The Digital Transformation Playbook: Rethink Your Business for the Digital Age

Part of data strategy is developing a legal, risk management, and secu-
rity plan. Rather than letting fear of risk postpone action (and likely not 
really reduce risks), leaders need to establish assessment, responsibility, and 
planning, with appropriate outside partners to support them. The risks of 
data theft are unavoidable, but they can be reduced if risk reduction is a 
leadership priority.
Consumer attitudes are also crucial to data strategy. Beyond the threats 
of identity theft and cybercriminals, many consumers are more generally 
concerned about privacy and the increasing amount of information busi-
nesses gather about them. Much of the data about customers is collected 
in ways that the public is only vaguely aware of, at best. Advocacy around 
consumer data privacy has raised the possibility of government regulation 
in many markets. Start-ups like Datacoup, Handshake, and Meeco have 
argued that individuals should own their personal data and be paid for 
access to it. They hope to create tools that allow customers to store their 
interests, preferences, social data, and credit card transactions and choose 
how much of this information to sell to companies for a fixed price.
With rising concerns about ownership of personal data, it is increas-
ingly important that any data strategy be based on a transparent value 
exchange with the customer: an exchange in which the customer knows 
that data is being collected and sees the benefits they are receiving in 
return. This is the foundation of loyalty programs with points and rewards. 
It is also the reason customers willingly provide personal ratings on a 
service like Netflix and are not alarmed when Amazon suggests products 
based on their recent browsing history. When customers can easily see 
both the ways that companies are gathering data and the benefits they 
are gaining as a result, they will be more likely to allow sustainable access 
to businesses.
8
As sensors, networks, and computing become embedded in every part of 
our lives, the data that is available to business continues to grow exponen-
tially. For some managers, this data deluge will seem overwhelming. Other 
managers may tell themselves that “I don’t operate in a very data-intensive 
industry” simply because that was the case a few short years ago. But the 
world has changed. Every business now has access to data.
The strategic challenge for business is to develop the clear vision 
and the growing capability needed to put data to work in the service of 

---

122
T U R N D A T A I N T O A S S E T S
innovation and value creation. By treating data as a key intangible asset to 
build over time, every business can develop a data strategy that informs 
critical decision making and generates new value for business and cus-
tomer alike.
Data allows us to continually experiment, learn, and test our ideas. 
This means data can do more than power products, optimize processes, 
and deliver more-relevant customer interactions; it can also help change 
the way organizations learn and innovate. This different kind of learning—
through constant experimentation—is at the heart of a profoundly different 
approach to innovation. That new approach to innovation is the subject of 
the next chapter.

---

5
Innovate by Rapid Experimentation
Think of the last time you used a search engine. Every time you type a 
query into Google or a similar service, you are the subject of a human 
experiment. Google presents you with search results and measures which 
ones you click on, in what order, and how quickly. And in subtle ways, 
those search results that you see are constantly changing. Changes occur 
in the primary listings, in the search ads you are shown, and in the auto-
complete guesses that start to appear after you type your first letter. Google 
is constantly trying to learn more about how to innovate and improve its 
search service for users. Which links are you most likely to be looking for? 
How should it group them? (Local services vs. global ones? Recent news 
stories vs. company webpages? Links to subsections of a website? Biograph-
ical tidbits about the politician whose name you just entered?) To improve 
its products, Google doesn’t sit down with customer focus groups to discuss 
their search engine experiences. Nor does it convene a committee to vote 
on which new features to implement. Instead, the company is constantly 
experimenting, testing each of its new ideas, measuring customer response, 
and iterating on what it learns.
INNOVATION

---

124
I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N
We can define innovation as any change to a business product, ser-
vice, or process that adds value. This change can range from an incremental 
improvement to the creation of something totally new and unprecedented. 
For Google, an innovation may be launching a completely new product 
such as Gmail, Android phones, Google Maps, or its Chromebook lap-
top line. But innovation at Google also includes the continuous process of 
refining, adding and subtracting features, and evolving the user interface 
and experience. As Scott Anthony says, innovation is not just about “big 
bangs”; it is about anything new that has impact.
1
The fourth domain of digital transformation is innovation—the pro-
cess by which new ideas are developed, tested, and brought to the mar-
ket by businesses. Traditionally, innovation was singularly focused on the 
finished product. Testing ideas was relatively difficult and expensive, so 
decisions and early ideas were based on the analysis, intuition, and senior-
ity of managers involved in the project. Actual market feedback tended to 
come very late in the process (sometimes after public release), so avoiding 
a marked failure was an overriding concern.
In the digital age, enterprises need to innovate in a radically different 
fashion, based on rapid experimentation and continuous learning. Rather 
than concentrating primarily on a finished product, this approach focuses 
on identifying the right problem and then developing, testing, and learn-
ing from multiple possible solutions. Like the lean start-ups of Silicon Val-
ley, this approach focuses on developing minimum viable prototypes and 
iterating them repeatedly—before, during, and even after launch. At every 
stage, assumptions are tested and decisions are made based on validation 
by customer and market responses. Leaders are those who know how to 
pose the right questions, not claim the right answers. As digital technolo-
gies make it easier and faster than ever to test ideas, this new approach to 
innovation is essential to bringing new ideas to market faster and with less 
cost, less risk, and greater organizational learning. (See table 5.1.)
This chapter explores how rapid experimentation is transforming the 
way innovation happens and how digital technologies are making experi-
mentation both more possible and more necessary. We will consider two 
complementary methods of experimentation for innovators. We will also 
examine how organizations must change to become effective experimenters 
and what the real financial benefits are of learning to take an experimental 
approach to innovation. The chapter presents two strategic planning tools, 
each one offering a method for designing, running, and capturing value 
from innovation experiments. It also explores the four paths to scaling 

