Comparing groups of customers and analyzing the ways they behave over time helps SaaS entrepreneurs identify patterns and information that expose problems early and grow the business more efficiently.
Do customers you acquired last month act differently than ones you signed up the month prior? Do users who responded to a discount or promotion behave differently than those who purchased at full price?
A cohort analysis can provide answers to critical questions that guide effective SaaS growth strategies.
What is Cohort Analysis?
Cohort analysis is a type of behavioral analytics, which is done by breaking down customers into related groups in order to gain a better understanding of their behaviors. It’s an informative practice every business owner should have in their toolbox.
The following is a run-down on how cohort analysis works and why it’s a useful strategy for gaining insight.
Defining cohorts
In cohort analysis, a cohort is the group of customers being analyzed. More specifically, a cohort is a group of people who have something in common during a specific time period. The parameters of this group are generally identified based on the question you want to answer and the metrics you determine to be significant.
For the purposes of a cohort analysis for your business, the groupings are usually made up of users who performed certain actions during a chosen time frame, such as downloading your app during a particular month or finding your product via social media in a given week.
The time-boundedness is key: Customers grouped by behavior but without a time parameter are called segments, not cohorts.
How is cohort analysis used in SaaS?
This type of analysis is valuable due to the specificity of the information it provides. It allows SaaS startups to find answers to specific questions about customer behavior by analyzing only the relevant data.
When you can see the patterns in how different demographics use — or don’t use — your product, you can:
Determine when and how to best communicate
Identify flaws in your messaging or promotions
Find out who is best served by your product
Design incentives to keep customers engaged when they’re most likely to stop using your product
A good place to start is looking at cohorts grouped by time, segment, and size.
Using time-based cohorts
Time-based cohorts include customers who signed up for various products or services during a certain time frame. Looking at time-based cohorts can reveal telling patterns in customer behavior. If 75% of those who signed up after your Q1 promotion remain customers in Q3, but only 50% of those who signed up upon receiving your Q2 email are still with you in Q4, it may be a clue that you over-promised in your Q2 promotions.
Time-based cohorts can be particularly useful for analyzing churn. For SaaS businesses, churn will likely peak at the start of a time-cohort’s time frame, but will eventually stabilize. More customers might bounce off your product early, but hopefully the ones who stick with it for longer learn to love it and churn at lower rates. Without time-based cohorts you may not see the pattern, and that can lead to erroneous conclusions about customer retention.
Using segment-based cohorts
Segment-based cohorts break customer groups down by pricing tier or the type of product they purchased. If your SaaS company offers multiple pricing tiers or products, analyzing segment-based cohorts can help you determine which pricing tiers serve the target audience's needs best.
For example, if customers in your most basic pricing tier churn at a much lower rate than your most expensive, that may be a sign that users are either getting everything they need from the basic version or feel that the premium version is doesn't deliver enough value.
Using size-based cohorts
If you’re a typical early-stage SaaS company on the way to serious traction, you likely have a good mix of clients — small businesses, scaled companies, and enterprises. Comparing these groups can show you where your stickiest revenue streams come from, and potentially illuminate issues with your product.
Generally, small businesses and startups churn at a much higher rate than enterprise businesses. Knowing this, you may want to “asterisk out” the smallest, most churn-heavy cohort to get a better picture of their average churn. Try analyzing MRR across these cohorts, too — does the lion’s share of your MRR come from a high-churn cohort? If so, you may need to figure out how to make your product stickier for that group.
It can also be very helpful to compare CSAT scores and Net Promoter Score (NPS) from different sized customers. Churning small businesses but maintaining a NPS of 50 in that cohort means something very different than churning small businesses and having a NPS close to 0.
