The signup chart is climbing, monthly recurring revenue is higher and the sales team is busy. Yet a look at last quarter's new customers reveals that many have already cancelled. A subscription company can grow its headline totals while repeatedly replacing people who leave. Cohorts show what happens to customers after acquisition, while revenue retention distinguishes expansion within existing accounts from the loss of subscriptions. Those are the numbers that reveal whether growth is becoming durable.
Meanwhile, customers who joined three months earlier are quietly cancelling. A large account has upgraded, masking the loss of smaller clients. The headline numbers are accurate, but they conceal a product that may struggle to keep the people it attracts.
A subscription business needs two views at once: how many new customers arrive and what happens to the customers already there. New signups can help. Retention reveals whether the product continues to deserve a place in someone's budget.
Registrations are an event, not a relationship
Someone can create a trial account out of curiosity, never activate a key feature and leave without making a payment. Counting that signup as equivalent to a customer paying for a year exaggerates the progress of the business.
The first distinction is between registered users, activated users and paying customers. The exact definitions depend on the service. A product for property agencies might consider activation to have occurred when a manager adds a property and assigns work. A reporting tool might require a first completed report.
Pick an activation event that represents a useful outcome rather than a convenient click. A completed tutorial may help someone learn the interface while still leaving them unable to solve their original problem.
A founder should also track time. Did a customer complete the useful action on the first day, first week or never? Are new users returning without reminders? Does a trial lead to a paid subscription?
A conversion metric is easier to interpret when its numerator and denominator reflect a meaningful journey.
Follow the customers who arrived together
Cohort analysis groups customers according to a shared starting event, such as the month they first subscribed. Rather than blending every customer into one total, it examines how each group behaves over time.
The ChartMogul introduction to cohort analysis describes cohort views for customer churn, retention, revenue and conversion. A cohort table lets a founder see whether people who joined in January are still subscribed in March or June.
Imagine that a software company acquired 100 paying customers in January. By the end of February, 85 remain. By the end of March, 70 remain. The total customer count might still rise if the company keeps acquiring new users. But the January cohort tells a different story about sustained value.
The numbers are hypothetical. The purpose is to reveal a pattern a cumulative signup graph can hide.
A cohort is most informative when the starting definition is stable. If a company changes its trial rules or pricing, comparing cohorts may require additional explanation. A customer who reactivates after cancelling may also be treated differently under different analytics conventions.
The chart is a lens, not a judgment by itself.
Customer retention and revenue retention answer different questions
Customer retention asks what proportion of the original customers remain active after a period. Revenue retention asks what has happened to the recurring money associated with the original customer group.
The difference becomes important when customers pay different amounts. A SaaS business may retain nearly every small account but lose one large customer. Its customer retention could look strong while recurring revenue drops. Another company could lose several low-paying accounts but gain expansion revenue from a few customers who need more of the service.
ChartMogul's customer-retention documentation explains how to analyze groups of paying subscribers across their lifecycle. It also notes that reactivation and subscription status choices affect the metric.
Do not compare raw percentages unless the underlying definitions match. Monthly versus annual subscriptions, different trial treatments and data corrections can change the numerator or denominator.
A founder who wants to improve customer experience should inspect both the account count and the money attached to those accounts.
A worked example of net revenue retention
Consider a fictional subscription company starting a quarter with $10,000 in recurring revenue from a particular group of existing customers.
During the quarter, customers from that starting group upgrade by $2,000, downgrade by $500 and cancel subscriptions worth $1,000. The group therefore ends the quarter with $10,500 in recurring revenue.
Under the simplified formula, net revenue retention is ($10,000 + $2,000 − $500 − $1,000) ÷ $10,000, or 105%.
The business has expanded recurring revenue from its existing cohort even though some customers left. Revenue from entirely new customers is excluded from this calculation. That separation is fundamental.
The ChartMogul net MRR retention guide describes how expansion, contraction, churn and reactivation affect the measure. Stripe's SaaS metrics guide also explains the role of net revenue retention as a measure of the revenue associated with existing customers.
An NRR above 100% can indicate that expansion is more than offsetting losses. It does not prove that every original customer is happy or that cancellations are insignificant. A large upsell can obscure a concerning pattern of smaller accounts disappearing.
Review customer counts and the distribution of expansion, not just the final ratio.
Gross retention reveals the losses that upgrades hide
Net revenue retention includes expansion. Gross revenue retention excludes it, focusing on revenue preserved after downgrades and cancellations.
In the same example, starting recurring revenue is $10,000. The company loses $1,500 through downgrade and cancellation, leaving $8,500 before upsells. Simplified gross revenue retention is 85%.
