The Thomke Talks Episode 5: Why analyzing data is not enough

A lot of companies now, they are interested in data analytics.

And they have data scientists, and they've invested a lot. And That's a really important field, but there was only one problem with it. And that is often in data analytics, we have to actually rely on existing data.

Now, here's sort of a set of issues that we typically run into, and that's why we need experimentation.

First of all, when it comes to innovation, we often don't have the data because if the data was already there, it wouldn't be very innovative because someone has already done it before, so it's not innovative. So by definition, innovation means that we usually don't have the data to analyze to begin with. Second of all, context matters.

Just because something works in one sort of area, doesn't mean that it works in another area. And then finally, when we do data analytics, we usually get correlations.

You know, that is variables correlate with each other.

But correlations is not the same as causation.

The most fundamental question that we need to know as managers is whether an action that we want to take, a management action or decision actually causes some outcome that we sort of aspire to. The only surefire way of actually addressing causality is to run rigorous controlled experiments.

So if you have a data analytics program and you're just relying on existing data, you're you're you're not there yet. What you have to do is you actually have to add experimentation to your toolbox as well, and the two will actually augment each other. Yep. So that's what it's all about.