What Cohort Analysis Is and How It Can Help Your Business

In 2022, customer retention is one of the focus areas for marketers, as it’s five to six times more expensive to acquire new customers than to retain existing ones.
To make sure customers want to continue using your product as long as possible, you need to get a better understanding of why they chose you, how often they use your product, and what causes them to leave.
You could use churn analysis, but it gives you answers after customers have already left. Another way to get a clearer picture of customer behavior is by using customer cohort analysis.
Customer cohort analysis evaluates the effectiveness of your marketing efforts and helps you make more detailed forecasts based on the collected data.
Having a deep understanding of your customers helps you choose the most effective marketing strategies. To analyze your customers, you can perform customer segmentation (by gender, age, purchase activity, etc.).
Keep in mind the following: segmentation by one characteristic isn’t indicative of customer behavior. For example, looking at people from one gender group doesn’t take into consideration the way they communicate with a brand. There may be new and loyal customers within one gender group and their behavior is completely different.
To track customer behavior across all touchpoints, marketers usually resort to customer cohort analysis. These are the results you can achieve:
A cohort is a group of users that share a common characteristic within a certain period of time. These can be users who signed up for a service at the same time, i.e. they performed the same action within a certain time span. The time factor is what makes cohorts different from segments.
Customer cohort analysis is one of the most popular practices in marketing and product analytics. It tracks the changes in KPIs over time within each cohort. Knowing when users exhibited certain behaviors lets you see how close you are to achieving your goals (like the values of Conversion rate, ROI, CAC, LTV, Retention Rate) and find what helps you achieve them. For instance, you can notice that the conversion rate for four-week users is higher than that for users of one week. If you performed user onboarding on week three, this means it helped you convert users into customers.
Before you start performing the customer cohort analysis, you need to define:
Absolute values can be less indicative than relative ones. For example, instead of analyzing the absolute number of customers, you should calculate customer retention rate.
You need to choose the parameters based on the needs and peculiarities of your business. The cohort type and size depend on the key metric you choose.
For a more detailed customer cohort analysis, you can create cohorts based on several characteristics at once. You can group people who subscribed to the newsletter and made their first purchase within one month. You can choose any traits you find necessary (traffic source/gender/age/country/etc.).
Many companies prefer to look at generalized statistics to measure the effectiveness of email marketing: open rate, click-through rate, unsubscribe rate for every month or for a particular email. Still, this information doesn’t tell you who interacts with your emails and how. One of the ways to learn this is customer segmentation. It is good but not good enough — there can be ‘new’ and ‘old’ subscribers in one segment. Their behavior and expectations can vary greatly. To get a more detailed understanding of your customers, you need customer cohort analysis.
Here’s how you can use cohort analysis to improve your email marketing strategy:
You can add more segments to cohort analysis to get a more in-depth insight into your customer base.
In this article, we won’t go into this method — we will focus on the basics of customer cohort analysis for you to get the gist of it instead.
In this case, the goal of customer cohort analysis is to see how subscribers interact with emails based on the Click-Through Rate.
The analysis will be based on the following parameters:
The parameter values will determine our future steps.
We gather the following data about subscribers into Table 1:
Table 2 contains information about interactions (one line for one action):
The first table already has the number of unique clicks. If your source data includes total values, you should remove duplicate lines from the set before starting. They can appear when a user clicks a link twice, for example. Since some platforms share all the data which they have, this results in duplicate lines — the only thing that is different is the event time.
Step 1. Group the users by subscription date, taking cohort size into consideration. This way we’ll know the number of subscribers in each cohort.
Step 2. Combine tables 2 and 1 using a common key field (subscriber ID).
Step 3. Sort the combined table you got at step 2 by the fields “date of action” and “subscription date” and calculate the total number of clicks.
Step 4. Calculate the key metric values for each pair.
The results are shown in a matrix table where rows contain cohorts, columns represent months, and values represent the key metrics (CTR).
For example, users that subscribed in February 2018 showed a CTR of 2,11%. In March 2018, they were more active — 14,20%.
What can we learn from this matrix table?
1. Cohorts “January 2018”, March “2018”, and “April 2019” were critically inactive. Looking at the number of subscribers in these cohorts makes the picture clearer.
In January and March 2018, there were a lot of new subscribers, which can be explained by successful marketing campaigns. The average metrics month by month look pretty convincing — the campaigns did their thing resulting in increased CTR.
Customer cohort analysis gives you a clearer view of what’s going on. The February 2018 cohort influenced the increase in CTR. Notice that most of the subscribers who signed up in January and March 2018 are irrelevant and don’t fall under the target audience. The April 2019 cohort doesn’t really stand out. We need more data to define the reason behind such low results: find out where the customers came from, whether there were any tests running, whether the marketing strategy has changed, and so on.
2. People who signed up at the beginning of 2019 are more active than subscribers from 2018.
3. In October 2018, CTR was drastically low. Even the cohorts that showed higher loyalty levels, engaged with the emails poorly.
There can be several reasons for that. The company might have changed their content (they tried a format that didn’t work, so they came back to the previous one). Maybe there was a problem with links (they weren’t clickable). Setting the cohort size at one day or week, you can track the effectiveness of email campaigns and anything else.
There is one more way to do cohort analysis with the help of a matrix table: you can use the month following the month when users signed up.
This way it is easy to compare cohorts based on their behavior and thus to calculate customer lifetime value (LTV).
Cohort analysis is a versatile method for data analysis that allows you to work with different types of data.
You can use cohorts to see how ads in a particular channel influence customer buying behavior. Without cohort analysis, you might think that a drop in ROI a month later indicates that the channel doesn’t work. However, if you create cohorts for users that engaged with the ads in January, you can see that they did make a purchase later (e.g., in March) — they just took more time to think the purchase over.
To predict conversion rate for a new design of a website banner, one part of users will be shown a new design, while the other — the old one. Then, a few weeks later, we will need to compare the results for both cohorts.
Use a cohort based on the number of new website visitors for a certain time period to compare the metrics of different marketing channels. This way, you’ll be able to track their engagement. You can motivate the loyal customers and target them with more sales. As for inactive customers, you can stimulate their interest with special offers.
Customer cohort analysis helps you identify user drop-off after releasing each mobile app version.
It is difficult to calculate the average LTV (lifetime value) while a customer is still active. The good news is you can calculate LTV for a cohort within a time frame (a month or half a year, for example) to forecast how much the cohort will be worth in the long run.
You can use tools like Excel and Google Sheets for customer cohort analysis. Still, these free tools have some downsides: the amount of source data is limited, it’s impossible to recalculate everything once the parameters change, there’s no auto-update, etc.
Some marketing and analytics platforms already have everything for customer cohort analysis. The user just has to choose the parameters to get a cohort report at once.
Still, this method isn’t perfect:
Here are some of platforms that offer cohort reports:
Business analytics tools (like Power BI) are optimal for building customer cohort analysis. Here’s what you can do with their help:
Customer cohort analysis can help you:
If you struggle with customer cohort analysis or any sort of business analytics, contact our professional analytics team. We strive to optimize business processes and help you achieve your best results powered by data.