TL;DR: Cohort analysis is how you see whether your SaaS business is actually improving over time or just growing by throwing more money at acquisition. By grouping customers by signup date and tracking their behavior over months, you can spot problems like degrading retention, identify which acquisition channels bring better customers, and forecast revenue with actual data instead of wishful thinking. Most companies track aggregate metrics that hide critical trends. Cohort analysis reveals the truth.
Every SaaS company tracks overall churn rate. Let’s say yours is 3% monthly. Seems stable. Your board is happy. You’re planning growth based on that number holding steady.
Here’s what aggregate churn hides: Your January cohort is churning at 2%, your March cohort at 4%, your June cohort at 5%. Your overall churn rate looks stable because you’re averaging together cohorts with wildly different behavior. By the time the aggregate number moves enough to notice, you’ve got a serious problem that’s been building for months.
We see this constantly with SaaS companies that can’t understand why their growth is slowing despite hitting new customer targets. The answer is usually in cohort data they’re not tracking. Newer customers are retaining worse than older ones, which means unit economics are degrading even as top-line growth looks fine.
Cohort analysis forces you to look at groups of customers who signed up together and compare their behavior over time. This reveals whether you’re actually building a better business or just masking problems with growth.
Start with the simplest version: a table showing retention by signup month. Each row is a cohort (customers who signed up in a specific month), each column is months since signup.
Here’s what it looks like:
Cohort | Month 0 | Month 1 | Month 3 | Month 6 | Month 12
Jan 2024 | 100% | 94% | 89% | 85% | 81%
Feb 2024 | 100% | 95% | 91% | 87% | –
Mar 2024 | 100% | 93% | 88% | 84% | –
Apr 2024 | 100% | 92% | 86% | – | –
Read this table horizontally to see individual cohort behavior. The January cohort retained 94% of customers after one month, 89% after three months. Read it vertically to compare cohorts at the same point in their lifecycle. Did the April cohort retain better or worse than January at month 1?
If newer cohorts show worse retention than older ones at equivalent points, you have a problem. Either product-market fit is degrading, you’re acquiring worse-fit customers, or something in your onboarding broke.
If newer cohorts retain better, that’s validation that you’re improving. Maybe you fixed bugs, improved onboarding, or started targeting better customers.
Build this table monthly. Track logo retention (percentage of customers who remain) and revenue retention (percentage of revenue that remains after accounting for churn and expansion). These tell different stories. You might lose 15% of customers but retain 105% of revenue because the customers who stay are expanding.
The Retention Curve That Predicts Your Future
Every SaaS product has a retention curve shape. Most show high early churn in months 1-3 as poor-fit customers self-select out, then flattening as you reach customers who are getting real value.
A healthy B2B SaaS retention curve might look like: 95% month 1, 91% month 3, 88% month 6, 85% month 12, 83% month 24. The curve flattens because customers who make it past month 6 are sticky.
An unhealthy curve never flattens. You lose 5% every month indefinitely. That signals fundamental product-market fit issues. If customers keep churning at constant rates forever, you don’t have a retention problem, you have a value problem.
The shape of your curve predicts your business trajectory. A company with flattening curves can afford longer CAC payback periods because customers who survive onboarding stick around for years. A company with linear decay curves needs sub-12-month payback because customer lifetime is short.
Plot your cohort data as curves, not tables. The visual pattern reveals things that tables hide. We worked with a SaaS company whose tables looked fine but whose curves showed a concerning pattern: every cohort had a secondary churn spike at month 8. Turned out their annual contracts came up for renewal at month 8 (they gave everyone 4 months free to start), and they’d built no renewal motion. We fixed that, and month 8 retention improved from 78% to 92%.
Logo retention tells you how many customers stay. Revenue retention tells you how much money stays. For most SaaS businesses, revenue retention matters infinitely more.
Build revenue cohort tables showing MRR by cohort over time:
Cohort | Month 0 | Month 1 | Month 3 | Month 6 | Month 12
Jan 2024 | $50K | $49K | $51K | $54K | $59K
Feb 2024 | $52K | $51K | $53K | $57K | –
Mar 2024 | $48K | $46K | $47K | $51K | –
The January cohort started with $50K MRR, dropped to $49K in month 1 (some churn), recovered to $51K by month 3 (expansion offset more churn), and is now at $59K in month 12. That’s 118% net revenue retention for that cohort.
This reveals expansion patterns that logo retention hides completely. You might be losing 10% of customers but growing revenue by 20% because the customers who stay are expanding aggressively. Or you might retain 95% of customers but see revenue decay to 85% because customers are downgrading.
Calculate net revenue retention for each cohort: ending MRR divided by starting MRR. Track this as a separate metric. A company with 90% logo retention but 115% net revenue retention has a very different business than one with 90% logo retention and 90% net revenue retention.
