The Data Trap

Data is a drug, you always want more but it can be deadly

• Data driven innovator

• Data driven marketing consultant

• Data driven growth

• Data and AI driven specialist (this is a new one)

These are just a few of the job titles that people use on Linkedin these days. I always thought that if data driven is a positioning, is ‘not being data driven’ also one? :)

And if not, shouldn’t everyone just be data driven?

Well we are not here to discuss this. We are here to discuss the data trap when it comes to creating strategy.

There is no denying the fact that data is important. It is important for decision making, it is important for tracking progress, it is important for resource allocation. But it cannot answer everything. Further it is prone to bias, even when analysing historical data.

Data is biased to interpretation and impacted by variables

I was recently reading two different books - The Blue Ocean Strategy and Freakonomics. Both are well revered and best sellers in their own domain. Now both books approach the topic of “fall in the crime rates in New York in 1990s” with two different interpretations of the same data set.

The Blue Ocean Strategy paints a rosy picture of the NY police commissioner at the time, William Bratton, and how his strategies created a blue ocean on how policing should be done leading to a drop in crime rates without an increase in additional resources via identifying hot spots, resource reallocation and what not. On the other hand, Freakonomics dismisses that Bratton and his policies had any major effect on the NY crime rate and calls it a minor contributor at best.

So you see, even with all the available data around this historical event, two authors have formed two completely opposite views of the same event and its causation.

Why is that? Well for starters no one is untouched by bias, be it a strategist or an economist. Your bias always walks around with you, trotting proudly and making observations even when not asked for an opinion.

Your bias is always beside you!

Secondly, the variables are so many that even when you account for and try to take care of them, you will still not be able to take into account the 100 other variables which are either not known or have no data available for them.

So when we find it so hard to interpret historical data, how about using data to predict the future, which is where strategy plays out?

Data projections are usually off by miles

The prediction that car market would be limited to the ultra wealthy was thrown out of the window when Ford came out with Model T.

The prediction by Mckinsey that mobile phone markets will be limited to 900,000 by 2000 was shattered as the numbers hit 109,000,000 in the US only.

The prediction was that Segway would be selling 10,000 units per week and it never even got close.

It is generally hard to predict the future and no matter how much data you use or how smart you are, chances are that you will be wrong.

The Arabic proverb aptly describe the use of data to predict the future

“He who predicts the future lies, even when he tells the truth”

So does it mean that you cannot trust data and should not use it? Well not really.

Data projections are very good at projecting the future demand for markets that are more stable and replacement driven. So for example, how many shoes are going to be sold next year or how many cars are people going to buy. But they falter (both overshoot and undershoot) when they are used to predict demand for a more disruptive technology (product or service).

Data is a clue not a crutch

Data should be used as one of the clues to solve a puzzle and not a crutch to support your entire strategy on.

A lot of managers out there want data driven business cases to justify their plans. Well I have news for them. Any business case can be made to look profitable depending on how you want to read data. If you do not believe me ask Masayoshi San about the WeWork business case I guess.

My proposal is this:

- Use data to help identify your biggest challenges (where are you leaving money on the table, operational challenges are best identified with data)

- Use data selectively to identify your biggest opportunities (if your market is stable and replacement driven, data is great. If it is a new or untapped market, data might be useful only directionally)

- Conduct experiments to create data for new markets. (better served by pilots instead of large scale capex heavy projects)

- Use data for tracking progress (outcomes are very well served by data analysis)

Remember, data is a tool. It is a means to an end and not the end itself.

---

So this was it for this week. Do share your thoughts about data and its usage in strategy. If you like this and want to read more stuff like this, do subscribe and share. See you all next week!