Professional Documents
Culture Documents
Buidling A Culture of Experimentation Case Study
Buidling A Culture of Experimentation Case Study
Article
Innovation
Building a Culture
of Experimentation
by Stefan Thomke
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
Harold Edgerton was known for his
experiments with high-speed photography
and used stroboscopic equipment to
capture moments in time.
Building a Culture
of Experimentation
It takes more than good tools. It takes
a complete change of attitude.
I N D E C E M B E R 2 0 1 7 , just before the busy Gillian Tans, Booking.com’s CEO
holiday travel season, Booking.com’s at the time, was skeptical. She worried
Stefan Thomke director of design proposed a radical that the change would cause confusion
Professor, Harvard
Business School experiment: testing an entirely new among the company’s loyal customers.
layout for the company’s home page. Lukas Vermeer, then the head of the
Instead of offering lots of options for firm’s core experimentation team,
hotels, vacation rentals, and travel deals, bet a bottle of champagne that the test
Harold Edgerton ©2010 MIT. Courtesy of MIT Museum
as the existing home page did, the new would “tank”—meaning it would drive
one would just feature a small window down the company’s critical perfor-
asking where the customer was going, mance metric: customer conversion,
the dates, and the number of people or how many website visitors made a
in the party, and present three simple booking. Given that pessimism, why
options: “accommodations,” “flights,” didn’t senior management just veto
and “rental cars.” All the content and the trial? Because doing so would have
design elements—pictures, text, but- violated one of Booking.com’s core
tons, and messages—that Booking.com tenets: Anyone at the company can
had spent years optimizing would be test anything—without management’s
eliminated. permission.
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
Booking.com runs more than 1,000 “In an increasingly digital world, if that succeeds, nearly 10 don’t—and in
rigorous tests simultaneously and, by you don’t do large-scale experimentation, the eyes of many organizations that
my estimates, more than 25,000 tests in the long term—and in many industries emphasize efficiency, predictability, and
a year. At any given time, quadrillions the short term—you’re dead,” Mark “winning,” those failures are wasteful.
(millions of billions) of landing-page Okerstrom, the CEO of Expedia Group To successfully innovate, compa-
permutations are live, meaning two told me. “At any one time we’re running nies need to make experimentation an
customers in the same location are hundreds, if not thousands, of concur- integral part of everyday life—even when
unlikely to see the same version. All rent experiments, involving millions of budgets are tight. That means creating an
this experimentation has helped trans- visitors. Because of this, we don’t have to environment where employees’ curiosity
form the company from a small Dutch guess what customers want; we have the is nurtured, data trumps opinion, anyone
start-up to the world’s largest online ability to run the most massive ‘customer (not just people in R&D) can conduct or
accommodation platform in less than surveys’ that exist, again and again, to commission a test, all experiments are
two decades. have them tell us what they want.” done ethically, and managers embrace
Booking.com isn’t the only firm to But in studying more than a dozen a new model of leadership. In this
discover the power of online experi- organizations and analyzing ano- article, I’ll look at several companies that
ments. Digital giants such as Amazon, nymized data on experiments from have managed to do those things well,
Facebook, Google, and Microsoft have upwards of 1,000, I have seen that Book- focusing in particular on Booking.com,
found them to be a game changer when ing.com, Expedia, and their ilk are the which has one of the strongest cultures
it comes to marketing and innovation. exception. Instead of running hundreds of experimentation I have found.
They’ve helped Microsoft’s Bing unit, or thousands of online tests a year, many
for instance, make dozens of monthly firms run no more than a few dozen that
improvements, which collectively have have little impact.
CULTIVATE CURIOSITY
boosted revenue per search by 10% to If testing is so valuable, why don’t Everyone in the organization, from the
25% a year. (See “The Surprising Power of companies do it more? After examining leadership on down, needs to value sur-
Online Experiments,” HBR, September– this question for several years, I can tell prises, despite the difficulty of assigning
COPYRIGHT © 2020 HARVARD BUSINESS SCHOOL PUBLISHING CORPORATION. ALL RIGHTS RESERVED.
October 2017.) Firms without digital you that the central reason is culture. a dollar figure to them and the impossi-
roots—including FedEx, State Farm, and As companies try to scale up their online bility of predicting when and how often
H&M—have also embraced online test- experimentation capacity, they often they’ll occur. When firms adopt this
ing, using it to identify the best digital find that the obstacles are not tools mindset, curiosity will prevail and people
touchpoints, design choices, discounts, and technology but shared behaviors, will see failures not as costly mistakes
and product recommendations. beliefs, and values. For every experiment but as opportunities for learning.
IDEA IN BRIEF
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG
It’s actually less risky to run a large number of experiments than a small
number. If a company does only a handful of experiments a year, it may have
only one success—or none. Then failure is a big deal.
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.
FOR ARTICLE REPRINTS CALL 800-988-0886 OR 617-783-7500, OR VISIT HBR.ORG
SCENARIO #1 SCENARIO #2
Hypothesis Hypothesis
Highlighting a neighborhood’s walkability helps users make better Displaying the checkout date when
decisions about property location. users select the age of children in their
party improves their experience.
A B A
The Control The Treatment The Control
Shows the site’s current practice Adds walkability information Shows the site’s current practice
B
The Treatment
Adds the checkout date above children’s ages
In 2015 experimentation wasn’t a core Then, Ari Sheinkin, IBM’s head of mar- offered training for everyone, and made
activity at IBM; the company’s IT func- keting analytics at the time, took over online tests free for all business groups.
tion offered to run tests, but they were experimentation and, with the backing He also conducted an initial “testing
costly, were charged back to business of the chief marketing officer, empow- blitz” during which the marketing units
units, and had to follow a rigid process. ered over 5,500 marketers worldwide had to run a total of 30 online exper-
The testing capacity consisted of just one to conduct their own tests. To induce iments in 30 days. After that, he held
specialist, who was also the gatekeeper them to do so, Sheinkin took a number quarterly contests for the most innova-
and who rejected many proposed exper- of steps. He installed easy-to-use tools, tive or most scalable experiments. He
iments because he felt that they weren’t created a center of excellence to provide also employed more-forceful tactics:
strong-enough candidates. As a result, support, introduced a framework for IBM tied part of marketing units’ budgets
the company ran only 97 tests that year. conducting disciplined experiments, to experimentation plans. These efforts
R E A L I Z I N G T H E T R A N S F O R M AT I V E
power of experimentation requires
a sustained commitment. Over time
experiments will result in thousands
of small and not-so-small changes that
collectively generate huge benefits.
Providing the right tools, while essen-
tial, is the easy part and isn’t enough
to make experimentation a way of
life. Vismans put it best: “If I have any
advice for CEOs, it’s this: Large-scale
testing is not a technical thing; it’s a
cultural thing that you need to fully
embrace. You need to ask yourself two
big questions: How willing are you to be
confronted every day by how wrong you
are? And how much autonomy are you
willing to give to the people who work
for you? And if the answer is that you
don’t like to be proven wrong and don’t
want employees to decide the future of
your products, it’s not going to work.
You will never reap the full benefits of
experimentation.”
The lesson is that it’s not so import-
ant whether any one experiment
succeeds or fails; what matters is how
decisions are adjudicated under uncer-
tainty in an organization. They should
not be based on faith or personal opinion
alone. If they can be put to the test, they
should be. HBR Reprint S20021
This article is part of a series. The complete Spotlight package is available in a single reprint. HBR Reprint R2002B
This document is authorized for use only in Murugan. P's Business Experiments for Decision Making at Indian Institute of Management - Shillong from Jan 2022 to Jul 2022.