Your overall churn rate is 5% per month. Is that good? Is it getting better? Is it masking something worse? You cannot answer any of these questions with aggregate metrics alone.

Cohort analysis is the single most revealing exercise a SaaS company can perform. It groups customers by when they signed up and tracks their behavior over time. Instead of seeing a blurred average, you see whether each generation of customers is better or worse than the last.

According to Amplitude's research, companies that regularly perform cohort analysis are 2-3x more likely to identify product-market fit signals early. And ChartMogul's SaaS data shows that improving cohort retention by just 5 percentage points can increase customer lifetime value by 25-95%.

This guide covers how to build both retention and revenue cohorts, how to read the results, and the mistakes that lead teams to wrong conclusions.

What Cohort Analysis Actually Is

A cohort is a group of customers who share a common characteristic -- usually the month they signed up. Cohort analysis tracks how each group behaves over subsequent months.

Here is the core concept: instead of asking "what is our churn rate?", you ask "what is the churn rate for customers who signed up in January, February, March?" This reveals trends that aggregate numbers hide.

Consider two companies:

Company A: 5% monthly churn, stable for 6 months
Company B: 5% monthly churn, stable for 6 months

Identical on the surface. But cohort analysis might reveal:

Company A's January cohort churns at 8%, February at 6%, March at 4%, April at 3%. Each new cohort retains better. Product-market fit is strengthening.
Company B's January cohort churns at 3%, February at 4%, March at 6%, April at 8%. Each new cohort retains worse. The product is attracting the wrong customers as it scales.

Same headline number, completely opposite trajectories. This is why cohort analysis matters.

Retention Cohorts vs Revenue Cohorts

There are two types of cohort analysis, and you need both.

Retention Cohorts (Logo Retention)

Retention cohorts track whether customers are still active. Each cell shows what percentage of the original cohort is still paying.

Cohort	Month 0	Month 1	Month 2	Month 3	Month 4	Month 5	Month 6
Jan (40 customers)	100%	88%	80%	75%	72%	70%	68%
Feb (55 customers)	100%	90%	84%	79%	76%	74%	--
Mar (60 customers)	100%	92%	87%	83%	80%	--	--
Apr (70 customers)	100%	93%	89%	85%	--	--	--

Reading this table: the January cohort retained 88% of customers after month 1 and 68% after month 6. The April cohort retained 93% after month 1 -- a clear improvement. Your product is getting stickier.

Use a SaaS metrics calculator to compute retention rates from your own customer data quickly.

Revenue Cohorts (Dollar Retention)

Revenue cohorts track how much revenue each cohort generates over time, compared to their starting revenue. This is where expansion revenue and contraction show up.

Cohort	Month 0	Month 1	Month 2	Month 3	Month 4	Month 5	Month 6
Jan ($8K MRR)	100%	92%	88%	90%	93%	96%	99%
Feb ($12K MRR)	100%	94%	91%	94%	98%	102%	--
Mar ($15K MRR)	100%	95%	93%	97%	103%	--	--
Apr ($20K MRR)	100%	96%	95%	100%	--	--	--

Key insight from this table: even though customers churn (logo retention drops), remaining customers upgrade enough that revenue recovers and eventually exceeds the starting amount. The February cohort is at 102% by month 5 -- that is net revenue retention above 100%.

This is the signature of a healthy SaaS business. Customers leave, but the ones who stay spend more. For a deeper dive into why this matters, see our net revenue retention guide.

How to Build a Cohort Table

Step 1: Define Your Cohorts

Group customers by signup month. If you have very few customers (under 10 per month), consider grouping by quarter instead. Cohorts need enough members to be statistically meaningful.

Step 2: Choose Your Metric

For retention cohorts: active/inactive status each month. For revenue cohorts: MRR attributed to each customer each month.

Step 3: Build the Matrix

For each cohort (row), track the metric at each time period after signup (columns). Month 0 is always 100% -- it is the starting point.

Step 4: Calculate Retention Rates

For each cell:

Retention cohort: (Customers still active in month N) / (Original customers in cohort) x 100
Revenue cohort: (Total MRR from cohort in month N) / (Total MRR from cohort in month 0) x 100

Step 5: Color-Code the Results

Use conditional formatting to make patterns visible at a glance:

Green: Above 90% retention (or above 100% for revenue)
Yellow: 70-90% retention
Red: Below 70% retention

The visual pattern tells the story. If the bottom-left of your table is greener than the top-left, your product is improving. If it is redder, you have a problem.

Reading Cohort Results

Pattern 1: Improving Cohorts

Each new row retains better than the one above it. This means your product, onboarding, or customer selection is improving. This is the strongest signal of product-market fit.

What to do: Double down on acquisition. The customers you are attracting are increasingly right for your product.

Pattern 2: Degrading Cohorts

Each new row retains worse. This often happens when a company scales marketing aggressively and starts attracting less-ideal customers.

What to do: Tighten your ICP. Improve onboarding. Look at which channels produce the worst-retaining cohorts and reallocate spend.

Pattern 3: Flattening Curves

Within each row, retention drops steeply in months 1-2, then flattens. This is healthy -- it means the customers who survive the first two months tend to stick around long-term.

What to do: Focus onboarding improvements on the first 60 days. That is where the leverage is.

