Main image for article

The Cohort Analysis Framework for Improving NRR: A CFO's Guide

Your step-by-step guide to building insightful NRR cohort tables.

Deeply understanding your Net Revenue Retention (NRR) has never been more important. 

The days of “growth at all costs” for B2B SaaS businesses are gone, and investors are focused on finding companies that are efficient. In times like these, you absolutely have to be able to keep, and grow, your customer base over time. 

And, it’s not just enough to know what your NRR is. If you want to increase it (and you surely do) then you need to know why your NRR is what it is. That’s why the most data-driven and strategic finance leaders rely on cohort analysis when looking to understand their NRR.

In this post, we’ll explore exactly how you can look at your cohorted NRR to lead to the right strategic decisions for your business. 

The Cohort Analysis Framework: Step By Step

Step 1: Cohort NRR by the month the customer started with you

As you build your cohorts, start by cohorting based on when your customers started with you. Then, track the NRR of that cohort by month over time. If you sell annual deals, then you can expect most of your movement to happen at month 13. 

💡 For a refresher on understanding cohort tables, check out this post.

Your table will start out by looking something like this:

Why is this helpful?

This is helpful because you can:

  • Understand what is happening with your NRR and when it is happening.
  • See how recent cohorts are performing compared to older cohorts. If something has gone wrong, you can act fast. If something has gone right, you know right away and can double down. For example, you can understand the impact of pricing changes by tracking your NRR by month before and after the change. Is the increased revenue making up for any customer churn you might see?
  • Understand the impact of seasonality. Do customers that start in January always end up performing worse than those in April? If so, should you change the amount you invest in those months? (Answer: Probably.)
  • Understand how your retention changes over multiple years which can impact your LTV and how much you’re willing to spend to acquire a customer. For example, if your NRR is 105% at month 13, and 105% at month 25 then what you can invest is very different from if it’s 105% at month 13 and 145% at month 25. Your whole strategy might just change.

Step 2: Add the raw number of customers to your table

To make your cohort analysis even more helpful, add another row that lets you track how the raw number of customers is changing by month. Now you’ll be able to see both:

  • How revenue is growing or shrinking over time from a group of customers
  • How the number of customers changes over time (and when those changes tend to happen)

This will make your table look like this (the percentage represents the NRR, and the number represents the number of customers):

Why is this helpful?

This is helpful because you can:

  • Keep an eye on your NRR and customer retention at a glance.
  • If NRR takes a huge drop, now you’ll know if it was one BIG customer or a bunch of average-sized customers that made the drop.

Step 3: Segment by your metadata

Now, add another layer to your table to see how the data changes for different segments of your customers. For example, you can slice your cohorts by “Size of Company” to understand the difference between SMB and Enterprise customers. Or, you can slice your cohorts by “Geography” to understand NRR by region.

This will make your table look like this (in this particular example, "Large," "Medium," and "Small" represent the size of the customer):

💡 Note: If you’re doing this in Subscript, Subscript Insights will automatically tell you the segments that perform better or worse than others.  

Why is this helpful?

It takes some getting used to, but for every cohort, I now understand what happened to my NRR, when it happened, and why it happened. And, you’ll have your insights in minutes, all through one cohort table.

Of course, all of this is only possible if you’re using Subscript. Otherwise, you’ll be buried in spreadsheets for weeks trying to pull this off. 😄

Other considerations

Once you have your cohort tables built, there are a few other things to keep in mind that might impact your analysis. 

  • Parent/Child Relationships - If you have customers that roll up to a larger “Parent” customer then you need to understand how NRR changes if you consider just the parent company, or if you look at every company underneath the parent. For example, if Coca-Cola is a customer then you probably don’t care that much if a few small bottling plants churn, as long as the overall NRR of Coca-Cola as a whole is increasing. 
💡 Note: There’s a button you can push in Subscript to do this analysis instantly. Toggle back and forth with ease. 
  • Usage-based Revenue - If you have usage-based revenue you should understand the impact that it has on your NRR cohorts. When doing your cohort analysis, do it once where you include your usage-based revenue and once where you don’t include it. When you include usage-based revenue, you’re learning whether, on the whole, you’re making more or less money from a set of customers this year than you did last year. When you measure NRR without usage-based revenue, you’re finding out how your contracts are growing over time.
💡 Note: There’s a button you can push in Subscript to do this analysis instantly. Toggle back and forth with ease. 


As you’re guiding your business, it’s critical that you understand NRR at a very deep level. Sure, you need to track the number over time, but you also need to know why it’s growing or shrinking. Frankly, that’s the only way you can make the right decisions to get the number going up and to the right.

To do this, one of your best tools is going to be a cohort table. It gives you one place to look at that can provide nearly everything you need to know about your NRR.

Historically, cohort tables have been extremely difficult to build. But, with the help of Subscript, this sort of analysis only requires a few clicks. If you’re interested to learn more then we’d love to show you around.