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

Myth 2: Correlation Is All That Matters
Spotting a pattern is not (always) enough. Some commenters on big data have 
reported that data science is no longer concerned with causation, just correlation. 
The belief is that underlying patterns across data sets are a truth unto themselves 
that does not need to rely on foggy human ideas of cause and effect.

---

T U R N D A T A I N T O A S S E T S

103
This is simply not true. It is critically important that managers understand the 
difference between simple correlation and causation—and know when this differ-
ence matters and when it doesn’t. A simple rule of thumb: if you are only making 
predictions, data correlation is sufficient. But if you are looking to change the 
precondition, you need to know there is causation as well.
Think of Stringer, the city comptroller who discovered the data correlation 
between declining budgets for tree pruning and rising lawsuits against the city. 
If the tree-pruning budgets weren’t actually causing the accidents that led to 
lawsuits, his decision to restore the pruning budget would not have helped. In 
Stringer’s case, causality mattered a great deal.
On the other hand, imagine your ad agency has determined that married 
women in Ohio are more responsive to advertisements for your new hair care 
product. You are not going to try to grow your shampoo sales by encouraging 
Ohioans to get married (that would be influencing the precondition). You are 
just going to use this information to target more of your ads to married Ohioans 
instead of single ones. In a case like this, simply knowing a data correlation is fine.
Myth 3: All the Good Data Is Big Data
It would be a mistake to conflate big data with data strategy. In many cases, com-
panies can build valuable data assets and apply them to strategic ends without 
delving into the messy world of big data.
Data does not always need to be “big” (i.e., unstructured) in order to be useful 
to a business. Powerful insights can be derived from the analysis and application of 
traditional, more structured data such as customer clickstream behavior (Where 
do customers click on a website, scroll down the page, spend more or less time, 
put things in shopping carts, etc.?). Even at a big-data powerhouse like Facebook, 
home to some of the biggest server clusters in the world, most queries run by 
engineers on a given day are of a scale that could be processed on a good laptop.
8
The point of your data strategy should be to generate value for your customers and 
business. Sometimes that will involve big data, and sometimes it won’t.
Where to Find the Data You Need
As you begin to put together a data strategy, you will start with the data 
you are generating in your own business processes. However, you will likely 
identify gaps in the data you need for some of your goals. Finding the right 
additional sources of data is critical to filling in gaps and building your data 

---

104
T U R N D A T A I N T O A S S E T S
asset over time. Important sources of data from outside your organization 
include customer data exchanges, lead users, supply chain partners, public 
data sets, and purchase or exchange agreements.
Customer Value Data Exchange
One of the best ways to generate additional data is to invite customers to 
contribute data as part of interacting with your business or in direct exchange 
for value you offer them. As mentioned in chapter 2, the navigation app 
Waze built both its map data and its real-time traffic data through user con-
tributions. Waze was designed from the beginning around generating data. 
Whenever a customer has the app turned on, it is pinging their phone’s GPS 
once a second. In densely populated areas, this approach provides excep-
tional real-time awareness of traffic conditions and allows for superior 
rerouting compared to competitors’ apps. (After it reached 30 million users, 
Waze was bought by Google for $1.3 billion.) Because it does not sell directly 
to consumers, Coca-Cola historically has had little consumer data. But with 
the help of its MyCokeRewards loyalty program, the company has built up 
a data view on 20 million of its customers, the linchpin of its data asset. The 
Metropolitan Museum of Art was able to gather 100,000 new, valid e-mail 
addresses simply by asking visitors for their e-mail addresses in exchange for 
access to the Met’s free Wi-Fi. What makes consumers willing to share their 
information with businesses? In a global research study that I conducted at 
Columbia University with Matt Quint, we observed four key factors: the type 
of value or rewards offered, the presence of a trusted relationship with the 
business, the type of data being requested, and the industry of the business.
9