The long term success of a software solution does not only mean getting people to register or create an account.

Success is also about getting repeat visits from users and increasing the total active users. To get value from your SaaS, you need to go beyond vanity metrics that only measure growth to metrics that really provide insight into your retention. One of the key tools to leverage in evaluating retention is cohort analysis.

 

Cohort Analysis: What is it?

A cohort is a subset of customers/users grouped by common characteristics. In the business analytics context, cohorts are segmented by acquisition dates. For example, we classify a user into a specific cohort depending on the first time this user visited your website or used your app.

Cohort analysis is the comparison of behavior and metrics of different subsets (cohorts) over time. You can identify the highest performing cohorts and the factors driving this success.

The cohort analysis report is an underrated feature on Google Analytics. Why is cohort analysis so critical? The metric allows you to isolate the impacts of various marketing activities on specific cohorts.

Running a cohort analysis is like running an experiment on your business. As a marketer, you can run a limited-time campaign with the parameters you wish to test. You can test ad content, target audience, marketing channel, landing page design, etc. You can then compare different marketing campaigns on the lines of engagement, reach, and conversion. With the results, you can appreciate the factors of a specific campaign that add value to your business. See Analytics Help for information on how you can configure cohort analysis on Google Analytics.

 

Uses for Cohort Analysis

Cohort analysis is an essential business analytics technique. The metric allows you to compare different changes and variables between your digital marketing campaigns. You can use the analysis to isolate the effect of certain website modifications on user behavior.

Below are factors that you can analyze with cohort analysis:

  • Ad content
  • Channels
  • Target audience
  • Experiment/campaigns
  • Website redesigns
  • Discount/sales/promotion campaigns
  • New service offerings and product lines

Although, in theory, you can analyze any of these parameters through cohort analysis, not all analytics tools (Google Analytics) allow you to analyze the impact of these factors on user behavior.

 

Limitations of Cohort Analysis in Google Analytics

Theoretically, cohort analysis is instrumental. However, the analysis report in Google Analytics has many limitations.

1. Google Analytics can Only Classify Cohorts Based on Acquisition Dates: In theory, you can run a cohort analysis on cohort classification based on any shared characteristics. Google Analytics limits you only to acquisition dates.

2. Google Analytics is Imprecise in Tracking Retention and Returning Users: Let's say John is a user on your website. John visits your website today. If John visits your website tomorrow, then Google Analytics should register John as a returning user.

However, Google Analytics will not be able to track his next visit as a returning session if John takes these actions:

  • Clears browser cookies
  • Visits the site on a different device or browser
  • Visits the website on incognito mode

The average digital consumer owns an average of four (3.64 to be precise) devices. About 36% of Americans own a tablet, a smartphone, and a personal computer. With these factors to consider, tracking users across devices, sessions, and browsers has become quite a challenge.

Cohort Analysis Usage

let us understand the usage of cohort analysis with an example:

Below is a representation of daily cohort users who have launched an app for the first time and revisited in the next ten days.

From the table, it is easy to notice a triangular formation. This right-angled triangle infers the following:

  • About 1358 users launched the app on Jan 26. The retention rate on day one was 31.1%,12.9% on day seven, and 11.3% on day nine. By day seven, one in eight users who launched the application on Jan 26 was still active on the app.
  • Out of all the users captured during this test (13,487 users), 27% are retained on day one, 14% by day five, etc.

There are two main benefits of reading the cohort table above:

Product Lifetime: Comparing different cohorts at similar stages of their lifecycle can depict product lifetime. From the comparison, you can see the percentage of users that are coming back to the app. The early lifetime months have an association with product quality and performance.

User Lifetime: It allows you to appreciate the long-term relationship with people in any cohort. You can use the data to ascertain how long people are coming back to the app and how strong or how valuable a specific cohort is.

Cohort analysis lets you see how certain parameters develop over the product lifetime as well as customer lifetime.

 

Cohort Analysis to Improve Retention

Cohort analysis involves the observation of groups of users over time. It is all about studying behavior changes.

From cohort analysis, you can track how customers behave over time or how similar behavior replicates between different cohorts. How do you break users into cohorts?

Below are some of the most popular cohorts:

Acquisition Cohorts

These are divisions based on the dates of acquisition. You may classify acquisition cohorts by day, week, month, or year. Acquisition cohorts can help you track customer retention.

Back to our earlier example, here is a table of acquisition cohorts.

From the table, we can chart a retention curve of these cohorts over time. The line graph below makes it easier to infer customers who are fleeing.

From the curve, you can notice that about 75% of users abandoned the app after the first day. After day one, users dop the app gradually to about 12% on day ten.

From the chart, you can tell that above 75% of users are not getting the core value of the app. There is a need for improvement in the onboarding department.

Behavioral Cohorts

These cohorts classify users based on the behaviors that they exhibit over time. Behavior could be several discrete actions that a user takes. These actions could be app installation, app launch, app uninstallation, etc.

Another example of behavior worth tracking is whether or not a user reads app reviews. Tracking such a parameter could answer questions like:

  • Do users who read reviews have higher conversion rates?
  • Do users who read reviews engage more with the app?

Let's test user behavior by comparing retention between the cohorts below:

You can see that both user segments had the intention of transacting on the app. One segment chose to proceed with checkout, and the other decided to cart abandonment.

From the retention tables above, we can conclude that cart abandonment issues did not re-engage with the app, not even one day after launching the app. So, we have less than 24 hours to re-target these users.

 

The Power of Cohort Analysis

From the raw data, you can develop a quantitative and systematic approach to understand how users fall in love with your application. You can use your approach to re-ignite the excitement and combat churn. Also, you can formulate strategies that will increase retention.

The power of cohort analysis is that it highlights which customers leave and when they leave. From the data, you can trace possible reasons for certain customer behavior. Cohort analysis is a key tool in tracking customer retention.



Author Information

Insivia is a Strategic Growth Consultancy helping software & technology companies scale through research, brand strategy, integrated marketing, web design, and retention.