Everybody is talking about it, but what are these cohorts, and how can they help my business? 🤯
A cohort is a group of people grouped together due to a common identifier.
This might be the year they were born in, or how much they spend on shopping each week, or the frequency with which they exercise. Any metric can be used to create a cohort, but it usually refers to a time-dependent grouping. Cohorts are more specific than standard demographic groupings, as they use specific markers to segment the groups.
So a cohort analysis is the study of behaviors of groups over time – know in research as longitudinal study.
Cohorts can help entrepreneurs better monitor their product performance, and make it possible to compare users over time, gauging performance to find out what’s shifted and which campaigns are working well. They are a valuable tool for precisely targeted marketing campaigns, making it possible to tailor and adapt messaging and activity. They allow you to identify potential issues and build more accurate forecasts.
How To Read Cohorts?
Lets take an easy to understand example: Imagine you were looking at how the employment rate evolved during the life of different cohorts.
People who turned 25 between 1974 and 1978 would be cohort one. People who turned 25 between 1979 and 1983 are cohort two etc. What’s crucial to remember is that people wouldn’t stay static. As they aged their employment rate evolved.
To analyze the evolution of a precise cohort, you would read the row. Whereas to analyze how cohorts are evolving, you would read the column.
By breaking down cohorts into granular sets you can get much more specific nuances and identify patterns that you might not be able to when looking from a macro perspective. Looking at the data can help you identify how user behaviors affect your business, understand acquisition and retention and the resulting customer churn, calculate customer lifetime value, optimize the conversion funnel and create effective customer engagement strategies.
Different Types Of Cohorts
When it comes to technology, cohorts indicate how valuable and sticky your users are becoming, and helps you identify odd behaviors or particulary issues. There are two types of cohorts investors like to see, showing user stickiness and product stickiness.
1. User Stickiness
First organize cohorts by periods (typically by month, trimester, or semesters) and then add in the second column the actual numbers of users who signed up. The third column is the % of users active the first period. That number is always 100%. The fourth column are the number of users still active divided by the number of users who originally signed up. If some people have stopped using the service, hence you are experiencing churn. So basically this cohort analystic shows if your users are sticking around and like the product. It validates whether you’re getting product buy in.
You can then find the lowest-performing cohorts, and identify what factors are driving this performance. It could be ad content, channels, campaigns, sales, service offerings – all sorts of activity can impact the cohort, and by running a cohort analysis you can work out what’s working – and what isn’t.
2. Product Stickiness
Product stickiness is different. The goal is to calculate how much your remaining users are spending compared to your first period, and it acts as a tool to highlight whether you can upsell to your clients on additional products or services.
The first column are the cohorts organized by periods (typically by month, trimester, or semesters). The second column are the 1st period revenues from users who signed up. It is basically the number of users who signed up multiplied the average revenue per user. The third column are the revenues from the cohort from the first period divided by the revenues from that same cohorts at sign up. That number is always 100%. Then in the next are the revenues from the cohort from the second period divided by the revenues from that same cohorts at sign up. So as users churn, that number goes down unless those who decide to stay add more accounts, or pay for product add-ons. If your product cohorts show increasing numbers 100%+ you are experiencing Net Negative Churn.
Venture Capital funds love it! 😍
Building Your Own Cohort Analysis
In order to perform a proper cohort analysis by your own, there are four main stages:
💡 Work out what question you want to answer, so that you get actionable information that you can act on at the end.
💡 Define the metrics that will enable you to answer that question.
💡 Work out which specific cohorts are relevant – not everyone will be.
💡 Perform the analysis. As part of this, always check the answers make sense.
There are lots of tools out there to help you, and if you’re looking at online marketing campaigns, Google Analytics can be very helpful. This article from Christoph Janz provides you with an Excel spreadsheet you can use, showing you where to enter the raw data and how to read it, and here Andrew Chen explains what you can do with an Excel sheet.
Range User Retention
One metric you might want to look at is range user retention. First you define a period of time, say a year, and then break it down into smaller chunks, like months. Then you will start to ask questions – users who made first contact in January: what proportion came back one, two … 11 months later. Those in February, who came back one, two and 10 months later. When it comes to November, we look at those who came back one. Here we calculate a triangular table containing relevant retention rates. You can look at the specific calculations here.
What Software Can Help You
If you’re really into it, you can’t create all cohorts in this granularity manually. But who has time for that? RetentionX does all the work for you and compares tons of cohorts by unlimited number of segments with each other, so that you can evaluate and assess each one to the max. There are an unlimited number of metrics such as revenue, activity, customer lifetime value and more that can be used. And you don’t have to rely on historical data – RetentionX can use real time cohorts with fresh data every day.
Sources:
Raphaël Vannson, „Range-user-retention. What is it? Why use it? How to calculate it?“
Augustin Sayer, „Cohorts, cohorts, cohorts!“
