How to Forecast Churn by Customer Cohort in SaaS

Your churn rate can look fine and still wreck next year’s revenue.

That’s the problem with blended churn. One average number can hide a strong enterprise cohort, a weak self-serve cohort, and a new pricing tier that’s falling apart after month two. For SaaS founders and finance leaders, that matters because churn shapes runway, hiring, fundraising, and how hard you can push for growth.

Forecasting churn by cohort is simply this: group customers together, watch how they behave over time, then use that pattern to estimate future losses and revenue risk. It sounds technical. It doesn’t need to be.

What customer cohorts tell you that overall churn cannot

Overall churn is tidy. Boards like tidy numbers. The trouble is that tidy numbers often hide the story.

A cohort view groups customers who started in the same period, or share another common trait, then tracks what happens next. That lets you compare one group against another instead of treating the whole customer base as if it behaves the same way.

Blended churn gives you the headline. Cohorts show you where the damage starts.

Why cohorts make churn patterns easier to spot

Once you line cohorts up month by month, patterns jump out. You can see whether customers tend to leave after onboarding, at trial end, after the first invoice, or at annual renewal.

That’s useful because churn often clusters around moments, not averages. A pricing change in March might only affect the April and May cohorts. A weak onboarding process might hit Month 1 retention hard, then flatten out later. A product issue might show up first in low-usage customers, not across the whole book.

Cohorts also help you compare retention, churn, expansion, and contraction in one view. You stop asking, “What is churn?” and start asking, “Which customers are leaving, when, and why?”

Which cohort splits are most useful for SaaS teams

Start simple. For most SaaS businesses, sign-up month or first paid month is the best first cut. Monthly cohorts usually give the right balance between detail and readability. If your sales cycle is long and contracts are annual, quarterly cohorts may make more sense.

After that, the most useful splits tend to be plan type, acquisition channel, customer size, or product behaviour. For example, you might find that outbound mid-market customers retain well, but paid social SMB customers fall away after Month 2.

Don’t overdo it. If a cohort has eight customers in it, the pattern is probably noise. Good cohort forecasting needs segments that are simple, stable, and large enough to trust.

The key numbers to track before you forecast churn

Before you forecast anything, get the inputs right. Recent 2025 to 2026 benchmark studies across more than 900 B2B SaaS firms put monthly logo churn at roughly 3% to 5% for SMB, 1.5% to 3% for mid-market, and 1% to 2% for enterprise. That’s useful context. It isn’t a substitute for your own data.

Retention rate and churn rate

Retention tells you what share of a cohort is still active after each month or quarter. Churn tells you what share you lost in that period. Those two numbers sit at the centre of the whole exercise.

In a cohort table, retention is often easier to read because it shows the surviving share over time. Churn is still important because it tells you where the drop happens. A cohort that goes from 100% to 90% in Month 1, then to 87% in Month 2, has a different problem from one that stays flat until renewal and then drops hard.

MRR movement, expansion, and contraction

Customer count only tells part of the story. Revenue can move even when logos stay steady.

That is why you need to track starting MRR, expansion, contraction, churned MRR, and, where relevant, reactivation. A cohort may lose a few small customers but still grow in value if the remaining accounts upgrade. The reverse happens too. You can keep the customers and still lose revenue through downgrades.

If you’re not already tracking this properly, a key SaaS metrics dashboard helps keep customer churn and revenue churn separate, which is where many SaaS forecasts go wrong.

Net revenue retention and why it changes the picture

NRR shows what happens to cohort revenue after churn, downgrades, and expansion. It answers a simple question: does this cohort shrink or grow after it lands?

That matters because a cohort with modest logo churn can still be healthy if expansion is strong. In many SaaS businesses, newer cohorts with NRR above 100% are in much better shape than raw churn alone suggests. Once you see that, your forecast stops being a customer count exercise and starts becoming a revenue planning tool.

How to build a simple cohort churn forecast step by step

This is where most teams overcomplicate things. Don’t.

Start with one clean cohort definition

Pick one cohort rule first and stick to it. For most SaaS businesses, use first paid month. That avoids muddying the picture with free users who never became real customers.

Be clear on your definitions. What counts as churned? Is a failed payment churn straight away, or only after a recovery window? Is a paused subscription active or lost? If those rules move around, your forecast will wobble for the wrong reasons.

Map retention over time with a cohort table

Build a table with cohorts down the left and months across the top. Then fill each row with retained customers or retained revenue.