Pattern 4: Continuous Decline

Within each row, retention never flattens. Customers keep leaving at a steady rate even 6-12 months in. This signals a product problem, not an onboarding problem.

What to do: Investigate why long-tenured customers leave. Run exit surveys. The problem is likely missing features, poor reliability, or better competitors.

Pattern 5: Revenue Expansion

Revenue cohort percentages climb above 100%. Remaining customers are spending more than the cohort originally spent. As our SaaS churn rate benchmarks show, top-quartile SaaS companies achieve 120%+ net dollar retention.

What to do: This is working. Invest in upsell/cross-sell capabilities. Consider whether you can accelerate expansion through pricing changes or feature gating.

Dollar Retention Cohorts in Detail

Dollar retention cohorts are especially powerful because they combine four forces:

Churn (customers leaving -- reduces cohort revenue)
Contraction (customers downgrading -- reduces cohort revenue)
Expansion (customers upgrading -- increases cohort revenue)
Reactivation (churned customers returning -- increases cohort revenue)

The net of these four forces determines whether your cohort revenue grows or shrinks over time. Use a customer lifetime value calculator to translate cohort retention curves into LTV estimates.

Calculating Dollar Retention for a Cohort

For each month after signup:

Dollar Retention = (Churn Revenue Lost + Contraction Revenue Lost + Expansion Revenue Gained + Reactivation Revenue Gained) / Starting MRR

If a January cohort starts with $10,000 MRR and in month 6 generates $9,500 MRR, dollar retention is 95%. If it generates $11,000, dollar retention is 110%.

What Good Looks Like

Based on SaaS churn benchmarks:

Below 80% annual dollar retention: Serious retention problem. Fix the product before scaling.
80-100%: Acceptable for early-stage, but you need a plan to get above 100%.
100-120%: Good. Expansion is offsetting churn.
120%+: Excellent. Characteristic of best-in-class SaaS companies.

Tools for Cohort Analysis

Spreadsheet Method

For early-stage companies (under 200 customers), a spreadsheet works fine. Export your billing data, group by signup month, and calculate retention percentages. The formulas are straightforward.

Advantages: Free, flexible, you understand every number. Disadvantages: Manual, error-prone at scale, no automation.

Analytics Platforms

Tools like Amplitude, Mixpanel, and ChartMogul build cohort tables automatically from your event or billing data.

Advantages: Automated, handles large datasets, visualization built in. Disadvantages: Cost, setup time, may require engineering to instrument events.

Billing System Reports

Stripe, Chargebee, and Recurly all offer some form of cohort reporting built into their dashboards.

Advantages: No setup, uses your actual billing data. Disadvantages: Limited customization, may not segment the way you need.

Common Mistakes

Mistake 1: Too-Small Cohorts

A cohort of 5 customers is not statistically meaningful. If 1 customer churns, that is 20% churn -- but it might be random noise. You need at least 20-30 customers per cohort to see reliable patterns.

Fix: If monthly cohorts are too small, use quarterly cohorts. If quarterly cohorts are too small, you may not have enough customers for cohort analysis yet -- focus on individual customer conversations instead.

Mistake 2: Ignoring Seasonal Effects

If your product has seasonal usage patterns, comparing a January cohort to a July cohort may be misleading. The January cohort might retain better not because of product improvements, but because January buyers are more committed.

Fix: Compare year-over-year cohorts (January 2025 vs January 2026) in addition to sequential cohorts.

Mistake 3: Not Segmenting

An overall cohort table blends all customers together. But your self-serve customers and enterprise customers have completely different retention profiles.

Fix: Build separate cohort tables for each segment: by plan tier, by acquisition channel, by company size, by use case. The segment-level view is where actionable insights live.

Mistake 4: Only Looking at Retention Cohorts

Logo retention tells you how many customers stay. Revenue retention tells you how much money stays. A company can have declining logo retention but improving revenue retention if it is moving upmarket. You need both views to understand the business.

Mistake 5: Analyzing Without Acting

The most common failure is building a beautiful cohort table, nodding at the patterns, and doing nothing. Every cohort analysis should end with a specific action:

"March cohort retained 8% better than February. What changed? Can we replicate it?"
"Enterprise cohorts retain 2x better than self-serve. Should we shift acquisition spend?"
"Month 2 is where we lose the most customers. What happens at the 60-day mark?"

A Practical Monthly Rhythm

Here is how to integrate cohort analysis into your monthly operating review:

Week 1 of each month: Update your cohort table with the previous month's data
Compare: How did last month's cohorts perform relative to the month before?
Identify: Which cohort is performing best? Which is performing worst?
Investigate: What is different about the best and worst cohorts? (Channel? Pricing? Onboarding version?)
Act: Pick one specific improvement based on the analysis
Track: Monitor whether the improvement shows up in the next month's cohort data

Cohort analysis is not a one-time exercise. It is an ongoing discipline that gets more valuable as you accumulate more data and more cohorts to compare.

Sources

Amplitude -- "The Complete Guide to Cohort Analysis" (2025)
ChartMogul -- "SaaS Cohort Analysis Report: Benchmarks and Best Practices" (2025)

SaaS Cohort Analysis: Revenue and Retention