Entrepreneurial Innovation at Google

Loading...

Taking too long?

Reload document

| Open in new tab

C OV ER F E AT U RE

Entrepreneurial
Innovation
at Google

Alberto Savoia and Patrick Copeland, Google

To fully realize its innovation potential,
Google encourages all of its employees to
think and act like entrepreneurs.

L arge organizations have enormous innovation

potential at their disposal. However, the innovation
actually realized in successful products and services
is usually only a small fraction of that potential.
The amount and type of innovation a company achieves
are directly related to the way it approaches, fosters,
selects, and funds innovation efforts. To maximize in-
novation and avoid the dilemmas that mature companies
face, Google complements the time-proven model of top-
down innovation with its own brand of entrepreneurial
innovation.

INNOVATION POTENTIAL

The concept of innovation potential is a critical, but
often overlooked, element in the discussion of innovation;
defining and understanding this potential is important
because it is the source of all innovation within a company.

Intellectual. The company possesses significant know-
how and intellectual property in many areas—most
notably in crawling, storing, indexing, organizing, and
searching data on a massive scale and with an extremely
fast response time.

Physical. Google has a network of datacenters as well as
a variety of custom, open source, and commercial hard-
ware and software to harness this computing power and
make it easily and seamlessly accessible to both customer-
facing products and internal tools.

Market. Hundreds of millions of people use Google’s
products each day. These products generate revenue as
well as goodwill that is useful to the company when it
needs to try out, and get feedback on, its latest innovations.
Leveraged. Google fosters an ecosystem that allows
other companies to prosper by providing additional value
and content on top of its services. By lowering the imped-
ance between itself and the outside community, Google
facilitates a symbiotic relationship that enables and accel-
erates innovation for all.

Financial. The company has the ability to invest signifi-
cant capital in many speculative projects and innovative
ideas.

Resources

A company’s innovation potential can be defined as the
combination of its human, intellectual, physical, market,
leveraged, and financial resources. Consider, for example,
Google’s assets.

Human. Google has more than 20,000 employees
spread across several functions such as engineering,
operations, marketing, and sales. In addition to exper-
tise in their field, all Googlers bring to the company their
individual passions and interests, which play a key role
in driving innovation.

Example: Google Search

Google’s Search platform, and the development of the
infrastructure on which it runs, illustrates how the combi-
nation and interaction of all of these resources helped the
company to grow and enabled—and sometimes forced—it
to innovate.

In the beginning, passion for a better way to search the
Internet led to Google’s innovative PageRank algorithm.1
The quality of PageRank results generated a lot of interest
in, and millions of users for, Google in a very short time,
requiring the company to scale its infrastructure as fast

56

computer

Published by the IEEE Computer Society

0018-9162/11/$26.00 © 2011 IEEE

as possible. To keep costs reasonable, Google built its
servers in-house with cheap commodity parts and Velcro
fasteners for fast swapping of components, and it oper-
ated these servers with open source software. At the time,
this approach was considered highly novel. Google had to
develop its own software layers to make its commodity
servers work seamlessly and in a fault-tolerant way. This
led to several additional innovations such as MapReduce,2
the company’s patented software framework to support
distributed computing on large datasets on clusters of
computers.

The highly scalable, distributed, and fault-tolerant
hardware and software infrastructure and tools originally
developed for Search, combined with access to a massive
user base and a growing number of employees, made it
possible for Google to conceive, launch, test, and rapidly
scale many new products like Gmail, News, and Ads.

INNOVATION MODELS

As a company grows, its innovation potential grows
along with it and, more often than not, so does its need
for innovation. The amount and type of innovation that a
company actually realizes are determined by its cultural,
organizational, and technical beliefs and practices with
respect to innovation, which can result in various models.
As Figure 1 shows, two different innovation models
starting from exactly the same innovation potential will
produce dramatically different subsets of actual inno-
vation. While some models produce innovation that is
closely aligned and easily integrated with existing tech-
nology and products, others lead to innovation in areas
peripheral to the organization’s current focus. Some
models result in incremental innovations; others are
highly disruptive.

The concepts of open and closed innovation de-
veloped by Henry Chesbrough3 are examples of such
models. Open innovation involves ongoing collaboration
with, and contributions to and from, people outside the
company. In contrast, closed innovation is kept in-house
and “under wraps” until the product hits the market.
Not surprisingly, a company that favors open innovation
will achieve very different results from one that favors
its counterpart.

Google has many projects that follow either the open
or closed model, and others that do not cleanly fit either
stereotype. Android and Chrome OS are examples of
permeable interfaces between Google and the outside
community, and would be defined as open on the sur-
face. However, both projects periodically “go dark” on
the community to surprise the market. In a sense, they
are both open and closed depending on business needs at
any given time. Google Wave is a good example of closed
innovation because it was developed without significant
external influence.