A simple example looks like this:

You can read that table in seconds. March is tracking below January and February early on, so unless something has changed for the better, Month 6 is unlikely to land at 70% plus. That’s the whole point. The pattern becomes visible.

Turn historical cohort patterns into a forward view

Use completed cohorts as the base for your forecast. If the last six monthly cohorts behave in a similar way, take the average or median retention curve and apply it to newer cohorts that haven’t matured yet. Median is often safer if one cohort had a freak result.

Then adjust for known changes. Maybe onboarding improved in February. Maybe a pricing rise hit lower-value customers in April. Maybe customer quality dropped when you pushed a new acquisition channel too hard. Put those assumptions in the model, but only when you can explain them.

If you want this linked properly to revenue, headcount, and cash, a cohort-based financial model gives the forecast somewhere useful to live.

Check the forecast against reality and refine it

A churn forecast is not a one-off exercise. Compare forecast versus actual every month or quarter.

If you predicted 85% retention at Month 3 and landed at 79%, find out why. Was it poor onboarding, bad-fit customers, failed payments, or a product issue? Then update the assumptions. Good forecasts get better because the business learns, not because the spreadsheet gets prettier.

How to make your forecast more accurate and useful

A decent forecast uses more than subscription dates.

Use product, billing, and support data together

Churn rarely has one cause. Low usage, missed onboarding milestones, failed payments, repeated support tickets, and long time-to-value all point to different risks.

Billing data tells you about involuntary churn, which still accounts for a meaningful slice of losses in many SaaS businesses. Product data shows adoption. Support data shows friction. Put them together and the cohort story gets sharper. A customer who hasn’t used a core feature for 30 days and has two open tickets is not the same risk as one who logs in daily and pays on time.

Watch for churn triggers and risk windows

Most churn doesn’t happen at random. It tends to bunch up around certain windows.

The common ones are early onboarding, trial end, first renewal, annual renewal, pricing changes, failed payments, and unresolved support issues. If you know where the risk sits, you can forecast earlier and act sooner. That means your forecast stops being a post-mortem and starts helping the team.

Avoid over-segmenting small groups

This is where smart teams trip themselves up. They create dozens of tiny cohort views, each with a different story, and end up trusting none of them.

Keep the first version broad. Prove the pattern. Then add detail where the data is stable enough to support it.

If the cohort is too small, you’re reading luck, not behaviour.

How to use the forecast to make better decisions

This is why the work matters. A cohort churn forecast is not for admiring in a dashboard.

Spot where to focus retention efforts first

The forecast tells you which cohorts are most at risk and when the drop is likely to happen. That helps you direct effort where it will matter most.

If churn spikes in the first 30 days, fix onboarding. If annual contracts fall away at renewal, tighten customer success coverage and product education before that date arrives. If a certain acquisition channel keeps producing weak cohorts, stop feeding it and rethink your spend.

Use cohort forecasts in board and investor conversations

Boards and investors don’t want a vague line about “improving retention”. They want to see whether newer cohorts are better than older ones, whether expansion is offsetting churn, and what that means for forward revenue.

A clear cohort view makes your numbers more credible because it shows cause and effect. It also makes cohort revenue forecasting easier to defend in diligence, where weak definitions get exposed fast.

Link churn forecasts to runway and growth plans

Every churn assumption flows into revenue, cash, and hiring. If you lose customers faster than planned, growth slows, collections soften, and runway shortens. If NRR improves, the same sales engine can carry much further.

That is why churn should sit inside an investor-grade 3-statement forecast, not in a side tab no one trusts. A one-point improvement in monthly churn can change hiring pace, fundraise timing, and how hard you can back growth. A one-point miss can do the same in the wrong direction.

Conclusion

Cohort-based churn forecasting is about seeing the story behind the average. Once you know which groups stay, expand, shrink, or leave, your revenue planning gets sharper and your risks show up earlier.

Start with one clean cohort definition, one retention table, and one regular review cycle. Over time, the forecast will get better, and so will the decisions built on it.

The smartest SaaS teams don’t guess at churn. They track cohorts, learn the pattern, and plan from the truth.

Kishen Patel, BFP ACA Founder, Consult EFC · ICAEW Chartered Accountant · Fractional CFO

Over 12 years across Big Four audit, investment banking and corporate advisory. Kishen works with UK SaaS and AI companies on financial strategy, fundraising and board-level CFO support. ICAEW regulated. Big Four trained. Based in London.