Churn is the percentage of customers who cancel their subscription in a given period. If you start January with 200 customers and end it with 190, you have 5% monthly churn. If that rate holds all year, you lose more than half your customer base annually. Not to a competitor. Just to attrition.
This is why churn is the most important metric to manage in a subscription business. Revenue growth looks fine until you do the cohort analysis and see that every month you are pouring water into a bucket that is draining faster than you fill it.
The two types of churn
Voluntary churn is when customers actively decide to cancel. They found an alternative, their needs changed, they did not get value from the product, or their budget was cut.
Involuntary churn is when customers are cancelled without actively deciding to leave. Payment failures, expired credit cards, and account issues cause a surprising amount of churn for most SaaS products. In many early-stage businesses, 20-40% of total churn is involuntary.
Fix involuntary churn first. It requires minimal effort and the customers often want to stay. Dunning emails (automated reminders when a payment fails), smart payment retry logic, and proactive outreach when cards expire can recover a significant percentage of involuntary churns.
What causes voluntary churn
The reasons customers cancel fall into a small number of categories:
Never got value. The customer signed up, struggled through onboarding, never reached the moment of value, and cancelled after a month of paying for something they barely used. This is an onboarding problem, not a product problem.
Got value but moved on. The customer used your product to solve a specific problem, solved it, and then had no reason to keep paying. This is a product scope problem: your product is too narrow, too one-time-use, or not embedded enough in the customer's ongoing workflow.
Found a better alternative. A competitor launched a feature you do not have, or a new product entered the market at a lower price with sufficient functionality. This is a competitive problem.
Budget was cut. External economic pressure caused the customer to cut tools they were not willing to fight to keep. This often reveals that your product was "nice to have" rather than "essential." The products that survive budget cuts are the ones that would hurt to lose.
Customer expectations were not met. The product did not do what the customer expected based on how it was sold or described. This is a sales and marketing alignment problem.
Each category requires a different response. The mistake is treating all churn as one problem with one solution.
The churn interview: the most underused tool in SaaS
Every churned customer should receive a personal outreach within two weeks of cancellation. Not an automated survey. A personal email from a founder or account manager.
The email should be short: "I noticed you cancelled your account. I want to understand what we did not deliver for you. Can we talk for 10 minutes this week?"
Many will not respond. The ones who do will tell you something useful that no amount of product analytics can reveal. They will tell you the specific moment the product stopped working for them, the specific alternative they chose instead, or the specific expectation that was not met.
The patterns across these conversations are your product roadmap. If seven of the last ten churned customers mention the same missing feature, you now know what to build. If they all mention that the product was confusing to set up, you know what to fix first.
Churn prevention metrics to track
Early engagement metric. Identify the action in your product that predicts which users will stay. For most SaaS tools, this is a specific action taken in the first three to seven days. Track what percentage of new users take this action. Users who do not are high-churn risks.
Feature adoption rate. Users who use only one feature of a multi-feature product are more likely to churn than users who use three or more. Low feature adoption is an early churn warning signal.
Login frequency. A user who logs in weekly is much less likely to churn than one who logs in monthly. Sudden drops in login frequency predict imminent cancellation.
Support ticket volume. A counterintuitive signal: users who have never opened a support ticket churn at higher rates than users who have. If they have never asked for help, they may have given up without telling you. Monitor inactive users proactively.
The retention features that work
In-app usage summaries. Show users what they have accomplished with your product. A weekly email or in-app notification that says "You sent 12 invoices last month and saved an estimated 4 hours" reminds users of the value they are receiving. This is particularly effective for products where the value is not visible in the interface day to day.
Customer success for high-value accounts. For accounts above a certain revenue threshold, proactive check-ins from a customer success manager make a measurable difference in retention. This does not have to be a dedicated hire: founder-led customer success for the first 20-30 accounts is common and valuable.
Feature announcements for inactive users. When a user has not logged in for two weeks, send them a targeted email about a specific feature they have never tried. This reactivates some percentage of users who would otherwise drift to cancellation.
Commitment mechanisms. Annual plans have lower churn than monthly plans, almost universally. Not because annual customers are more committed but because the decision to cancel is harder when the billing is less frequent. Offer meaningful discounts for annual plans and actively promote them.
The technical infrastructure for churn management
Managing churn at scale requires knowing which customers are at risk before they cancel. This means tracking engagement data and surfacing patterns automatically.
A basic churn prediction setup requires:
- Event tracking for key product actions
- A way to flag users who have not taken those actions recently
- Automated outreach (email or in-product message) triggered by inactivity
Our data analytics services can help you set up this kind of behaviour monitoring for your product. You do not need a data science team. You need well-instrumented event tracking and a simple reporting layer.
If you are building a new product and want to make sure churn monitoring is built in from day one rather than retrofitted later, talk to our team. The cost of doing this correctly at build time is a fraction of the cost of adding it to an existing product that was not designed for it.