Innovation
potential

Innovation model A

Innovation model B

Actual
innovation

Actual
innovation

Figure 1. Starting from exactly the same innovation poten-
tial, two different innovation models, A and B, will produce
dramatically different subsets of actual innovation.

TOP-DOWN INNOVATION

Arguably, nothing has more influence on a company’s
future than the innovation models it chooses to adopt. In
addition to open and closed innovation, Google practices
both top-down and entrepreneurial innovation.

Top-down innovation is the default model for most es-
tablished organizations after they reach a certain size and
maturity. It is characterized by several traits including

• the creation of one or more entities focused on re-
search or advanced development—for example,
Bell Labs, Xerox PARC, and Sun Microsystems
Laboratories;

• recruitment of dedicated researchers, including many

PhDs, to staff the research organization;

• a small number of ambitious and often expensive
long-term projects that are usually chosen, or at least
vetted, by the organization’s top layers;

• formal and extensive research proposals, plans, and

reviews; and

• a relatively closed and secretive environment, with
limited sharing of resources and information with
other parts of the company.

An example of top-down innovation at Google is its
Translate technology. Language translation is widely
acknowledged as a very difficult problem in computer
science. Given Google’s goal to organize the world’s in-
formation, having an effective translation technology is
critical to its mission and business. To address this prob-
lem, the company hired many leaders in the industry and
gives them time to innovate at their own pace. Google
integrates this group’s innovations in its products as they
become available—the company recently introduced a
feature that lets Chat users who speak different languages
send instant messages in real time.

Another notable example of top-down innovation at
Google is the ambitious self-driving car project led by Stan-
ford University’s Sebastian Thrun, a pioneer researcher in
this area and codeveloper of Street View.

AprIL 2011

57

The key role of managers at Google is
to guide and connect, not control.

innovation.

C OV ER F E AT U RE

Top-down innovation has produced groundbreaking
results and is irreplaceable for innovation that requires long-
term commitment, substantial investment, and significant
domain expertise. However, it is unsuitable for pursuing
innovation that requires limited time and resources and is
best served by an open, lightweight approach.

ENTREPRENEURIAL INNOVATION

In The Innovator’s Dilemma, Clayton M. Christensen4
described the challenges that established companies, es-
pecially industry leaders, face delivering breakthrough
innovation. As a company grows in size and market stat-
ure, keeping existing products competitive and satisfying
customers’ needs with incremental features can easily
consume all of its resources, leaving the door wide open
for start-ups with disruptive offerings. The dilemma Chris-
tensen poses is more than just a possibility—it is the most
likely outcome for most businesses.

Google believes that the best way to stay on top of the
market and remain competitive over the long term is to
promote, foster, and invest in entrepreneurial innovation
in all areas of the company. The ability to drive, and par-
ticipate in, innovation is not limited to a select few PhDs
working in designated research labs—it is open to all
employees. Further, Google’s entrepreneurial innovation
model mimics, with some obvious and necessary limits,
the experience that entrepreneurs would have in a start-up:
fighting for funding and resources, dealing with competing
products, and, if successful, earning significant financial
rewards for their efforts.

Two core beliefs drive Google’s approach to entrepre-

neurial innovation.

The business of Google is innovation. If you randomly
ask 10 people what business Google is in, most will say
Internet search and advertising. But if you ask Google’s
CEO, Eric Schmidt, or its founders, you will get different
answers, such as “our business is innovation” or “we take
our jobs to be innovators and we are failing if we are not
innovating quickly enough.”

While Google invests heavily in maintaining, sup-
porting, and continually improving already established
products such as Search, Ads, and Gmail, it realizes that to
ensure long-term growth and success, it must also commit
resources to innovation in several areas. Schmidt made
that clear in 2009 when he said that “innovation is the
technological precondition for growth.”5

58

computer

Expect innovation from every employee. Google strongly
believes that innovation can come from any employee
at any time. “We prefer [our engineers] to run rampant,”
Schmidt explained in 2005. “The cleverest ideas don’t
come from the leaders, but rather from the leaders listen-
ing and encouraging and kind of creating a discussion.”6
In his 2009 commencement address at Carnegie Mellon
University, Schmidt said, “You cannot plan innovation,
you cannot plan invention. All you can do is try very hard
to be in the right place and be ready.”7

What does it mean to “try very hard to be in the right
place and be ready,” and what does that take? For Google,
it means organizing the entire company to foster and sup-
port “unplanned” innovation and entrepreneurship.
Google puts these beliefs into practice through

• a flat, data-driven organizational structure;
• a “20 percent time” policy;
• open and powerful development environments;
• services and tools to help launch, test, and get user

feedback as early as possible; and

• generous rewards and recognition for successful

Each of these efforts requires significant commitment,
investment, and participation by almost every company
group and project. As practiced at Google, entrepreneurial
innovation is not a matter of simply making a few tweaks
and adjustments here and there; it is woven into the very
fabric of the company, coloring every activity.

