Drive Your Supply Chain

through the supply chain to inform all the other supply chain functions, coordinat-
ing POS data with inventory information from RFID sensors. This is the role of the
Retail Link platform that serves to connect and coordinate the entire supply chain,
matching supply with demand. Information at any one point in the supply chain—
say, a shortage in supply, delayed shipments in transportation, or production stop-
page in operations—is conveyed to all other supply chain functions informing them
and coordinating action. By making supply and demand signals visible between
retail stores and suppliers, the company optimizes all its supply chain decisions—
from customer fulfillment to inventory tracking to automatic purchase orders
through its supplier portal.
Another example is Zara, the major division of the Spanish retailer Inditex.
Zara is a global leader in “fast fashion,” an industry characterized by the ability to
generate quick turnover of merchandise at affordable prices. Low prices coupled
with freshness of assortment are requirements. Success mandates rapid informa-
tion access and quick supply chain response,42 dependent upon a data-driven
coordinated supply chain. The first hallmark is Quick Response, a set of standards
for information exchange and supply chain management, enabling significant
reductions in lead times and efficiency due to rapid information exchange. This
includes real-time information on item sales, stocking levels, and inventory move-
ments that are then quickly used to modify production, distribution, and procure-
ment decisions. This information is also used to create dynamic assortment, which
are frequent assortment changes that can occur monthly, weekly, or even daily
and that appeal to high-fashion customers. The result is low inventory coupled
with high responsiveness. To further enhance these capabilities Zara recently
implemented an RFID system to track items from factory to point of sale, getting
a real-time picture of which fashions are selling, prompting an instant order to the
stockroom each time a garment is sold, and keeping track of all this data.43 Like
Walmart, the data and analytics are shared across functions and the supply chain
in real time. However, in Zara’s case, this coordination is made easier as Zara is
vertically integrated, owning its stores and managing its own assortments.
Yet another example is Seven-Eleven Japan, the largest convenience store net-
work in the world. With a professional customer base, the company experiences
demands similar to those seen in fast fashion, such as constantly changing customer
preferences and last minute customer orders for prepared items. The company has
a strategy with a focus on freshness, and its Just-in-Time supply chain relies on a
sophisticated information system that gathers point-of-sale data shared with suppliers
and distributors.44 The data is also used to forecast future trends and, like Walmart
and Zara, mined to optimize assortments. By analyzing hourly sales trends for individ-
ual items, the company optimizes delivery schedules and minimizes waste. Based on
data analysis it even changes individual product assortments, such as that of milk
products rearranged a few times a day to better suit customers’ evolving needs. One
assortment is small containers of milk in the morning for workers to pick up on
way to work. By midday, the assortment is changed with lunch-sized servings offered
for students. At the end of the day, larger cartons are placed for parents bringing milk
home in the evening. Every aspect of such assortments—timing, volume, display, and
deliveries—is optimized based on data and coordination with suppliers.45

34 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

These examples show that leading companies use state-of-the art analytics
applications to not only gather data and local intelligence, but to connect the supply
chain. These companies understand that the supply chain is a system and, as such,
functions need to be linked and coordinated for the system to work. Optimizing
costs along one function typically results in increased costs in another. Marketing,
for example, may use analytics applications to customize product offerings.
However, if operations are not able to produce the desired product versions and
quantities, or if logistics is not prepared to deliver them, the system will underperform.

Implementation Hurdles
Our survey of senior executives and managers finds that the majority
(seventy-eight percent) believe that big data analytics is a technological priority
for the future. Sixty-eight percent expect to be making some degree of supply
chain technological software investment in the coming year. However, eighty-three
percent expressed substantial concerns regarding costs of the technology and the
adoption decision for their individual needs, considering costs versus capabilities
needed. Sixty-eight percent stated they were awash in data and seventy-two percent
engaged in fragmented implementation efforts.
Based on our findings, we identify four hurdles that prevent companies
from taking advantage of the big data revolution.
§ Needle in a Haystack—Many companies are feeling the need to rush and
implement analytics applications to keep up with the hype. As a result,
they are often using analytics randomly in search for causation and rela-
tionships with the hope that something will eventually turn up. Rather
than being focused, analysts are in many cases engaging in mere number
crunching, often uncovering relationships that are false positives resulting
in wasted time and money. An executive at a financial services firm
described a scenario where the company’s analysts are “number crunch-
ing” through huge amounts of data in the hope of identifying ways to
improve forecasts. The problem, as he put it, is that “we are just running
random experiments to see if anything will show up.” As commented in
a recent Harvard Business Review article, “a pure data mining approach often
leads to an endless search for what the data really say.”46
§ Islands of Excellence—Often companies select applications designed for a
specific process. Although the algorithm may optimize that process and
be very efficient, there may be little bearing on optimizing the supply chain
if the process isn’t linked across the supply chain. An executive at a mid-
sized company described his company’s effort to optimize transportation
fuel costs without linking it to other variables such as lost sales and prioriti-
zation of deliveries. Another interviewed analyst proudly exclaimed: “I have
an algorithm to optimize the content of our kitchen cabinets!”
§ Measurement Minutiae—Many companies find that they have too many met-
rics. This is the hallmark of Measurement Minutiae, which is trying to mea-
sure everything. Few companies have the diligence to actively manage all of
the metrics they have created and most of the companies don’t know which

