What is Cohort Analysis and Why Should You Care?

cohort analysis from MyTheresa S1

Cohort Analysis is a framework you can use to better understand your business. Learn more about how to apply this framework to a consumer products business.

Cohort analysis comes from the tech startup world. Historically it’s been used to track the performance of consumer apps, software or subscription products and analyze user behavior.

But this framework can be applied to any business with customers, as long as you have the right data. So let’s zoom in on how to use cohort analysis for physical product businesses that sell direct to consumer.

What Is A Customer Cohort?

A cohort is simply any group of customers that share a common characteristic. It can be any characteristic–when they bought, what they bought, demographic information or behavioral data. If you can track the data reliably, you can build cohorts with it.

Building a cohort enables you to analyze the behavior of a subset of customers and compare their behavior to that of the average customer. It also enables you to track an audience’s behavior over time in comparison to your business as a whole.

What Is Relevant Cohort Analysis For A DTC Business?

Year or month acquired

Group customers by the date they were acquired.

Cohort Analysis from the MyTheresa S1 filing
Example of annual cohort analysis from the MyTheresa S1 filing

You can use this analysis to understand how much of today’s revenue was driven by customers acquired in a given year. That will help you determine if your retention rate is getting better or worse over time and how heavily you’re leaning on highly tenured customers for today’s sales.

This analysis will also help you understand if customers you acquire in certain months are more or less valuable than others. Hint–the holiday shopping season often produces some of the least loyal, least valuable customers.

New vs returning customers

Determine what percentage of sales over a set period of time were driven by returning customers vs newly acquired customers.

This analysis is critical in understanding why sales may not be growing as fast as you want them to. Sales from retained customers have a relatively fixed upper limit, whereas the only limit to customer acquisition is profitability.

Analyzing which products are being purchased by new vs returning customers can help you understand if new launches are achieving your strategic goals and if existing customers are adopting your latest wares.

Acquisition Channel

Group customers by the marketing channel that drove their first purchase.
This one is a little tricky because it gets into questions of marketing attribution. You can always use last click attribution data to make a first pass at this analysis.

The obvious use case here is determining which acquisition channels and campaigns produce the highest-value customers. You can also study the behavior of each channel cohort over time to develop lifetime value estimates by channel and use that to inform budgeting strategy.

Category or item of first purchase

Group customers by the category or item of their first purchase.

You can use this analysis to better understand the value of each category in your assortment. This is especially valuable if you’re running a multi-category business with a complex assortment. Your brand may be known for one type of product, but a relatively small or unknown part of the assortment could be bringing in your highest-value customers.

If you find yourself with a hot item on your hands, [link] this analysis will help you understand how likely all your new customers are to return to the brand for subsequent purchases.

Why Is Cohort Analysis Useful?

Cohort analysis helps you unlock additional layers of understanding of your direct to consumer business. As we know, your business is made up of channels, but your channels are made up of customers.

Let’s look at a fairly common scenario: you’re gearing up for a “tentpole” selling event and expecting sales to grow 20% from the same event last year. You reach the end of the event period and, unfortunately, sales were fairly flat despite an increased investment in performance marketing.

So what happened? Traditional analytics will only give you part of the picture: how much traffic visited the site, conversion rates, average order value, and channel-level digital marketing performance. All of that is useful, but sometimes it’s not as actionable as you’d like it to be.

Cohort analysis can reveal the rest of the story. What percent of sales during last year’s event was driven by new vs returning customers? What about this year? Maybe you missed your plan because fewer returning customers came back, or because you failed to drive enough acquisition.

What products drove the majority of sales during last year’s event, and how were they priced? How does that compare to the assortment and pricing this year? Maybe your event has become associated with a hero product you no longer stock, or didn’t stock in sufficient quantities.

This is just scratching the surface of the value of cohort analysis. Trends in conversion rate, traffic and AOV are all tied to customer behavior, and cohorts help you isolate that behavior.

Cohort analysis can also be used as a forecasting tool. Forecasting in consumer goods businesses–especially “fashion” businesses– is hard. But if you can predict the revenue coming in from returning customers with a low-ish margin of error it becomes a lot more straightforward.

How Do I Run Cohort Analysis?

Unfortunately there isn’t a free and easy tool for cohort analysis. You have two options–use a language like SQL or Python to run queries on your own data, or partner with a CRM or CDP vendor to create a graphic user interface for your database queries. If your business is relatively young and has less data, you may be able to get away with using Excel.

Something I’ve pondered for a bit–where is the Google Analytics of cohort analysis? I think the answer is that customer and product data are rarely structured consistently over time. The inputs are typically owned by two separate teams who are working to optimize their own tech stacks vs thinking about mashing up the two data types.

Kicking off a cohort analysis project will always require some data cleaning. You need to land on a definition of “one customer” that every stakeholder can agree on, and you need to ensure that the revenue in your analysis aligns with the “source of record” within your company.

If you don’t do this, it’s easy for people to poke holes in your findings if they don’t like them.

All the work is worth it though! Cohort analysis will help you uncover the strategic advantage within your business and enable you to develop strategies that go beyond “best practices”. And that’s what Modern Lifecycle Marketing is all about, isn’t it?