FLAT ORGANIZATIONAL STRUCTURE

The most immediately visible evidence of Google’s
policy of entrepreneurial innovation is its organizational
structure and associated management philosophy.

On the surface, Google is organized and managed like
most other companies. It has different groups—for ex-
ample, engineering, finance, human resources, operations,
product management, and sales—and within each group
are vice presidents, directors, managers, project leads, and
so on. But that is where the similarities end.

One of the first things people new to Google notice is its
very flat management hierarchy. While the company has a
traditional job ladder with familiar titles, it has always tried
to keep the ratio of engineers and other individual contribu-
tors to managers as high as possible. It is not unusual for 30
to 40 people to report directly to a manager, or even to a di-
rector or VP. In addition, the key role of managers at Google
is to guide and connect, not control. As one senior executive
put it, “I am a very expensive e-mail router.” While no two
groups or managers are exactly alike, titles and seniority
do not carry as much weight at Google as they do at most
companies, especially when it comes to making product
decisions and launching or assigning activities.

The most notable effect of Google’s flat management
hierarchy is that, at any given time, there is a certain
amount of chaos. But the company is not only comfort-
able with this, it sees some chaos as a necessary ingredient
for innovation. Shona Brown, Google’s senior VP of busi-
ness operations and coauthor, with Kathleen Eisenhardt,
of Competing on the Edge: Strategy as Structured Chaos
(Harvard Business School Press, 1998), summarized this
philosophy in a 2006 Fortune interview: “The company’s
goal is to determine precisely the amount of management
it needs—and then use a little bit less. … If I ever come
into the office and I feel comfortable, if I don’t feel a little
nervous about some crazy stuff going on, then we’ve taken
it too far.”8

While Google’s organizational structure can result
in, for example, project duplication, it also increases the
number of projects and accelerates their time to market.
By “letting more flowers bloom,” the company can collect
more feedback on what customers consider valuable.

TWENTY PERCENT TIME

Another notable aspect of Google’s culture derived
from its core beliefs is its well-known 20 percent time
policy, which allows engineers to invest roughly a day
each week pursuing projects outside their official area of
responsibility.

The most important thing about 20 percent time is not
how long employees are allowed to spend on side proj-
ects, but that Google encourages them to think and be
entrepreneurial. There is no formal accounting of time
spent—some people use more, others less. Googlers
engrossed in their primary responsibilities may not be
inclined to work on anything else right then; others may
choose to spend approximately a day each week on a side
project or accumulate their 20 percent time over several
months and then spend several weeks in a row on the
project.

Google employees working on 20 percent projects
often join forces and create the internal equivalent of a
small start-up, recruiting their first “employees” from the
company ranks. As in the real world of start-ups, most
20 percent projects do not make it to the next level. But
the few that achieve critical mass eventually lose their 20
percent status and become official Google projects—the
equivalent of a start-up getting venture capital funding.
We estimate that about half of Google’s products, including
Gmail and News, started out as 20 percent projects.

POWERFUL DEVELOPMENT ENVIRONMENTS
One advantage that entrepreneurs outside Google do
not have is access to its unparalleled computing resources
and the ability to use and leverage all of its code. Unlike
other organizations of comparable size, at Google thou-
sands of engineers share a single, gigantic code base. This

means that someone working on Maps can see, use, and
even modify the code developed by colleagues working
on Ads, Gmail, or Calendar—and vice versa. The lack of
“code silos” enables all kinds of code reuse, mashups, and
product cross-pollination that inspire innovation.

Allowing thousands of developers to work concur-
rently on hundreds of projects in an open development
environment of this magnitude and diversity—Google
supports several programming languages and frame-
works—requires a massive investment in tools and
computing power. Predictably, most commercial or open
source software development and testing tools were
never meant to scale to Google’s requirements in terms
of size, diversity, and speed. To address this problem, the
company had no choice but to think like entrepreneurs
and innovators.

Google employees working on 20
percent projects often join forces
and create the internal equivalent of
a small start-up.