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 35

 How to Use Big Data to Drive Your Supply Chain

ones to focus on. An executive at a large shelf-stable food company told us:
“We have over one hundred different metrics. My primary job is to drastically
cut this number so we know what matters.” Many of the companies we inter-
viewed were settling on easier to measure metrics, such as margins, revenues,
and inventory turns, resulting in metric proliferation. By contrast, a few com-
panies had taken the time to develop a smaller number of customized metrics
that actually measured relevant performance.
§ Analysis Paralysis—Many companies are overwhelmed with the rapid change
of technological capability. They understand they must do something but are
in a state of paralysis. Many indicate they are drowning in data—gathered
from POS systems, websites, internal transaction processes, and social media.
For most, it is difficult to digest all of the data, technology, and analytics that
are available to them. Many indicate that they cannot absorb the new analyt-
ics technologies being made available to them. As a result, they find that they
are in a state of paralysis. In our interview, a CPG executive exclaimed: “We
have so much data but have no idea how to use it! What do we do with all this
data? Where do we begin?” This was a common sentiment frequently stated.

Coordination Imperative
Indeed, companies leading in big data analytics are “number-crunchers” with
superior technology. However, not all have the latest analytics applications. Our
study finds that these leading companies do much more than just acquire technol-
ogy. They have the right focus so that the acquired applications are not random but
directly support the value proposition of the business. To drive entire supply chains,
their efforts are coordinated across all supply chain levels.
Analytics efforts are resource intensive. As a result, companies must choose
where to direct these efforts. It is not possible to effectively focus on everything. Big
data implementation efforts that tackle too many areas can easily become diffused.
This can lead companies to lose sight of the business purpose behind the effort. We
find that successful implementation requires companies to focus their big data
efforts by strategically selecting areas or initiatives that support their overarching
strategy. A good example is UPS, which had initially focused their analytics efforts
on improving logistics operations.47 Since then the company has expanded analyt-
ics efforts to provide superior customer service. Further, internal analytics efforts
must be coordinated with customers and suppliers. Recall that Walmart requires
its suppliers to use its Retail Link system to monitor product movement by store,
to plan promotions and layouts within stores, and to reduce stock-outs. Procter &
Gamble has a similar platform called Joint Value Creation. It allows data analysis
for both its retail customers and suppliers and helps improve responsiveness and
reduce costs.48
Coca-Cola provides another example of a big data driven, end-to-end supply
chain. Analytics applications are used along all supply chain functions that “talk” to
one another to ensure coordination of activity. First, an algorithm is used to engineer
the taste of its orange juice. Next, a computer model is used to direct everything from
picking schedules of oranges to the blending of ingredients needed to maintain a

36 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

consistent taste. The company has spent $114 million to expand its U.S. juice bottling
plant, which is completely technology driven and claims to be the largest in the
world. At this plant, the company uses a computer algorithm it calls Black Box,
which can be viewed as a “secret recipe.” Black Box contains granular data from
more than 600 flavors that make up what customers perceive as an “orange taste.”
The algorithm then matches this data to the taste of each batch of raw juice. The
profile of each batch specifies acidity, sweetness, and other taste attributes. The algo-
rithm then blends batches of raw juice to replicate the taste and consistency per-
ceived as “orange,” including the amount of pulp added. Further, the algorithm is
tied to satellite images of fruit groves to ensure the fruit is picked at the optimal time
for Coca-Cola’s bottling plants. To generate the computations, the algorithm also
considers external factor such as crop yield, current prices, and weather patterns.
Every aspect of the supply chain is optimized and standardized, but can be quickly
updated. In fact, the mathematical model can quickly create new plans—in a matter
of five or ten minutes—with any new information. Jim Horrisberger, Director of Pro-
curement, put it this way: “You take Mother Nature and standardize it.”49

The Framework
Clearly big data analytics has a transformative potential for supply chain
management. However, based on our case studies we find that companies who
have actually achieved this have three things in common. First, their efforts are
driven to support the company’s strategy rather than random efforts of explora-
tion. Second, these companies use applications along all supply chain functions
in a coordinated manner rather than optimizing just one function. Third, they
measure performance through carefully selected metrics and use these for contin-
uous improvement, further guiding their analytics efforts. Based on this, we have
developed the framework shown in Exhibit 2.

Segment
The first step in the framework is to use analytics to create optimal supply
chain segments with clear attributes. This defines exactly how the company
intends to compete in each segment and helps establish the needed focus. It helps
companies avoid the hurdle of needle in a haystack or creating islands of excellence.
Indeed, companies have been segmenting customers for decades by attributes
such as demographics, purchase patterns, shopping attitudes, and behavior. These
segments have different product and customer service requirements, and they are
served through different channels and different supply chains.
Almost twenty years ago, a seminal article described the then technological
capability and asked: “Why haven’t the new ideas and technologies led to improved
performance?”50 It went on to explain that the nature of product demand requires
different supply chains and suggested dividing products into “functional” (e.g.,
stable product demands with long life cycles) and “innovative” (e.g., uncertain
product demand with short life cycles and great variety), each having different
supply chain structures. The underlying premise still holds true. However, today’s
analytics enables data scalability with granular data being aggregated in an infinite

