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

Divergent Experiments
Divergent experiments are generally not built around a causal question. 
Looking back at Intuit’s development of Fasal, at the beginning of its 
experimentation the question was quite broad: “How can we increase the 
revenue of rural Indian famers?” It was far too early to form any specific 
hypotheses about a choice between two product features or marketing mes-
sages or design layouts.
Once the Intuit team had some initial solutions in mind and began to 
prototype and present them to possible customers, they were not looking 
to measure customer response in terms of a single number. They were 

---

132
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
looking for a range of qualitative feedback: “It was confusing.” “I would 
use this only if others were already using it.” “I don’t know what to do 
with the information I am seeing.” “I like this, but I need it more quickly.” 
And so on.
The process of divergent experimentation is therefore more informal 
than that of convergent experimentation. But that does not mean divergent 
experimentation is simply ad hoc. It is still structured and benefits greatly 
from a clear process for conceiving of options or ideas, creating meaningful 
prototypes, testing these to gather real-world feedback on critical assump-
tions, and using that information to make decisions about whether to pro-
ceed and how to launch an eventual solution.
Common mistakes in divergent testing mostly center on testing too late, 
as when “product testing” of a new innovation occurs after development is 
nearly complete. In these cases, because of resources already committed 
and organizational momentum, the testing serves merely as “validation” for 
a course of action that has already been committed to.
Some of the key writers on divergent experimentation include Nathan 
Furr and Jeff Dyer (for established businesses) and Eric Ries and Steve 
Blank (for start-ups).
10
Why You Need Both
To innovate successfully, you will need both convergent and divergent 
experiments at different stages and in different parts of your business. Suc-
cessful innovation must balance both exploratory learning (to generate and 
develop new ideas) and confirmatory learning (to verify and refine ideas). 
A/B testing alone will never tell Wawa what new food product it should 
try in its stores, nor will it write the e-mail subject lines to be tested by a 
political campaign. Likewise, showing iterative design prototypes to cus-
tomers in a lab will never tell you what the final pricing should be, what the 
optimal marketing mix is, or how customers will behave with your product 
once they are using it in the real world.
To some degree, the type of experiment you use may be shaped by the 
area of your business in which you are innovating. For innovations intend-
ing to improve your existing core business, you are more likely to rely on 
convergent experiments. For innovations intending to develop new busi-
ness areas and generate substantially new products, services, or processes, 
you are more likely to rely on divergent experiments.

---

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

133
The two types of experimentation may also happen at different stages 
within the same innovation project. Imagine you are a financial services 
company planning to offer a new mobile app to assist customers in their 
financial planning. You might start with an iterative divergent process to 
test out broad ideas, learn what does and doesn’t work, and develop the 
core value proposition and focus of the new innovation. Then, as you final-
ize the design, you might shift to a convergent process to test and optimize 
key elements (features, design, pricing, marketing messages for the launch). 
Once your app is in the market and you have established a large user base, 
you can apply more convergent experiments to determine what features are 
adding the most value for the customer, driving repeat customer engage-
ment, and increasing customer retention or lifetime value.