Several years ago, a few Googlers got together and cre-
ated the Engineering Productivity organization to design
and implement the tools, infrastructure, and services
required to support the company’s uniquely demanding
development environment. The “customers” of Engineer-
ing Productivity are the thousands of Google developers;
the following statistics indicate the scale and speed of de-
velopment that this organization has made possible:

• 6,000 developers in more than 40 offices,
• 2,000 projects under active development,
• 100,000 builds each day,
• 150 million test case executions each day,
• 20+ code changes per minute,
• 50 percent of code changes every month, and
• a single monolithic code tree with mixed-language

code.

Google’s very-large-scale development and testing
tools are among its most valuable assets—and they sig-
nificantly accelerate the company’s ability to experiment
and innovate.

LAUNCH, TEST, AND FEEDBACK TOOLS

Startups can move fast and take the kind of risks that
their bigger counterparts are unable or unwilling to take. It
is difficult to be entrepreneurial if launching a new product
takes months or years instead of weeks. To foster innova-
tion, Google has a “launch early and iterate” philosophy.
How early? One of the company’s rules of thumb is: “If you

AprIL 2011

59

C OV ER F E AT U RE

are not embarrassed by your first launch, you have not
launched early enough.”

In addition, Google makes internal entrepreneurs’ new
ideas accessible to thousands of coworkers and millions of
external users who are ready and willing to try out applica-
tions and tell the company what they think of them. Two
early-access programs, Internal Labs and Labs, provide
innovators with progressive market exposure and visibility.
As the name suggests, only Googlers have access to In-
ternal Labs. This gives innovators an opportunity to obtain
feedback from 20,000 people on a proposed application,
when expectations are low. Among the categories of Inter-
nal Labs applications is one called “R.I.P.,” the final resting
place for innovation not considered worth pursuing. Most
applications end up here, which shows that Internal Labs
is effective at helping employees test ideas and fail fast.

To foster innovation, Google has a
“launch early and iterate” philosophy.

Innovators who are ready to share early versions of
their innovation with external users can take advantage
of Google Labs. As described in its FAQ (www.googlelabs.
com/faq):

Google Labs is a playground where our more adventurous
users can play around with prototypes of some of our wild
and crazy ideas and offer feedback directly to the engineers
who developed them. Please note that Labs is the first phase
in a lengthy product development process and none of this
stuff is guaranteed to make it onto Google.com. While some
of our crazy ideas might grow into the next Gmail or iGoogle,
others might turn out to be, well, just plain crazy. …

Google engineers and researchers are always looking
for a way to show off their pet projects, and Google Labs
seemed like a great way for them to get feedback without
forcing every new feature on all of our users. So, please
follow the “Details and Feedback” link under each experi-
ment and post a comment to let them know what you think
of how they’ve been spending their time—and be frank. It
doesn’t help anyone if a bad idea is encouraged to spread
like a noxious weed.

Note the emphasis on getting honest user feedback.
While Google Labs provides employees an opportunity to
showcase their innovation with the rest of the world, its
primary function is not self-promotion but weeding out
bad ideas.

Some established products have their own version of
Google Labs where users can experiment with new fea-
tures. Product-specific labs, along with their taglines,
include

• Calendar Labs: Latest ideas from the Calendar team,
• Gmail Labs: Dozens of Gmail experiments,
• Google Maps Labs: Experimental Maps features,
• Search Experiments: Alternate search views and

more, and

• YouTube TestTube: YouTube’s ideas incubator.

The importance of these labs in the context of in-
novation and entrepreneurship at Google cannot be
overemphasized. The expression “say it with numbers” is
an integral part of the company’s DNA. When it comes to
deciding whether to invest in new ideas, there is no more
compelling set of numbers than the actual usage data ob-
tained from labs launches.

REWARDING SUCCESSFUL INNOVATION

One major motivator for entrepreneurs is the prospect
of substantial financial rewards if their start-up is suc-
cessful. While for many Googlers the main incentive for
innovation is seeing their idea become reality and reach
millions of users, employees who take innovation from
idea to successful product receive both monetary and hon-
orary recognition.

The Google Founders’ Award, launched in 2004
to reward outstanding entrepreneurial achievement,
can amount to millions of dollars. The approximately
two dozen recipients of the first award shared around
$12 million worth of stock. As in a real start-up, the shares
were divided in proportion to recipients’ contributions;
the core contributors received awards of $1 million or
more. Google cofounder Sergey Brin explained that the
award was largely created “to give people incentives to
apply for jobs at Google even after the promise of getting
rich from the company’s initial public offering last August
had passed.”9

Google offers many other incentives and awards that
recognize and reward internal entrepreneurship. However,
because only a fraction of innovation succeeds, how the
company deals with failure is just as important as the way
it deals with success.

