Churn Cohort Analysis
Updated July 21, 2026
Grouping customers by attribute and tracking churn patterns over time to identify when and why attrition spikes.
Also known as: Cohort retention analysis, Retention cohort analysis, Vintage analysis, Cohort churn analysis, Cohort retention curve
Churn cohort analysis groups customers by a shared starting attribute, most often the month or quarter they signed up, though also plan tier, acquisition channel, or segment. It then tracks how each group retains or attrites over the periods that follow. Instead of reporting one blended churn number for a billing period, it decomposes that same set of cancellation events by starting group, producing a retention curve or a cohort-by-period table. The payoff is that it shows when customers leave relative to their own tenure, and whether customers acquired more recently are holding on better or worse than earlier ones.
A blended churn rate averages together customers who behave nothing alike, and that average can stay flat while the underlying picture rots. If newer signups are cancelling in their first month while a large, loyal base of older accounts props up the total, the aggregate looks stable right up until the loyal base thins out. Cohort analysis surfaces that divergence early by keeping each vintage separate and comparable.
The method is not specific to SaaS or to churn. The cohort study comes from epidemiology, where Wade Hampton Frost formalized it in 1935 to compare disease incidence across people grouped by birth period, and the word traces to the Roman military cohors. As subscription businesses matured through the early-to-mid 2010s, cohort analysis became a standard way to read retention, popularized in SaaS operator and investor writing and built into billing-analytics tools like ChartMogul, Baremetrics, and Stripe. Churn cohort analysis is simply that older technique pointed at attrition.
How a churn cohort table is built
The raw input is an event log: a row per customer with a user or account ID and dated events for signup, renewal, and cancellation. Customers are bucketed into cohorts by a starting attribute, most commonly the period in which they signed up. For each cohort you then compute, at every subsequent period, the share still active. That share is the retention rate, and its complement is the churn rate, which equals one minus retention.
Laid out as a grid, the result reads down the rows as cohorts and across the columns as periods since signup: month zero, month one, month two, and so on. Rendered as a heat map, the columns expose tenure-driven patterns: a sharp drop in the first column points to onboarding or activation failure, while a cluster at the twelve-month column points to renewal-date attrition. Reading down a single column instead compares the same lifecycle moment across vintages, which is how you tell whether recent cohorts are retaining better or worse than older ones.
Logo cohorts versus revenue cohorts
The same cohort framework can measure two different units, and they can tell opposite stories. A customer-count cohort tracks logo churn: what fraction of the accounts in a cohort are still active each period. A revenue cohort tracks dollar churn: what fraction of a cohort's starting MRR or ARR is retained, which moves with downgrades, expansions, and the size of the accounts that leave.
The gap between the two is diagnostic. A cohort can shed many small accounts while the larger ones stay and expand, so logo retention looks poor but revenue retention holds or even climbs above one hundred percent as expansion outweighs losses. The reverse also happens: a cohort keeps most of its logos but loses one anchor account and craters on a revenue basis. Reading both prevents a single view from flattering or alarming you about the wrong thing, which is why revenue-cohort analysis pairs naturally with net revenue retention as a durability check.
Churn cohort analysis versus general churn analysis
Cohort analysis is a structured, comparative technique: it groups customers and lines their curves up against each other over time to answer who churned and when, relative to other groups. General churn analysis is a broader umbrella that also asks why customers left, pulling in support tickets, exit surveys, downgrade reasons, and review sentiment, and does not necessarily group by cohort at all.
The two are complementary rather than competing. Cohort tables are strong at locating a problem in time: they will tell you that the March signups fell off a cliff at month three, or that every cohort since a pricing change retains worse than the ones before it. They are weak at explaining cause, because a grid of percentages carries no reason codes. The usual workflow is to let the cohort view flag the anomaly, then run qualitative churn analysis on the affected cohort to establish why. Retention analysis, for its part, is best treated as a subset of cohort analysis: the same grouping method, with retention as the metric in the cells.
Where competitive intelligence intersects churn cohorts
Cohort tables are an internal instrument, but the reasons a cohort deteriorates are frequently external. When a specific vintage starts churning faster than its predecessors, the question is what changed in its lifetime, and a competitor shipping a missing feature, cutting prices, or launching an aggressive displacement campaign is a common answer. Cohort timing narrows the search: knowing that decay began in a particular quarter tells you which competitor moves to line up against it.
This is where a competitive-intelligence workflow feeds the analysis. Monitoring competitor pricing pages, changelogs, and messaging gives you a dated record of external events to correlate against a cohort's inflection point, and win-loss and competitive-churn data can attribute departures to named rivals. The cohort grid says when and how much; competitive evidence helps explain whether the cause sits inside your product or across the market.
Stop looking terms up. Start tracking them.
meertrack watches your competitors' websites, pricing, and hiring, then alerts you when something meaningful changes.
Frequently Asked Questions
What is cohort analysis in SaaS?
It is a technique that splits customers into groups sharing a starting attribute, usually their signup period but also plan, channel, or segment, and follows a chosen metric for each group across later periods. In SaaS it is applied most often to retention and churn, so you can see how each vintage of customers behaves over its lifetime instead of collapsing everyone into one blended figure.
How do you calculate churn rate using cohort analysis?
Assign each customer to a cohort by their signup period, then for each cohort count how many remain active at every subsequent period. The retention rate for a cell is active customers divided by the cohort's original size; the churn rate is one minus that retention rate. Repeating this across all cohorts and periods yields a grid, often shown as a heat map, that exposes when attrition concentrates.
Why does a blended churn rate hide problems that cohort analysis reveals?
A blended rate averages customers with very different behavior into one number. A large base of loyal older accounts can mask newer cohorts that are cancelling quickly, keeping the aggregate flat until the loyal base finally thins. Cohort analysis keeps each vintage separate, so deteriorating retention in recent signups is visible immediately rather than after it has already spread through the whole book.
What is the difference between logo churn and revenue churn cohorts?
Logo cohorts track the share of accounts still active in each period; revenue cohorts track the share of the cohort's starting MRR or ARR still retained. They can diverge sharply: a cohort can lose many small accounts while its larger ones stay and expand, so logo retention looks weak but revenue retention is strong, or vice versa. Reading both avoids drawing the wrong conclusion from one.
What is vintage analysis and how does it relate to churn?
Vintage analysis is a near-synonym for signup-based cohort analysis, borrowed from finance, where customers are grouped by the period they were acquired, their vintage. Applied to churn, each vintage gets its own retention curve, and comparing curves across vintages shows whether the business is acquiring customers who retain better or worse over time. It is the same method as churn cohort analysis under a different name.
Related terms
Grouping competitors or events by shared characteristics and analyzing their trajectories in parallel.
Gross ChurnTotal revenue or customers lost without accounting for expansion or recovery.
Revenue Churn (MRR Churn)Percentage of recurring revenue lost from cancellations and downgrades.
Net Revenue Retention (NRR / NDR)Revenue from existing customers at period end divided by their starting revenue, after expansion, contraction, and churn. Above 100% = customers spend more over time.
Customer Churn Rate (Logo Churn)Percentage of customers who cancel or don't renew during a period, regardless of revenue impact.
Competitive ChurnCustomer attrition specifically caused by a switch to a competitor's product, as opposed to budget cuts or dissatisfaction.
Average Revenue Per User (ARPU)Total MRR divided by total paying customers. Tracks whether expansion is outpacing discounting.
Customer Acquisition Cost (CAC)Total cost of acquiring a new customer (marketing + sales spend / new customers).