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

Divergent Experimental Method
6. Build an MVP
Minimum cost
Maximum learning
2. Set limits
Time
Scope
Money
8. Decide
9. Scale up
10. Share learning
Proceed
Four paths to scaling
Pull the plug
Prep to launch
Pivot
3. Pick your people
1. Define the problem
4. Observe
5. Generate more than one solution
7. Field test
A
c
tion
It
er
ation
Preparation
Figure 5.3
The Divergent Experimental Method.

---

148
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
Step 1: Define the Problem
The first step of a divergent experiment is to define the problem you are 
seeking to solve. The problem should be rooted in an observed customer 
need or market opportunity and be a challenge that your organization is 
particularly well suited to solve. The advantage of defining your innova-
tion in terms of a problem is that it forces you to take the customer’s point 
of view. Your innovation should always focus on delivering value to the 
customer (even if that customer is an internal constituency) rather than 
on deploying the latest exciting technology or product feature or defeating 
your competitors.
The problem definition may include a quantified goal, but that goal 
should be both challenging and broad. Recall the experimentation that 
led to Intuit’s Fasal product: the defined goal was to raise Indian farmers’ 
income by 10 percent. This allowed the team wide latitude in thinking about 
how to reach it. When Steve Jobs tasked his team at Apple to develop the 
first iPod, he challenged them to help customers “put 1,000 songs in their 
pocket.” Notice that the challenge is not technical (“fit this much memory 
on a hard drive this size”) but describes the benefit or experience from the 
customer’s point of view.
Step 2: Set Limits
The second step is to set limits for your innovation process. Because diver-
gent experimentation is iterative and because we are naturally inclined to 
defer or delay before admitting failure, it is easy for your innovation project 
to keep running even when the prospects for success are dim. It is therefore 
essential to set limits at the outset.
Any divergent experiment should begin with three kinds of limits 
defined:
r
 Time limit: Finite time should be allotted for the project and its key 
approval stages. Many companies, including Mondelez, AT&T, Intuit, 
and Amazon, use three months as a limit for iterative project develop-
ment before a crucial decision is made on whether to proceed.
31
r
 Money limit: Budgeting for innovation projects is often best done 
in approval stages. IDEO charges clients for each stage of iterative 

---

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

149
product development, requiring buy-in before moving on to the next. 
As assumptions are tested and project risks are reduced, additional 
budget can be released.
r
 Scope limit: Companies should define up-front what they are not seek-
ing to accomplish. This provides helpful boundaries for even the most 
wide-open experiments. For Intuit’s Fasal project, the desired product 
and business model were unknown, but the target market (rural Indian 
farmers) established critical boundaries.
Step 3: Pick Your People
The last step of the preparation phase is to pick which people will work on 
your innovation experiment.
The first question is the size of your team. As a general maxim, an inno-
vation team should be as small as possible—but no smaller. Intuit’s popular 
SnapTax product was developed by a team of three people.
32
Jeff Bezos is 
famous for his “2 Pizza Rule” at Amazon: no meeting is to take place if the 
number of participants is too great to be fed with two pizzas. In my own 
experience running strategy workshops both within and across companies, 
a five-person team is usually ideal for innovation. J. Richard Hackman has 
studied team collaboration and found that the number of network links 
between team members poses an upper threshold for effective group size. 
As the number of group members increases linearly, the necessary lines of 
communication increase exponentially, as n(n – 1)/2. Hackman advises that 
a group of five is ideal and warns against ever going above ten.
33
In addition to size, diversity of team composition is crucial. This 
should include diverse skill sets that relate to the nature of your project. 
(For example, an innovation team working on new service options for a 
bank might include team members with backgrounds in IT, consumer 
behavior, employee training, and service design.) You should also strive to 
include participants with diverse biases and backgrounds. Look for people 
who don’t always work together or who may come from different parts of 
your organization. Include recent hires as well as someone who knows your 
organizational culture well.
It is valuable to change the innovation team over time rather than keep-
ing the same group for every project. You may want to introduce an ele-
ment of competition as well, with multiple small teams competing (at least 
in the initial stages) to develop the best solution to a common challenge.

---

150
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
You have now completed the preparation phase of the Divergent 
Experimental Method. Next come steps 4–8, the heart of your experiment. 
They will be done not once but in an iterative cycle until a decision is made 
to either terminate the project or move on to a public launch.