In the “outside world,” entrepreneurial success—in the
form of venture capital funding, acquisition, or an initial
public offering—is the result of actual market success.
Likewise for Google innovation, user adoption, not opin-
ion, largely determines a project’s future. A good example
of this is the genesis—and termination—of Google Wave.

Wave began as an idea to create a new paradigm for
online collaboration, with the ambitious goal of augment-
ing and possibly replacing e-mail. Its creators pitched the
idea to Google executives, who funded it much like a ven-
ture capital firm would a start-up. Wave received luxuries
not normally accorded to in-house projects including near
isolation (in Sydney, Australia) to allow for greater indepen-
dence, plentiful resources, and a long runway. However,

60

computer

Wave failed to reach and maintain a specific number of
active users by a given time and was cancelled.

While Wave’s end result was disappointing, it dem-
onstrates the effectiveness of Google’s entrepreneurial
innovation model. The application was given sufficient
time and room to flourish and only terminated when it
became evident that it was not popular enough to justify
the further commitment of resources. And the effort
was not wasted: many of Wave’s breakthrough ideas
and technology are finding their way into other Google
products.

Trying something new and not succeeding is an ines-
capable and important part of the innovation process.
Google knows that if it never fails, then it is probably
not being as innovative as it needs to be. When a project
fails to meet expectations, the company acknowledges it,
learns whatever lessons it can, and moves on to something
different.

P utting Google’s entrepreneurial innovation model

into practice requires significant commitment,
investment, and participation from all functions
and areas of the company. For example, the annual cost
of 20 percent time alone is hundreds of millions of dollars.
Overall, however, the ongoing stream of cutting-edge proj-
ects and features that directly result from entrepreneurial
innovation amply demonstrate that the model is working.
Quantifying the costs and benefits more precisely might
be possible, but it would be difficult—and perhaps point-
less. Entrepreneurial innovation is so tightly woven into
everything Google does that is hard to imagine the com-
pany without it.

References
1. L. Page et al., The PageRank Citation Ranking: Bringing
Order to the Web, tech. report, Stanford Univ. InfoLab,
1998; http://ilpubs.stanford.edu:8090/422/1/1999-66.pdf.

2. J. Dean and S. Ghemawat, “MapReduce: Simplified Data
Processing on Large Clusters,” Proc. 6th Symp. Operating
System Design and Implementation (OSDI 04), Usenix
Assoc., 2004, pp. 137-150; http://labs.google.com/papers/
mapreduce.html.

3. H. Chesbrough, Open Innovation: The New Imperative for
Creating and Profiting from Technology, Harvard Business
School Press, 2003.

4. C.M. Christensen, The Innovator’s Dilemma: When New
Technologies Cause Great Firms to Fail, Harvard Business
School Press, 1997.

5. A. Eustace, “Investing in Innovation at Google,” The Of-
ficial Google Blog, 2 Nov. 2009; http://googleblog.blogspot.
com/2009/11/investing-in-innovation-at-google.html.
6. R. McMillan, “Loosen the Reins, Says Google CEO,” Info-
World, 19 May 2005; www.infoworld.com/t/platforms/
loosen-reins-says-google-ceo-123.

7. E. Schmidt, Carnegie Mellon Univ. 112th commence-
ment speech keynote, 17 May 2009; www.cmu.
edu/cmufront/homepage/images/extras/transcripts/
ericschmidtcommencementkeynote.pdf.

8. A. Lashinsky, “Chaos by Design,” Fortune, 2 Oct. 2006;
http://money.cnn.com/magazines/fortune/fortune_
archive/2006/10/02/8387489/index.htm.

9. K. Hafner, “New Incentive for Google Employees: Awards
Worth Millions,” The New York Times, 1 Feb. 2005; www.
nytimes.com/2005/02/01/technology/01google.html.

Alberto Savoia is Engineering Director and Innovation
Agitator at Google. His research interests include rapid
prototyping tools, mobile computing, and developing ob-
jective metrics that help innovators predict the success
and adoption of new products or services. Contact him at
asavoia@google.com.

Patrick Copeland is a Senior Engineering Director at Google
and leads the Engineering Productivity Group. His research
interests include testing, developing, and creating tools for
large cloud-based applications and systems. Contact him at
copeland@google.com.

Selected CS articles and columns are available
for free at http://ComputingNow.computer.org.

LISTEN TO GRADY BOOCH
“On Architecture”

podcast available at http://computingnow.computer.org

AprIL 2011

61

This article was featured in

For access to more content from the IEEE Computer Society,
see computingnow.computer.org.

Top articles, podcasts, and more.

computingnow.computer.org