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 37

 How to Use Big Data to Drive Your Supply Chain

EXHIBIT 2. Framework for Implementation

SEGMENT
Create Customer
Segments & Define
competitive priorities

CONTINUOUS
IMPROVEMENT
CYCLE
ALIGN
Align functional efforts to support segments

SOURCE MAKE MOVE SELL

MEASURE
Develop strategically aligned KPIs

array of possibilities, making it possible to optimize segment categories in new and

novel ways, well beyond a binary set.
A critical part of creating segments is defining competitive priorities in each
segment, which specify how a company intends to compete in each segment.
Defining unique segments and their characteristics leads to a clear identification
of competitive priorities for each segment. These should be specific and may
include target levels for customer service, cost competition, quality, time, or
responsiveness. Each of these competitive priorities translates into different oper-
ational requirements.51 This results in different supply chain structures, different
suppliers, transportation and operational strategies, and thresholds of perfor-
mance for each segment. Consider that supply chain segments that focus on cost
have very different supplier requirements than those that compete on innovation,
quality, or customer service.
Each segment focuses on different objectives. The goal is to develop seg-
ments that optimize customer needs and supply chain requirements to serve each
segment. Analytics algorithms can then be selected to optimize decision processes

38 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

to support the competitive priorities in each segment—such as optimizing customer

service while keeping a threshold on costs. An excellent example is the apparel
retailer American Eagle Outfitters. The company used analytics to create clusters
of its more than 750 stores based on the types of assortments to which shoppers
were most responsive.52 They were then able to identify that customers in Western
Florida bought merchandise similar to those in parts of Texas and California. This
segmentation enabled the company to better target store assortments and better
control pricing in each segment.53

Align
The next step in the framework is to align organizational functions so that
their efforts support segment attributes and competitive priorities. Aligning means
integrating processes across the supply chain. As a result, each supply chain func-
tion applies its analytics efforts to support the stated competitive priority, rather
than engaging in random exploratory efforts. Alignment integrates the organiza-
tion and its supply chain horizontally. This is accomplished through the sharing
of intelligence across functions and supply chain partners, and engaging in joint
decision making, such as Sales & Operations Planning (S&OP). This is precisely
what was illustrated in the examples of Walmart, Zara, and Seven-Eleven Japan.
Alignment also avoids fragmented efforts. Staying true to alignment can be
challenging given the endless array of new analytics applications and the hype
surrounding them. However, functional alignment should drive application selec-
tion rather than analysts engaging in efforts that do not support competitive pri-
orities. Without alignment, no amount of data mining will yield a system-wide
competitive advantage.
One part of alignment is to use analytics to sync up supply and demand.
Big data can be a huge source of aid in this process as it enables “demand sensing”
and helps drive other supply chain decisions. An example is Ford, which uses big
data analytics to align its supply chain. FordDirect is a platform that provides an
interface for information sharing between the customer, dealer, and manufacturer
in real-time. It can be used to customize vehicles, manage inventory, and handle
customer service needs such as obtaining financing.54 The platform enables sharing
of information permitting supply chain coordination.

Measure
The third step in the framework is to measure performance. We find that
leading companies develop strategically aligned KPIs to measure segment attrib-
utes. Here the analytics initiatives are measured with targeted and measurable
KPIs that are routinely reviewed. The key is to identify the right metrics for the
phenomena the company needs to optimize. This can be accomplished using stra-
tegically aligned KPIs agreed upon by all process members. These metrics should
also measure degree of alignment, integration, and cross-enterprise cooperation.
We find that a number of leading companies use analytics to look for new mean-
ingful metrics that are driven by strategy, core competencies, and measure the
value proposition of the business. Big data analytics enables and necessitates
the development of new metrics that offer greater insight. An example of this type

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 39

 How to Use Big Data to Drive Your Supply Chain

of innovation is seen the movie Moneyball where traditional baseball metrics to

evaluate players, such as “batting average,” were changed to new and more
meaningful metrics, such as “on-base percentage.”
Finally, a feedback loop should exist between metrics—monitored on a
continuous basis—and the defined segments and their competitive priorities.
The metrics should be used to refine the segmentation process and realign the
competitive priorities. Through kaizen—continuous improvement—we have
learned that the best and sustained change occurs from gradual improvements.55
This should be an ongoing process. Big data algorithms can significantly assist with
this, such as automatically tracking these metrics and creating alerts when devia-
tions occur.

