Customer Level $ Retention

Another Way to Look at SaaS Cohorts

You can find a lot of good content on cohort analysis for SaaS companies (see a great template by Christoph here and a Q&A by me here for instance) and I would say most SaaS founders and investors are looking at cohorts in some kind of way.

These cohorts typically measure the number of customers signed up in a given month and what % of these stick around in the following lifetime months. This allows for easy benchmarking to other SaaS companies and helps in estimating the lifetime value and contrast it to CAC, the ultimate health check of the business model (or LTV/CAC ratio).

Recently, I started using a modified format, which I believe has some advantages to the standard template as on top of the above, it also answers some of the other common questions in a SaaS due diligence. This is an example of what Customer Level Dollar Retention could look like:

Note:

(1) This assumes the first customer was signed in January `16

(2) To make it fit, I cut off customers added in/after February `17

(3) For simplicity sake, I used round numbers and random segments

What can we learn from this data?

Since it is one level deeper, it gives us the basis for answers beyond the retention of customers (letters corresponding to the illustration above):

(A) The industry mix can be used to understand the current customer base and serves as a great data point for bottom-up TAM work (by extrapolating the early customer mix to the total addressable market); it also can show any concentration risks (by segments or customers). You can also expanded this with additional customer data like number of employees, location and or user accounts.

(B) Churned accounts are easy to identify.

(C) So are accounts that churned, but then came back for whatever reason.

(D) Up-sells are the best and in combination with B+C show net $ retention.

–– Plus: A mix of accounts (B)+(C)+(D) are great for reference calls ––

(E) Dividing total MRR by number of customers gives you ACV over time.

(F) On top of churned/downsold/new MRR, the number of new customers added over time shows momentum when looking at the monthly or quarterly number of net adds. As the company finds product/market fit and optimizes distribution, that number should increase. This can also help to identify seasonality if you`re looking into a longer time horizon.

Some caveats:

(i) Since you are digging one level deeper here, this is not as well suited for high N, low ACV SaaS businesses. It can get messy quickly when you are looking at 100s of customers – but for ca. 50–150 it works like a charm.

(ii) I would always supplement $ retention (or any kind really) with some kind of engagement data that shows that customers are really using the software and are expected to keep doing so. Engagement is the best predictor for retention.

(iii) The purpose of looking at cohorts this way should be to make the communication of this data easier for entrepreneurs, as well as easier to consume for investors. Overcomplicating it could lead to frustration and it can typically be a good starting point to look at whatever material is already prepared. That being said, I found most companies could easily export the above data from Salesforce and such and it definitely cut down the time my team is spending on back and forth communication on KPIs.

Thanks for reading!

I hope this is helpful and would appreciate any feedback and/or suggestions to do this better. :)

–– nico