Hypothesizing (Segmentation)

Now that we’ve gotten context and clarified each event, let’s take an example of a hypothesis we could propose and think about how to validate it.

Hypothesizing

One variable we can consider is that if fewer new users are joining the platform, this may account for fewer friend requests. If we had fewer new users joining the platform, this would imply that fewer friend requests are being made since our hypothesis states that new users would usually be the ones making the most friend requests.

We can then propose a metric analysis to understand if this is the case. We can look at the number of friend requests sent per user cut by:

• Users that joined a week ago
• Users that joined a month ago
• Users that joined a year ago
• Users that joined over a year ago

These segmentations would tell us if the user’s age is correlated with the friend request metric. Lastly, we could look at the daily active users from the number of total users in each of these buckets over time. If the number of new users joining the platform decreases, this would likely cause a loss in friend requests.

Common segmentations

The above hypothesis was just one example of many that we could dive into. In this case, we segmented our users by newer users and existing users. All product problems are vague and require segmenting in different ways to develop multiple hypotheses. The more hypotheses and validations we think of, the better our response will be.

Here’s a quick cheat sheet of segmentations we can make.