Implementation in Action
Our framework is prescriptive. It provides the steps companies should
follow in implementation. Technology and big data analytics are merely tools
in achieving a competitive advantage enabling better segmentation (Step 1),
alignment that unifies functional areas (Step 2), and use of metrics that drive
performance (Step 3). Lastly, the framework is cyclical following principles of
continuous improvement. It should be routinely refined and realigned, such
as part of the S&OP cycle (Step 4). Selection of analytics applications should fol-
low and support decisions at each step.
Dell offers an example of this type of implementation. Only a few years
ago, Dell competed on the configure-to-order model.56 Here customers could
order customized computer configurations on the company’s website, including
variations in models, software configurations, memory, screens, design, and every
other customizable feature. This supported the company’s strategy of customiza-
tion. However, from an operations and supply chain standpoint it was a night-
mare. The combinations of customer options resulted in over seven septillion57
possible configurations of Dell products. Although the options were good for cus-
tomers, the inventory implications were staggering.
Then the company moved to an analytics driven system that allowed it to
maintain the same strategy but optimize inventory decisions. They still wanted to sat-
isfy the diverse needs of a broad set of customers. However, they also wanted to keep
costs as low as possible. To respond to the challenge the analytics team decided to use
historical order data to run cluster analysis to determine the most common configu-
rations customers were selecting. The finding was that there was actually a great deal
of commonality in ordering. Analyzing the data, they were able to reduce the seven
septillion options to just a few million. They were then able to segment their product
configurations. They identified the models so common that they could stock these in
preconfigured inventory. Coordinating with their supply chain partners, these com-
mon configurations could be built ahead of time with the lowest margins and stocked
in inventory to be shipped. For this segment, next day delivery was made possible,
providing high customer service with low inventory cost. This level of segmentation
and analyses enabled Dell to sell precisely what the customer wants but produce and
deliver it in a cost-effective manner. This new system has brought the company an
additional $40 million in revenue. It also exemplifies the benefits of using big data

40 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

analytics to segment markets and customers, and then use analytics for product con-
figurations and coordination with supply chain partners.

Implementing Technological Capability

Our study of leading companies reveals a process for implementing analyt-
ics capability. There is no one big data solution. Instead it is a set of analytical tools
available with varying degrees of complexity, cost, and capability. We have
learned that not every company needs the most powerful analytics capabilities
to be successful. Further, implementation of these capabilities is best accomplished
through a gradual evolving process of maturity as discussed below. As a number
of interviewed executives noted: “It is a journey.”

The Maturity Map

Most organizations follow an evolution in their implementation of analyt-
ics. They utilize efforts that build big data capabilities over time and learn through
the process of continuous improvement. There are four stages of maturity in this
evolution, which are presented as a maturity map in Exhibit 3. The first stage is
digitizing and structuring the data, what we call “data structuring.” This is a pro-
cess that ensures that the data generated are clean, structured, and organized in
such a way that they can be used for further analysis. This involves a process
called “scrubbing” the data to remove errors and ensure data quality. Our inter-
views with executives reveal a repeated concern with “dirty data.” Many quoted
the adage of “garbage in and garbage out.” Companies are awash in data but
much of it may not be reliable.
The second stage of evolution requires making the data available to all
who need it when they need it. Available data can be a powerful driver of value
by itself and is an important first step in integrating datasets to create more
meaningful business insights. Further, data needs to be available in usable form
to support decision making in the respective areas and functions. Making the
right data, in the right form, available when and where it is needed is a signifi-
cant process challenge.

EXHIBIT 3. Implementation Maturity Map

STAGE 1 STAGE 2 STAGE 3 STAGE 4

Data Data Basic Advanced
Structuring Availability Analytics Analytics

“Clean” data. Make the right data Basic Analytics Tools Predictive Analytics
Ensure data quality. available when and Descriptive Statistics Automated
where needed.
Correlation Analysis Algorithms

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 41

 How to Use Big Data to Drive Your Supply Chain

The third stage is applying basic analytics to the data. This covers a range of
methodologies, such as basic data comparisons, correlations, and regression. Here,
relatively standardized quantitative analyses are used, such as descriptive analytics.
These do not require customized analyses designed by people with deep analytical
skills as in the subsequent stage. In fact, many companies do not need the highest
level of analytics to achieve huge gains. Many can gain success from this stage with-
out investing in more advanced technologies. Correlations and simple regression
alone can provide powerful insights. In fact, we found this to be particularly true
for a few small and mid-sized companies which, given their scale, were able to
more easily implement coordinated solutions that utilized basic analytics.
The fourth and highest level is applying advanced analytics, such as predic-
tive analytics, automated algorithms, and real-time data analysis that can create
radical new business insight. They allow new levels of experimentation to develop
optimal approaches to targeting customers and operations, and to opening new
big data opportunities with third parties. Leveraging big data at this level often
requires the expertise of deep analytical talent.
As companies gain competency, they move along the maturity map. It is a
mistake for companies to try to jump to the last stage of the maturity map with-
out first going through the first three stages. Companies need to go through the
organizational adaptation, learning, and continuous improvement needed to
move through the stages. We find that when leading companies implement ana-
lytics, they move through the maturity map. They tend to begin on a small, tar-
geted, and carefully selected pilot project. This is called a “process of purposeful
experimentation.” It can be the best path toward becoming an organization that
fully leverages big data. It also enables learning. This is a very different and more
effective approach than a complete plan for the enterprise prior to doing any
implementation. It can begin by selecting a few high-potential areas in which
to experiment with big data and then scaling upward. It is easier to create value
from such small projects rather than jumping directly to complex analytics
implementation. We found a few examples of the latter resulting in islands of
excellence.
Kaiser Permanente offers an example of staged implementation.58 The com-
pany began their big data efforts by concentrating on one IT project focused exclu-
sively on patients with long-term conditions. Then they moved along the maturity
map creating specific disease registries and panel management solutions, rather than
an all-encompassing IT solution addressing a range of problems. This type of slow
and targeted approach—particularly as it supports competitive priorities—is the best
strategy.
Lastly, as companies move through the maturity map they need to concur-
rently build their IT infrastructure with security in mind. Many companies address
security post hoc by layering new security on top of the existing architecture,
which often results in greater security gaps.59 Even the earliest stage of the matu-
rity map should include security considerations. Although integrating analytics
across the supply chain provides tremendous benefits, it unfortunately also creates
a host of cybersecurity challenges and vulnerabilities against which companies
need to guard.