Enterprise SaaS companies should track revenue cohorts more carefully than logo cohorts. Losing one $100K customer matters more than retaining ten $1K customers.
Don’t stop at simple time-based cohorts. Segment by anything that might affect behavior: acquisition channel, customer size, industry, product tier, geography.
Build separate cohort tables for customers acquired through different channels:
Organic Search Cohorts: Month 1 retention 96%, Month 12 retention 89%
Paid Search Cohorts: Month 1 retention 91%, Month 12 retention 78%
Direct Sales Cohorts: Month 1 retention 98%, Month 12 retention 94%
This tells you which channels bring customers who stick around. Organic and direct sales customers retain well. Paid search customers churn faster. Now you know where to invest acquisition dollars.
Segment by customer size:
Under $100 MRR: Month 12 retention 65%
$100-500 MRR: Month 12 retention 83%
$500+ MRR: Month 12 retention 91%
Small customers churn much faster than large ones. This informs pricing strategy, what features to build, and where sales resources should focus.
Segment by everything you track. Some segments will show surprisingly different behavior. We found a client whose customers acquired during Q4 retained 15% better than Q2/Q3 acquisitions. Turned out Q4 customers were coming in through end-of-year budget spend and were more committed buyers. They adjusted their sales strategy to front-load effort in Q4.
Cohort data lets you forecast revenue bottom-up instead of top-down. Instead of saying “we’ll grow 30% this year,” you model exactly what happens to each cohort.
Take your retention curves, apply them to existing cohorts to project their future revenue. Take your expected new customer acquisition, apply historical retention curves to project their contribution. Add it all up.
This produces much more accurate forecasts than trend-based projections because it accounts for cohort aging. A company with 100 customers and 3% churn faces different dynamics than one with 1000 customers and 3% churn because the cohort composition differs.
Build a model that:
1. Lists every active cohort with their current MRR
2. Projects each cohort forward using retention curves
3. Adds new cohorts from your acquisition plan
4. Applies retention curves to new cohorts based on their expected attributes
This tells you realistic revenue growth rates. We’ve seen companies project 100% growth based on doubling new acquisitions, then realize through cohort modeling that they’d only hit 60% growth because existing customer decay offsets new revenue more than they expected.
Run this analysis monthly to spot problems early:
Compare newest cohort to cohort from 6 months ago at month 1. Is retention improving or degrading? If the newest cohort is retaining worse than the cohort from six months ago, something changed for the worse.
Look for sudden breaks in retention curves. If all cohorts show consistent patterns except one, something happened that month. Maybe you launched a broken feature, changed pricing, or adjusted targeting.
Check if expansion is accelerating or slowing. Are recent cohorts expanding faster at month 6 than older cohorts did at month 6? That suggests improving product-market fit.
Verify that longer-tenured cohorts are still contributing revenue growth. If your year-old cohorts are stagnant while newer cohorts are expanding, you have different value props for new customers versus existing ones.
Companies that master cohort analysis have monthly dashboards showing:
– Logo retention curves for the last 12 monthly cohorts
– Revenue retention curves for the same cohorts
– Net revenue retention by cohort with trend lines
– Segmented cohorts for each major customer attribute
– Cohort-based revenue forecasts
They review this data in management meetings alongside new acquisition metrics. When retention degrades, they investigate immediately rather than waiting for aggregate churn to spike.
They use cohort insights to inform acquisition strategy (which channels bring better customers?), product roadmap (what features increase retention?), and pricing (which segments have best unit economics?).
Most importantly, they understand that a few percentage points of retention improvement compound into enormous valuation differences. Improving month 3 retention from 88% to 91% means more customers survive to become long-term revenue. Do that across all cohorts and your business value increases by millions.
—
Q: How far back should we track cohorts?
Track as far back as your data allows, but focus analysis on the last 12-24 months. Older cohorts tell you historical patterns but may not reflect current product or market conditions. For forecasting, use retention curves from cohorts in the last 12 months since they best represent current customer behavior. If you’re pre-revenue or very early stage, you need at least 6 months of cohorts before patterns become meaningful.
Q: Should we track monthly or quarterly cohorts?
Monthly for the first 24 months of customer lifecycle, quarterly after that. Monthly granularity reveals early retention patterns and lets you spot problems quickly. After month 24, customer behavior stabilizes enough that quarterly tracking works fine. If you have low customer volume (under 20 new customers monthly), quarterly cohorts might be necessary to have statistical significance.
Q: What do we do when cohort analysis shows retention is getting worse?
First, segment to isolate the problem. Is it specific to a channel, customer size, or product tier? Second, look for correlation with product changes, pricing adjustments, or targeting shifts. Third, run qualitative analysis with churned customers to understand why. Fourth, implement fixes and monitor whether subsequent cohorts improve. The key is acting fast—retention problems compound. A cohort that’s retaining 5% worse at month 3 might retain 15% worse at month 12.