SaaS Cohort Retention Analysis and What It Does to Valuation

SaaS companies with net revenue retention above 100% grow 43.6% a year on average. Those below 60% grow 13.1%, per ChartMogul’s SaaS Retention Report. Buyers know that spread, which is why the retention number on your metrics slide is the one figure they refuse to take at face value.

What they build instead is a cohort retention analysis. A cohort is every customer who signed up in the same month, tracked together from that month forward. Instead of asking “what is our churn,” it asks “what happened to the customers who arrived in March, every month since.” Aggregate churn reports an average. Cohorts show what is happening underneath it, and the two regularly disagree.

By the end you should be able to build the three cohort views a buyer asks for, read your curves the way their analyst reads them, and know which pattern raises your price and which quietly caps it.

Why One Retention Number Cannot Carry a Valuation

Start with why one number fails. A 2026 compilation of private SaaS benchmarks puts median net revenue retention at roughly 105% in 2021, drifting to about 101% by 2024. That median averages together businesses that barely resemble each other, and the clearest divider is what you charge.

So when your deck says “101% NRR, right at the median,” a buyer has learned almost nothing. Is that a stable base quietly expanding, or new-logo growth papering over a leak? Both produce the same headline, and they do not earn the same multiple.

The Three Cohort Views a Buyer Builds From Your Data

The more serious the buyer, the less they trust your dashboard. Institutional acquirers rebuild retention from the billing ledger themselves.

View one: the retention curve. For each join-month cohort, what share of its original revenue remains after 3, 6, 12, and 24 months? The shape is the verdict. A curve that drops early then FLATTENS means the customers who stayed are staying: a durable core a buyer can underwrite. A curve that keeps sloping down never found its floor, and nobody pays a premium for a base that is still draining.

View two: cohort quality over time. Are this year’s cohorts retaining better or worse than last year’s? This is where growth hides decay, and it is the view aggregate churn is structurally incapable of showing.

The mechanism is simple once you see it. Blended churn is weighted toward your oldest, most loyal customers, because they are the largest share of the base. A company whose January cohort holds 88% at month six, April holds 84%, and July holds 79% is degrading fast, and can still report a blended rate that barely moves, because those three cohorts are a minority of total revenue. The number is not lying. It is measuring the past.

Buyers therefore read cohort-over-cohort quality as a verdict on go-to-market, not product. Retention that degrades as volume rises usually means the newest customers fit worse than the early ones: a channel that converts poorly, a discount cohort that never adopted, a segment entered too early. A buyer paying for growth is buying the newest cohorts, not the average.

View three: where the expansion lives. Net retention above 100% means expansion outran churn. But within the cohorts, who expanded? If a few accounts carry it, your net retention is a concentration bet wearing a retention costume. It is the cohort-level version of the two-number test in our gross versus net retention guide.

One Cohort, Followed Home

Here is an illustrative January cohort starting at $100,000 of monthly recurring revenue. Gross retention counts only revenue that stayed. Net adds back what survivors spent on upgrades.

Read it like a buyer. Gross falls to 82% by month 12, so nearly a fifth of the cohort’s original revenue walked out. Net shows 104%, because expansion from survivors more than covered the hole. On a metrics slide, this is a 104% net retention story.

Then comes the question that sets the price: how many accounts produced that expansion? If the answer is two, this is a key-account story and gets priced with a concentration discount. Same table, opposite conclusion, decided by a column most sellers never show.

The Expansion Concentration Test

Here is the arithmetic a buyer runs on that 104%, and you can run it in one line. Take the cohort’s expansion dollars, then ask what share came from the top three accounts.

Above, survivors held $82,000 and the cohort finished at $104,000. Expansion was $22,000 on an $82,000 base, or 27%. Strip it out and this is an 82% gross retention story, which is unremarkable.

Now split that $22,000. If the top three accounts contributed $6,000, expansion is broad and the 104% is real. If they contributed $19,000, then 86% of your expansion sits in three logos and the buyer is underwriting three renewal conversations, not a retention curve.

The other half of the read is shape over time. What matters is not the percentage at any given month but whether the line is still falling. A curve still sloping down at two years has not found its floor, and a buyer cannot price where it stops.

The consequences for valuation are direct. A flattening curve with broad expansion defends your multiple and your ARR quality story. A curve that never flattens pushes buyers toward earn-outs, since they pay for revenue only once it proves it stays. Degrading cohorts trigger a growth-quality discount. None of this requires distrust: it is what churn benchmarks and a headline net retention figure structurally cannot show, made visible.

Package the Analysis Before Diligence Does It To You

All of this gets run on your business by someone. The only choice is who goes first.

Build the three views yourself, from the same billing export a buyer will request, not from a BI dashboard they cannot audit. Then annotate every dip. A pricing migration, a sunset segment, a discount experiment that ended: each explains a bad cohort honestly. An explained dip reads as competence. An unexplained one reads as decay, and the buyer assumes the least generous version.

Reconcile to one source. Your cohort file, headline retention numbers, and financial model must derive from the same ledger. The fastest way to lose a retention premium is three documents with three answers.

A clean cohort package makes a claim that is hard to fake: this base is inspectable, and I already know what it says. That alone moves price.

Frequently Asked Questions

Next Steps