42 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

Analytics Outsourcing
Successful use of big data analytics does not only depend on availability of data
and analytical tools. It also depends on the ability of organizational members—leaders,
managers, and employees—to be able to use them. This is a significant problem. Most
companies lack the capabilities to do all the analytical work that is required. As tech-
nological capability advances, the gap between available technology and the ability to
use it will continue to grow. Many companies need specialized software available
from external vendors. They may also need access to additional databases. Further,
they may need someone to orchestrate the use of all the technological elements. In
our interviews we found outsourcing analytics and hiring an outside vendor to be a
major concern for companies.
To quickly gain access to technological capabilities needed, many compa-
nies are turning to analytical outsourcing. In fact, analytical outsourcing is a rap-
idly growing trend. Leading companies outsource various aspects of their
analytics capability, from Walmart outsourcing to Mu Sigma to Limited Brands
and Pottery Barn working with Alliance Data.60 The need for these services is
obvious given the high level of skills required. As a result many providers, from
software vendors to consultants, are now providing analytical assistance. Exter-
nal providers can offer access to data and software. They are on the technology
frontier and can provide a breadth of analytical assistance. Further, many exter-
nal providers specialize in certain industry segments, say healthcare or fast-
fashion, and can quickly bridge the gap between the company’s needs and tech-
nical knowledge.
Two key dimensions of outsourcing are the scope and criticality of the
engagement.61 Scope is the degree of responsibility outsourced, while criticality
is the importance of the outsourced activities. The greater the scope of the out-
sourced task, the larger the relinquishing of control by the organization. Similarly,
the greater the criticality of the outsourced task, the greater the consequences of
poor performance. For example, the outsourcing engagement can be small in
scope, involving just one aspect of analytics—such as purchasing data or conduct-
ing analysis on one existing data set. It may mean supplementing current staff
with either onshore or offshore analytical consultants. On the other hand, it
may mean outsourcing most of the company’s analytics capabilities.
The potential benefits of outsourcing are large as they enable an organization
to tap into highly specialized skills it currently does not have. However, outsourcing
can create a range of risks and dependencies, not the least of which is the leaking of
proprietary information and issues of data security.
Companies should be especially careful when outsourcing activities with
large scope and criticality, where these risks can be particularly dire. Consider
an example where some 108,000 Florida state employees had their personal
information compromised when the outsourcing service provider improperly
allowed subcontractors in India to index state personnel files.62 Another example
is that of Target where a data breach was tied to employees of a company that
provided call center support service to Target National Bank, the issuer of the
Target Visa Card.63 These outsourcing engagements need to be monitored carefully.

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 43

 How to Use Big Data to Drive Your Supply Chain

Also, specifying liability for data breaches and adding specific data security
requirements into contracts are becoming increasingly common.64
Our interviews revealed a wide variation in engagements. As a result,
we suggest companies begin by first understanding their current capabilities,
processes, and needs. Otherwise they may find that they have purchased capabili-
ties that do not enhance the business value, are beyond their scope of understand-
ing, or worse yet solidify bad processes (as we observed in a few cases). Second,
companies should compare providers, beginning with those who offer analytics
applications for their specific industry versus those who work across industries.
Some companies, for example, preferred providers who had expertise in their spe-
cific niche, such as location applications in healthcare, while others specifically
wanted applications successful in other industries. The idea here is that it may
be possible to gain a competitive advantage by partnering with those outside of
one’s industry. IBM showcases the advantage of this by recently launching analyt-
ics solutions based on data collected from over 50,000 different client engage-
ments across a wide range of industries, from oil and gas to healthcare to retail
to banking. These analytics solutions cut across twelve different industries and
are customized per industry segment yet standardized to serve as “out of the
box” solutions with preset dashboards. The company could only achieve this capa-
bility given its large scale and industry breadth of best practices learned.65 Third,
selection of external providers should include their ability to seamlessly interface
with systems currently in place. A number of companies we spoke with used
multiple niche providers with the caveat that the systems were all able to com-
municate. Lastly, the ability to scale and go beyond the company’s immediate
need is a high priority. Some partners may build short-term technical capabili-
ties, while others can help to build long-term organizational capabilities that
are highly scalable. In fact, a few companies we interviewed ultimately chose
to develop their own internal analytics capability after a careful analysis of
options.
The need for outsourcing will grow as IT capabilities evolve requiring ever
more specialized skills. As a result, companies need to develop a skillfully formu-
lated analytics outsourcing strategy. This strategy must clearly delineate which
analytical capabilities the company wants to build for itself, and which they will
outsource from partners. It must also specify the long-term plan for building capa-
bilities over time, in addition to meeting short-term goals.