---

I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N

125
up an innovation and offers guidance on choosing the appropriate one. 
By applying these frameworks and tools, businesses can learn faster, fail 
cheaper and smarter, and shorten the time to successful innovation.
But, first, let’s look at a case study of a company using experimentation 
to rethink how it innovates for customers.
How to Grow the Innovation Premium: Intuit’s Story
Since its founding in 1983, Intuit has focused on designing and selling great 
accounting and finance tools for individuals and small businesses. With a 
track record of innovative products, the company grew from a start-up to 
a company worth billions. But after twenty-four years, founder Scott Cook 
realized the firm needed to change its model of product innovation if it was 
going to continue to grow. He started a new initiative with Kaaren Hanson 
that focused on rapid experimentation. When I met Hanson in 2013, she 
was chief innovation officer, and Intuit had run over 1,300 experiments in 
the previous six months. To provide a sense of how this new model for 
innovation worked, she described a project in India.
2
Deepa Bachu was the head of Intuit’s emerging markets team. The team 
had been tasked with developing a product for India’s farmers, who make 
up the bulk of the economy. After spending time immersed with small 
farmers to discover their pain points and customer needs, the team found 
a pressing problem for those who were selling perishable goods, such as 
Table 5.1
Innovation: Changes in Strategic Assumptions from the Analog to the Digital Age
From
To
Decisions made based on intuition and 
seniority
Decisions made based on testing and 
validating
Testing ideas is expensive, slow, and difficult
Testing ideas is cheap, fast, and easy
Experiments conducted infrequently,
by experts
Experiments conducted constantly,
by everyone
Challenge of innovation is to find the right 
solution
Challenge of innovation is to solve the right 
problem
Failure is avoided at all cost
Failures are learned from, early and cheaply
Focus is on the “finished” product
Focus is on minimum viable prototypes and 
iteration after launch

---

126
I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N
produce. These farmers, they learned, could afford to travel to only one 
market (or mandi) when it was time to find a merchant to buy their crop. 
When they did, they negotiated prices with a mandi agent, but there was 
a complete lack of market transparency. The mandi agents would actually 
put a cloth over their hand when indicating to one farmer the price they 
would pay for goods so that the next farmer in line could not see the price. 
Without access to refrigeration, the farmers had limited time to sell their 
perishables and no way to find the best buyer based on local supply and 
demand. In many cases, the farmers were forced to unload their produce 
for deeply discounted prices just to bring some income home. Bachu’s team 
set a goal: develop a product that could help farmers raise their income 
from crop sales by 10 percent. Then they set to work generating ideas.
3
The team’s first solution was to create an eBay-like marketplace where 
buyers and sellers could find each other and negotiate prices before sell-
ers loaded their produce and traveled to market. But when they presented 
mock-ups of the product to mandi agents, they discovered the agents would 
be unwilling to offer a price for produce without inspecting it first in person. 
The team’s second solution was to create a service that would let farmers 
alert each other to what crops they were growing so that each farmer could 
make a better guess as to what crops would be in higher demand. But when 
the Intuit team tested this idea, they found that farmers were unclear how 
to act on the information. The team’s third solution was to provide an SMS 
notification service that would inform farmers of the prices being offered 
at various markets before they left their farms. Bachu realized there were 
several assumptions behind this product idea: Could the farmers read the 
text messages? Would the mandi agents provide prices to Intuit to share? 
Would they honor those prices when the farmers arrived at the market? 
The team decided to run an experiment and recruited fifty farmers and five 
mandi agents willing to try out the notification service. For six weeks, two 
Intuit team members went into the markets to gather pricing information, 
while a third team member sat in a back office texting each farmer the 
prices of produce in various locations. This bare-bones operation would 
never scale, but it allowed the team to find out if the premise of an eventual 
mobile technology solution would actually work. At the end of the test, 
they found that both farmers and mandi agents had adopted it and that the 
farmers’ incomes were raised by 20 percent—twice the original goal. That 
impact continued as the final product, now called Fasal, was developed and 
rolled out as an automated service providing customized text messages to 
the more than 1 million participating farmers.
4

