Gain better customer engagement with cohortsLast edited: Mar 1, 2022
Ruben Ugarte is the founder of Practico Analytics which helps software companies implement an analytics stack and tools like Mixpanel. You can learn more about his company and the work they do with Mixpanel at practicoanalytics.com/mixpanel.
When it comes to data analysis, cohorts is one of the most underused features that we see with our clients. This is a shame because it is one of the most powerful ways to better understand your customers. Cohorts give users the power to break the large shapeless mass that represents everyone who has ever used a product into clearly-defined groups.
In this article, I’ll show you how to start creating cohorts and how to use them to analyze your data. Let’s start with the basics: what cohorts are and how to create your first one.
What are cohorts and how should you create them?
Cohorts are groups of users who share similar characteristics (user attributes or actions they took) over a period of time. Grouping your users in this way allows you to isolate certain behaviors and analyze them objectively.
To better understand this, let’s create 3 cohorts that apply to nearly every product.
- New Users: this cohort will include any user who signed up in the last 30 days.
- Active Users: this cohort will include any user who is active (as defined by your team) in the last 90 days and signed up more than 30 days ago.
- Inactive Users: this cohort will include any user who isn’t active in the last 90 days and signed up more than 30 days ago
With these 3 cohorts, we can now analyze certain parts of our product and see how these different groups behave. For example, let’s say that you just released a new product feature. Instead of seeing general feature adoption, you could see how specific cohorts are adopting this feature e.g. are new users adopting the feature as fast as active users?
You can also create cohorts based on dates. This is commonly used for retention analysis where you can group users based on their sign up date. You can then see if your retention is improving with each cohort.
In the screenshot above, cohorts are being created on a weekly basis based on an initial action (typically a sign-up event). We are then able to track the retention of each weekly group over time and see how it compares to future groups to answer questions such as: is week 4 retention improving over time?
In Mixpanel, you can create cohorts based on actions they took (events) or user attributes (people properties). You can also define multiple conditions or criteria e.g. a user signed up in the last 30 days AND it was done on mobile.
Besides using cohorts as segmentation options, you can also use them to improve your messaging campaigns.
Personalized messaging using cohorts
Once you have created your cohorts, you can also start to think about what kinds of messages would make sense for each one. For example, users in the “New Users” cohort could benefit from receiving messages that will help them onboard or see the value of your product.
On the other hand, users in the “Active Users” cohort already know the value of your product but could benefit from messages that can convert them into Power Users (which could be another cohort).
Finally, let’s look at a few other ways you can use cohorts to slice your data.
Comparing cohorts and finding distinctions among them
Building on our example from before, you can use cohorts to better understand why some users love your product while others abandon it after only a few days.
You can do this by analyzing your critical events and seeing how different cohorts perform them. In the example below, we can look at a critical event and the average times it is performed by our two cohorts: Active and Power Users.
You can do the same with other reports like Funnels. Let’s say that you have a key funnel inside of your product and you want to see how the overall conversion rate differs by cohort.
In the image above, we can see that the cohort of “Power Users” is more likely to make it to the end of this funnel than “Active Users”. Instead of simply looking at averages, we can start using cohorts to better understand how different groups of users are using our product.
Now get started
As you can see, there is a lot of different ways in which cohorts can help you better understand your customers. Start by creating 1-3 cohorts and working them into your regular analysis. You can also start to optimize your messaging to target specific cohorts.
There’s no such thing as one-size-fits-all when it comes to your product. There are specific groups that each deserve personalized messaging based on what they need to get more value out of your product. These groups can also help shape your product roadmap and overall company strategy.