Conclusion
Big data analytics is not just another technology. It is the nexus of software,
computing, and technological capabilities that has ushered in an era of radically
different competition and is a “tech” disruption of historic proportions. This study
was motivated by the need to understand how big data drives leading supply chains
and to identify the characteristics of successful implementation. Despite the hype,
the majority of companies have yet to leverage big data for their supply chain oper-
ations, they are engaging in random implementation efforts, and many do not
know how to proceed. Contributing to this are analytics capabilities that offer deep

44 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

insights but are hyper-specialized. This creates special challenges for supply chain
management where executives need to improve performance of entire organiza-
tional systems rather than mere functions.
We identified three distinct differences of today’s big data analytics capability:
unprecedented opportunity for inquiry, change in the nature of inquiry, and change
in the nature of experimentation. This has led to an array of application capabilities
that is vast, tempting, and even overwhelming. Thus it is more important than ever
to look to practices of firms succeeding in this arena and identify their common strat-
egies. We found a few key lessons from these companies. First, these companies use
big data applications that are coordinated across supply chain functions matching
supply with demand. They do not select applications to optimize single functions
or single decision areas in isolation. Second, their efforts are focused and driven
by strategy, resulting in targeted tactical applications rather than random efforts
of exploration. Third, they rigorously measure performance using carefully devel-
oped metrics and engage in continuous improvement. Fourth, leading companies
follow “a journey” in their implementation efforts, beginning with a carefully
selected pilot project, rather than jumping into wide-scale adoption. These are
important lessons that should guide the vast majority of companies that remain
uncertain as how to proceed in this new analytics era full of opportunity.

Notes
1. Michael E. Porter and James E. Heppelmann, “How Smart Connected Products are Transforming
Competition,” Harvard Business Review, 92/11 (November 2014): 64-88; D. Barton and D. Court,
“Making Advanced Analytics Work for You,” Harvard Business Review, 90/10 (October, 2012):
79-83; H. Chen, R. Chiang, and V. Storey, “Business Intelligence and Analytics: From Big Data
to Big Impact,” MIS Quarterly, 36/4 (December 2012): 1165-1188; S. Baker, The Numerati
(New York, NY: Houghton Mifflin, 2008).
2. T. Davenport and D. Patil, “Data Scientist: The Sexiest Job of the 21st Century,” Harvard Business
Review, 90/10 (October 2012): 70-76.
3. James Manyika, Michael Chui, Brad Brown, Jacques Bughin, Richard Dobbs, Charles Roxburgh,
and Angela Hung Byers, “Big Data: The Next Frontier for Innovation, Competition, and Produc-
tivity,” McKinsey Global Institute White Paper, May 2011.
4. Thomas H. Davenport and Jeanne G. Harris, Competing on Analytics: The New Science of Winning
(Boston, MA: Harvard Business School Publishing, 2007).
5. “A Different Game,” The Economist, February 27, 2010, pp. 6-8.
6. Thomas H. Davenport, “Realizing the Potential of Retail Analytics Plenty of Food for Those
with the Appetite,” Working Knowledge Research Report, Babson Executive Education, 2009,
pp. 1-42.
7. Thomas H. Davenport, “Competing on Analytics,” Harvard Business Review, 84/1 (January
2006): 98-107.
8. Kashmir Hill, “How Target Figured Out a Teen Girl Was Pregnant Before Her Father Did,” Forbes,
February 12, 2012.
9. Bill Schmarzo, “KPMG Survey: Firms Struggle with Big Data,” InFocus, February 6, 2014,
<https://infocus.emc.com/william_schmarzo/kpmg-survey-firms-struggle-with-big-data/>.
10. “The Promise and Challenge of Big Data,” A Harvard Business Review Insight Center Report,
September 2012.
11. P.B. Goes, “Big Data and IS Research,” MIS Quarterly, 38/3 (September 2014): iii-viii.
12. Fifty-two open ended interviews were conducted from industries including pharma (12%),
manufacturing (53%), healthcare (11%), retail (17%), hospitality (5%), and financial services
(2%). Interviews were conducted in person following an interview protocol and conducted at
the following annual professional conferences Council of Supply Chain Management Annual
Conference (CSCMP), Supply Chain Leader’s in Action Conference (SCLA), Institute for Supply