---

I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N

127
The experiment-driven approach to innovation was not isolated to 
emerging markets but became the hallmark of Intuit’s company-wide 
efforts to rethink innovation. “We have gone from a company of 8,000 
employees to 8,000 innovators,” Hanson told me.
5
Over the five years the 
company had been using this new approach, its innovation premium—the 
portion of its market capitalization attributable to future innovation—grew 
from 20 to 29 percent, adding $1.8 billion in value.
6
In shifting to a culture 
of rapid experimentation, the company had made a bet on running a large 
enterprise as a lab for continuous learning. That bet paid off big.
Experimentation Is Learning
Experimentation can be defined as an iterative process of learning what 
does and does not work. The goal of a business experiment is actually not 
a product or solution; it is learning—the kind of learning about customers, 
markets, and possible options that will lead you to the right solution.
When you innovate through experimentation, you don’t try to avoid 
wrong ideas; rather, you aim to quickly and cheaply test as many promising 
ideas as possible in order to learn which ones will work. This is very dif-
ferent than a traditional innovation process: analyze the market, generate 
ideas, debate internally, pick a solution, and then develop it through many 
stages of quality testing before launching it and getting feedback from actual 
customers. In developing Fasal for the Indian market, the Intuit team didn’t 
convene meetings to debate which of their three proposed solutions was 
the optimal one. To test their assumptions, they put their ideas, in rough 
form, in front of the actual farmers and merchants who would have to use 
the final product. This approach requires a paradigm shift from innova-
tion based on analysis and expertise to innovation based on ideation and 
experimentation for constant learning.
This shift toward a more iterative, learning-based model for innovation 
has been growing for several years and in many quarters. It is at the heart 
of Steve Blank’s customer validation model and Eric Ries’s writing on “lean 
start-up” methods. It is integral to the model of design thinking that prod-
uct development firms like IDEO and frog have been using with clients 
like Apple, JetBlue, Target, Disney, Intel, and SAP. With the rise of digital 
A/B testing, constant experimentation has become the norm for more and 
more products, services, and communication channels. It has become fash-
ionable to take the stance of a Silicon Valley start-up and assert that the 

---

128
I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N
product is never finished and that every new innovation should be released 
as a beta ready for continuous evolution.
But innovation in an enterprise (seeking to launch a new venture or 
offering or to improve an existing one) is not exactly the same as innova-
tion in a three-person start-up (whose new app may be the entire focus 
of the organization). And not every product can be launched to the full 
public in beta (e.g., think of a car). Some of the principles of experimenta-
tion therefore need to be adapted or translated to the context of an existing 
enterprise. And, in fact, not everything called an experiment is the same. 
Different types of business experiments may not be designed or run in the 
same manner or be used to answer the same kinds of questions. But all 
business experiments do have this in common: they seek to increase learn-
ing by testing ideas and seeing what works and what doesn’t.
Two Types of Experiments
Think back to the two examples we have seen so far: Intuit’s experimenta-
tion to develop Fasal and Google’s experimentation to continuously improve 
its search engine. Both companies are experimenting, but there are many 
differences. Google is testing on the actual product: the real search engine 
used by its customers. With Fasal, Intuit intentionally tested simple mock-
ups and a rough prototype of what an actual product might eventually be. 
Google’s testing is in real time, with thousands or millions of subjects whose 
behaviors can be compared scientifically to identify meaningful statistical 
differences. With Fasal, the experiments were conducted with small groups 
of customers, and the results would not appear to pass muster with any-
one’s statistics teacher (“What’s the standard deviation among five mandi 
agents?”). For Google, the goal of innovation is to improve something 
known. For Fasal, the goal was to develop something completely novel.
In fact, a wide range of practices can be called business experiments. 
The most fundamental difference is between more formal (scientific) 
experiments and the kind of informal experimentation that is common to 
new product development. This is not due just to the organizational culture 
of the business that is doing the experimenting (i.e., experimental “style”), 
nor is it due to the ready availability of a large sample size (even if Intuit had 
access to 1,000 farmers, it wouldn’t have made sense to use a formal scien-
tific experiment). Rather, we can see two types of business experimentation 
that are suited for two types of learning.

---

I N N O V A T E B Y R A P I D E X P E R I M E N T A T I O N

129
I will call these two types convergent and divergent because I prefer 
to name them by their function rather than their form (e.g., formal vs. 
informal). Convergent experiments are best suited for learning that elimi-
nates options and converges on a specific answer to a clearly defined ques-
tion (e.g., Which of these three designs is preferred by the customer?).