cohort analysis
Cohort analysis is grouping customers by when they joined you, so you can see how each group behaves over time instead of guessing from one big blur of numbers.
In plain English
Why did the customers who signed up in January spend more than the ones who signed up in March, even though you ran the same advert for both?
That question right there is what cohort analysis was built to answer, and once you start asking it, you cannot really go back to looking at your numbers the old way.
Here is the plain version. A cohort is simply a group of customers who share a starting point, usually the month or week they first bought from you or first downloaded your app.
Cohort analysis means you track each of those groups separately over time, instead of dumping everyone into one big number that hides more than it shows.
Say you run a skincare business in Lagos and you sell through Instagram. In January you got 200 new customers, and in February you got 200 more. If you only look at total sales, you might think business is steady. But cohort analysis lets you check how many of the January group came back in February, March and April, separately from the February group. Maybe the January customers are loyal and keep buying, but the February ones vanish after one purchase. That difference tells you something important happened, perhaps a change in your delivery speed or your product quality, and it happened right around February.
A subscription business in London tracking cohort analysis might notice that everyone who signed up during a big discount month cancels within two months, while those who signed up at full price stay for a year. That single insight can save the business from chasing cheap sign ups that never turn into real revenue.
For someone in the diaspora sending money home through a fintech app, the company behind that app is almost certainly running cohort analysis on you. They want to know if people who joined in your cohort still use the app six months later, because that tells them whether their product actually works or whether people try it once and calmly drift off.
A Nigerian ed tech startup notices that of the 500 students who enrolled in January, 300 are still active after three months, but of the 500 who enrolled in March after a discount promo, only 90 are still active after three months. Same number of sign ups, very different story, and the discount cohort is the one leaking money.
Questions people ask
Is cohort analysis only for tech startups?
No. Any business that has repeat customers can use it, including a food delivery service, a school, a savings app, or a shop that sells on instalment. The point is grouping by start date and watching behaviour, not the size or type of business.
What is the difference between cohort analysis and just looking at monthly sales?
Monthly sales tell you the total money coming in, but they blend old loyal customers with brand new ones, so you cannot tell which group is driving growth or which group is falling off. Cohort analysis separates the groups so the real pattern shows itself.
Do I need special software to do cohort analysis?
Not at the start. A simple spreadsheet with sign up dates and purchase dates can reveal a lot. As the business grows, tools that track user behaviour automatically make the job faster, but the thinking matters more than the tool.
Last reviewed