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 45

 How to Use Big Data to Drive Your Supply Chain

Management (ISM), and the INFORMS Analytics & Operations Research Conference (2014).
Interviewees included eighteen senior executives and SVPs, twenty-one mid-level managers
or directors, and thirteen senior analysts.
13. Survey of 300 companies with 102 respondents (response rate 34%) using Dillman’s Survey
Design Method (Dillman et al., 2009); respondents were high-level managers and executives from
manufacturing (53%), retail (31%), and wholesale/distribution (16%); t-tests for differences
in means between the initial and final quartiles of respondents was used for non-response bias
(Armstrong and Overton, l977). D.A. Dillman, J.D. Smyth, and L.M. Christian, Internet, Mail, and
Mixed-Mode Surveys: The Tailored Design Method (Hoboken, NJ: Wiley & Sons, 2009); J.S. Armstrong
and T.S. Overton, “Estimating Nonresponse Bias in Mail Surveys,” Journal of Marketing Research,
14 (1977): 396-402.
14. Andrew McAfee and Erik Brynjolfsson, “Big Data: The Management Revolution,” Harvard
Business Review, 90/10 (October 2012): 60-68.
15. Scholarly research on big data analytics is becoming extensive. Our major journals, such as
Management Science, MIS Quarterly, Production Operations Management, International Journal of
Forecasting and others, have commissioned special issues on these topics. A new journal called
Big Data has been recently launched with thousands of recently downloaded articles.
16. X. Zhang and L. Zhang, “How Does the Internet Affect the Financial Market? An Equilibrium
Model of Internet-Facilitated Feedback Trading,” MIS Quarterly, 39/1 (2015): 17-38.
17. “Increasing User Engagement with Mobile Analytics,” presented at INFORMS Conference on
Analytics & Operations Research, April 12-15, 2015, Huntington Beach, CA.
18. V. Dhar, “Data Science and Prediction,” Communications of the ACM, 56/12 (2013): 64-73.
19. Michael P.S. Brown, William Noble Grundy, David Lin, Nello Cristianini, Charles Walsh Sugnet,
Terrence S. Furey, Manuel Ares, Jr., and David Haussler, “Knowledge-Based Analysis of Micro-
array Gene Expression Data by Using Support Vector Machines,” Proceedings of the National Acad-
emy of Sciences of the United States of America, 97/1 (2000): 262-267.
20. One part of data science is focusing on “Deep Learning” using algorithms to extract high-level,
complex abstractions as data representations through a hierarchical learning process. Progressively
complex abstractions in the hierarchy are “learned” at a given level based on simpler abstractions
from the preceding level. Deep Learning enables analysis and learning from big data that is
unstructured. For a good representation see Maryam M. Najafabadi, Flavio Villanustre, Taghi M.
Khoshgoftaar, Naeem Seliya, Randall Wald, and Edin Muharemagic, “Deep Learning Applica-
tions and Challenges in Big Data Analytics,” Journal of Big Data, 2/1 (December 2015): 1-21.
21. “How LinkedIn Leverages Text Analytics to Hear Customer Voices in Today’s Social Media,”
INFORMS Conference on Business Analytics & Operations Research, April 12-14, 2015,
Huntington Beach, CA.
22. A.D.I. Kramaer, J.E. Buillory, and J.T. Hancock, “Experimental Evidence of Massive-Scale
Emotional Contagion Through Social Networks,” Proceedings of the National Academy of Sciences
of the United States of America, 111/34 (2014): 8788-8790.
23. McAfee and Brynjolfsson, op. cit.
24. Chen, Chiang, and Storey, op. cit.; also discussed in Goes, op. cit.
25. “Gartner Predicts Business Intelligence and Analytics Will Remain Top Focus for CIOs Through
2017,” December 16, 2013, <www.gartner.com/newsroom/id/2637615>.
26. William Ruh, Vice President of GE Software, “Industrial Internet,” keynote presentation at the
INFORMS Conference on Analytics & Operations Research, April 12-14, 2015, Huntington
Beach, CA.
27. Davenport (2009), op. cit.
28. David Kiron, Renee Boucher Ferguson, and Pamela Kirk Prentice, “From Value to Vision:
Reimagining the Possible with Data Analytics,” MIT Sloan Management Review Research Report,
March 5, 2013.
29. Davenport (2009), op. cit.
30. Davenport and Harris (2007), op. cit.
31. Victor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution That Will Transform How
We Live, Work, and Think (Boston, MA: Houghton Mifflin Harcourt, 2013).
32. Mary Siegried, “Find the Big Picture in Big Data,” Inside Supply Management, (January/February
2014), pp. 19-23. Also available at <https://www.instituteforsupplymanagement.org/pubs/
ISMMag/ismarticle.cfm?ItemNumber=24235 >.
33. Mayer-Schönberger and Cukier, op. cit.
34. Davenport and Harris (2007), op. cit.; Cindy Etsell, “Analyze This: Enhancing Operational
Excellence through Customer Data,” Digital Manufacturing, February 15, 2012.

