Let’s say you have a food delivery app. One use case is ordering lunch at work, another common use case is young couples ordering dinner at home, a third one is ordering food for a dinner party.

The time of day and the order size give you enough to bucket most people fall into one or more use case.

The problem is that I threw a dinner party recently and I ordered two of my favourite dishes and we cooked the rest ourselves. There’s no way your app could have known I was throwing a party.

After thinking about this a lot I’ve come to the conclusion that if you can’t tell, it doesn’t matter. If you define the dinner party use case as ordering 8+ items then it’s only relevant when I order 8+ items.

The key here is to define your use cases from the business’ point of view. Then you have to establish unique criteria to measure each use case. The payoff is that you can then segment your behavioural data by user intent.

I’ve found this way of thinking about segmentation helpful because it leads to a productive conversation about relevant product changes.

For example, we now know that people who order lunch at work are the most likely to continue using our app six months later, as compared to our other use cases. How can we make our product better for this use case? When can we start interviewing people in this segment to understand their sticking points? Does personalising the product experience make sense here? Could we run an experiment to see if we can encourage people at work to order lunch as a group?

Compare this to aimlessly segmenting your data by easily available information. Looks like most of our users prefer Chrome…on a laptop…in the US. Interesting.

Segmenting your data by user intent helps tame what can sometimes be an ocean of behavioural information to highlight regions of insight that can lead to productive conversations about relevant product changes.


Other questions to ask yourself to improve your retention…