Both numbers are valid under the stated assumptions. An NRR of 105% and gross retention of 85% tell different stories. The company is growing within some relationships while losing a significant share of starting revenue elsewhere.
A founder should resist treating the more flattering metric as the only one worth showing. Gross retention can expose underlying dissatisfaction, pricing problems or weak onboarding even when expansion remains strong.
Revenue metrics are also sensitive to how downgrades, annual contracts, one-time fees and reactivations are recorded. Accounting and subscription platforms may differ in implementation. State the measurement period and method when presenting a figure publicly.
The most useful dashboard supports investigation, not selective storytelling.
Find where customers leave the journey
A retention curve may decline sharply after the first invoice, after a trial ends or when customers encounter an advanced workflow. Each pattern suggests a different question.
If customers leave in the first month, examine what they expected to accomplish, how quickly they reached that outcome and whether setup required too much effort. If cancellations spike after a seasonal business cycle, the problem could involve usage patterns or pricing structure rather than interface quality.
A company should talk with customers who are leaving when appropriate, while respecting their choice and privacy. A cancellation reason selected from a menu can provide an initial clue, but it rarely explains the complete cause.
Look at usage in context. Low login frequency is not always evidence of low value. A payroll product may be used mainly at defined intervals. A property-reporting tool may matter once a month. Activity metrics must follow the actual job the software performs.
The question is whether customers successfully use the service at the moments when they need it and find the ongoing cost justified.
Acquisition can temporarily conceal a retention problem
Suppose a company adds 200 new customers each month but loses 180 existing ones. The total count increases. Marketing reports growth, yet the company needs a constant stream of new customers just to maintain its base.
That treadmill can become expensive. Advertising, onboarding, sales assistance and customer support all cost money. If customers leave before delivering enough contribution margin to cover acquisition spending, new registrations may make the financial problem worse.
A founder should connect retention analysis with acquisition cost and customer economics. Do not assume that every paying customer contributes the same profit. Some require intensive support or consume more infrastructure resources than their monthly fee covers.
Use customer cohorts to observe how much revenue and support demand arise over time. A growing subscription count becomes more persuasive when older cohorts remain healthy and new cohorts show a similar or improving pattern.
The objective is a durable relationship between product value and customer willingness to keep paying.
An executive dashboard that fits on one screen
A useful small-SaaS dashboard can start with six measures: new paying customers, customers lost, recurring revenue added from new customers, recurring revenue lost from cancellations, expansion and contraction within existing customers, and cohort retention over time.
Add contextual information such as activation and customer-support issues where the product's workflow calls for it. Resist adding dozens of disconnected vanity metrics.
Make the date period explicit and compare like with like. A company billing annually should account for that structure rather than treating payment receipt in one month as proof the product was used during that month.
Allow managers to click from a summary into the underlying cohort or customer segment without exposing unnecessary private information. A meaningful operational dashboard should explain the change, not merely display an attractive number.
A warning flag might be falling gross retention despite rising NRR. Another could be a new cohort activating more slowly than earlier cohorts. Each should lead to a concrete investigation.
How to use the numbers without overreacting
Small samples can be volatile. If a company has ten customers, losing one produces a large percentage change. That deserves attention but not a universal conclusion about product-market fit.
Look at the individual circumstances behind movements and compare several periods. One cancellation may arise from the customer closing their business. Several may reveal the same missing feature or pricing mismatch. Revenue retention alone cannot distinguish those causes.
Benchmark comparisons should account for company stage, contract value, customer type and industry. An enterprise SaaS vendor with annual contracts has a different lifecycle from a consumer subscription app. An impressive retention rate in one category is not an automatic target for every other model.
ChartMogul's cohort documentation is most useful when the team uses it to locate customer behavior, not to chase a single percentile.
Product decisions should combine numerical evidence, customer conversations and economics.
The number worth showing investors and employees
A company that attracts customers and keeps delivering value usually has a more credible growth story than one that relies on repeated acquisition to replace customers who leave. The evidence appears in successive cohorts, renewals and the actual economics of serving those accounts.
A founder can still celebrate new signups. They are the beginning of a possible relationship. Retention shows whether that relationship developed.
When a dashboard announces record growth, ask one follow-up question: “What happened to the people who signed up three months ago?” The answer may change which feature deserves the next week of development.
References and background
Our explainers link to source documents so readers can explore the evidence and interpret it in context. Websites and policies may change after publication.
- ChartMogul, *Cohort Analysis*help.chartmogul.com
- ChartMogul, *Customer Retention Cohorts*help.chartmogul.com
- ChartMogul, *Net MRR Retention Cohorts*help.chartmogul.com
- Stripe, *Essential SaaS Metrics*stripe.com