46 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU

 How to Use Big Data to Drive Your Supply Chain

35. M.A. Huselid and B.E. Becker, “Bridging Micro and Macro Domains: Workforce Differentiation and
Strategic Human Resource Management,” Journal of Management, 37/2 (March 2011): 421-428.
36. M.R. Leenders, P.F. Johnson, A.E. Flynn, and H.E. Fearon, Purchasing and Supply Management
(New York, NY: McGraw-Hill, Irwin, 2010), p. 6.
37. Brad Stone, “The Secrets of Bezos: How Amazon Became the Everything Store,” Bloomberg
Businessweek, October 10, 2013, <www.businessweek.com/articles/2013-10-10-/jeff-bezos-
and-the-age-amazon-excerpts-from-the-everything-store-by-brad-stone>.
38. Shel Israel, “How Walmart and Heineken Will Use Shopperception to Put Your In-Store Expe-
rience in Context,” <www.forbes.com>, January 27, 2013.
39. Nigel Hollis, “Starbucks’ New Logo: A Risky Move,” HBR Blog Network, January 7, 2011,
available at <www.blogs.hbr.org/2011/01/starbucs-new-logo-apple-or-le/>.
40. “Data, Data Everywhere,” The Economist, February 25, 2010.
41. A great deal has been written about Walmart’s unifying platform Retail Link. Some good sources
are: Mayer-Schönberger and Cukier, op. cit.; “A Different Game,” The Economist, February 25,
2010, available at <http://www.economist.com/node/15557465>; Sam Hornblower, “Always
Low Prices,” <www.pbs.org/wgbh/pages/frontline/shows/walmart/secrets/pricing.html>.
42. Marshall L. Fisher and Ananth Raman, The New Science of Retailing: How Analytics are Transform-
ing Supply Chains and Improving Performance (Boston, MA: Harvard Business Press, 2010);
Marshall L. Fisher, Ananth Raman, and Anna McClelland, “Rocket Science Retailing Is Almost
Here: Are You Ready?” Harvard Business Review, 78/4 (July/August 2000): 115-124; F. Caro
and V. Martínez-de-Albéniz, “How Fast Fashion Works: Can It Work for You, Too?” IESE
Insight, 21 (Second Quarter 2014): 8-65.
43. Christopher Bjork, “Zara Builds its Business around RFID,” Wall Street Journal, September 16,
2014, pp. B1-B2.
44. Caro and Martínez-de-Albéniz, op. cit.; Sunil Chopra “Seven-Eleven Co., Japan,” Kellogg
School of Management, February 2005.
45. Caro and Martínez-de-Albéniz, op. cit.; Sunil Chopra, op. cit.
46. Barton and Court, op. cit.
47. Mayer-Schönberger and Cukier, op. cit.
48. J. Maniyaka et al., “Big Data: The Next Frontier for Innovation, Competition, and Productivity,”
McKinsey Global Institute, May 2011.
49. “Coke Has a Secret Formula for Orange Juice, Too,” Bloomberg Businessweek, February 4, 2013,
pp. 19-21.
50. Marshall Fisher, “What is the Right Supply Chain for Your Product?” Harvard Business Review,
75/2 (March/April 1997): 105-116.
51. M.E. Porter, “From Competitive Advantage to Corporate Strategy,” Harvard Business Review,
65/3 (May/June 1987): 43-59; M.E. Porter, “What Is Strategy?” Harvard Business Review,
74/6 (November/December 1996): 61-78; M.E. Porter, On Competition (Boston, MA: Harvard
Business School Publishing, 1998); R.H. Hayes and S.C. Wheelwright, “Link Manufacturing
Process and Product Life Cycles,” Harvard Business Review, 57/1 (January/February 1979):
133-140.
52. Davenport (2009), op. cit.
53. <https://www.adobe.com/content/dam/Adobe/en/customer-success/pdfs/americaneagle-case-
study.pdf>.
54. “FordDirect Debuts DealerConnection Elite Service for Dealers at the National Automobile
Dealers Association (NADA) Convention,” PR Newswire, February 4, 2012.
55. Masaaki Imai, Gemba Kaizen (New York, NY: McGraw-Hill, 2007).
56. Travis Hessman, “Putting Big Data To Work,” Industry Week, 262/4 (April 2013): 14-18.
57. This is a number with 1024. Essentially it is a huge amount of data.
58. Manyika et al., op. cit.; Basel Kayyali, David Knott, and Steve Van Kuiken, “The Big-Data Rev-
olution in US Health Care: Accelerating Value and Innovation,” McKinsey & Company, April
2013.
59. Oliver Bossert, Wolf Richter, and Allen Weinberg, “Protecting the Enterprise with Cybersecure
IT Architecture,” McKinsey & Company, March 2015.
60. Manyika et al., op. cit.; Kayyali, Knott, and Van Kuiken, op. cit.
61. N.R. Sanders and A. Locke, “Making Sense of Outsourcing,” Supply Chain Management Review,
9/2 (2005): 38-45.
62. Robert McMillan, “Offshore Outsourcing Cited in Florida Data Leak,” Computerworld, March 26,
2006; Saritha Rai, “Security Breaches Worry Outsourcing Industry,” The New York Times,
October 5, 2006; “The Leaky Corporation,” The Economist, February 24, 2011.

CALIFORNIA MANAGEMENT REVIEW VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU 47

 How to Use Big Data to Drive Your Supply Chain

63. Paula Rosenblum, “Target Hit by One of Most Sophisticated Data Thefts Ever, But It Won’t
Hurt the Retailer,” Forbes, December 19, 2013; N. Eric Weiss and Rena S. Miller, “The Target
and Other Financial Data Breaches: Frequently Asked Questions,” Congressional Research
Service, February 4, 2015, <www.crs.gov>.
64. Stephanie Overby, “IT Service Providers and Customers Battle Over Data Breaches,” CIO Mag-
azine, March 9, 2012; “Outsourcing Risk: Who Pays Most When Customer Data is Breached?”
W.P. Carey Magazine, Spring 2015.
65. Thor Olavsrud, “IBM Unveils Industry-Specific Predictive Analytics Services,” CIO Magazine,
May 28, 2015.

California Management Review, Vol. 58, No. 3, pp. 26–48. ISSN 0008-1256, eISSN 2162-8564. © 2016 by
The Regents of the University of California. All rights reserved. Request permission to photocopy or
reproduce article content at the University of California Press’s Reprints and Permissions web page,
http://www.ucpress.edu/journals.php?p=reprints. DOI: 10.1525/cmr.2016.58.3.26.

48 UNIVERSITY OF CALIFORNIA, BERKELEY VOL. 58, NO. 3 SPRING 2016 CMR.BERKELEY.EDU