How to Do A Cohort Analysis
How you go about performing cohort analysis depends on what question you’re trying to answer. You’ll need to select the following information from whichever data-management solution you use:
The characteristics of your cohort (what defines the group)
An inclusion metric (the action that precipitated inclusion in the group)
A return metric (the thing you want to know about them)
Example 1
Let’s say you are a mobile game developer, and you want to determine if users on iOS devices have been more or less profitable than users on Android devices over the last quarter. Since equal resources have been used to promote the app on both platforms up to this point, you decide to measure how valuable users are on each platform by comparing the average revenue per user (ARPU) between users on iOS devices and Android devices.
In this case, the characteristics of the cohorts are defined by the mobile operating system each user has (iOS or Android). The inclusion metric for both would be being an active user over the last quarter. And the return metric for both would be ARPU.
Your inclusion metric tells you that the iOS cohort has 400,000 users and the Android cohort has 500,000 users. Over the last quarter, the iOS cohort has had 200,000 active users while the Android cohort has had 250,000. The return metric indicates the iOS cohort has an ARPU of $3 while the Android cohort has an ARPU of $2.
From this cohort analysis, you might conclude that iOS users are less likely to download the game but slightly more profitable on a per user basis than Android users; and therefore, you may choose to appropriate a greater portion of the company’s marketing budget towards promoting the iOS version of the app for the upcoming quarter.
Example 2
You have a cloud-based time tracking app. Let’s say it’s December and you want to compare the retention rates of the customers you acquire from two distinct marketing campaigns:
Customers who signed up from a Mailchimp drip email campaign in April; and,
Customers who signed up from a Google Adwords campaign in May.
The characteristics of your cohorts are defined by the marketing campaign attributed to the new customer (email or Adwords). The inclusion metric for both is signing up. And the return metric for both is the customer's status (current or lapsed) in December.
Say the inclusion metric tells you that the email cohort has 200 customers while the Adwords cohort has 300. The return metric shows that the email cohort has 100 remaining current customers come December, while the Adwords cohort has 250. The retention rates are 50% for those who signed up in from the email campaign, and 83% for those who signed up from the Adwords campaign.
From this cohort analysis, you can conclude that the retention rates for customers who signed up from the Adwords campaign are significantly higher than those who signed up from email marketing. Therefore, you might choose to focus future marketing campaigns on Adwords or even test some other combination of search engine marketing (SEM) and display marketing strategies for future analysis.
Dive deeper...
With the new information you have from your cohort analysis, you can now cross-reference it with other data to try to figure out why the difference between the groups is so large. You may want to run the same analysis on February, March, and June sign-up cohorts, for example, or to look at how the customers in these cohorts converted into customers.
Applying Cohort Analysis Insights to Grow Your SaaS
Cohort analysis can provide all manner of useful insight into what works best to engage, convert, and retain customers. It’s something savvy SaaS entrepreneurs return to frequently to answer both basic and complex questions about the company’s progress in order to grow smarter.
Here are 5 ways you can apply cohort analysis insights to effectively grow a healthy, sustainable SaaS business:
Understand how user behaviors affect your business. Cohort analysis allows you to see how actions by those in the cohort translate to changes in business metrics, such as acquisition and retention.
Manage customer churn. You can marshal your data to assess your hypotheses regarding whether one customer action or attribute leads to another, such as whether sign-ups related to specific promotions reduce churn.
Calculate customer lifetime value. Analyzing cohorts based on acquisition time period, such as grouping customers by the month they signed up, allows you to see how much customers are worth to the company over time. You can then further group these cohorts by time, segment, and size to assess which acquisition channels lead to the best customer lifetime value (LTV).
Optimize your conversion funnel. Comparing customers who engaged in various ways with your sales process at given times can show how user experience throughout your marketing funnel translates to value in your customers.
Improve customer engagement. As you see patterns in how various cohorts engage with your company, SaaS website, and product you can take steps that will encourage all customers to take various actions more efficiently.
How does your startup stack up to other SaaS businesses?
Enter your own data to compare your startup's performance to the key metrics in our SaaS Benchmarks Report.
Our one-of-a-kind SaaS Benchmarks Calculator makes it easy to evaluate your results alongside peers in your industry.