Do the Churn Math First
Before you change anything, quantify the problem. Monthly churn is the percentage of customers (or revenue) you lose each month. The number sounds small and is not. At 5 percent monthly churn, you lose nearly half your customers in a year. At 7 percent, the average customer stays only about fourteen months. Small percentages compound into large holes.
Separate two kinds of churn, because they have different fixes. Customer churn counts logos leaving. Revenue churn counts dollars leaving. They can diverge: you might lose many small accounts but grow revenue if your remaining customers expand. Track both, and for any business that can grow accounts over time, watch net revenue retention, which folds in expansion from existing customers. Net revenue retention above 100 percent means your existing base grows even with zero new sales.
The math also sets a ceiling on growth. If monthly churn is high, you eventually hit a wall where new customers only replace the ones leaving and growth stalls no matter how much you spend on acquisition. Knowing your numbers turns churn from a vague worry into a specific target, and tells you whether retention or acquisition is the more urgent fire.
- Calculate monthly customer churn and monthly revenue churn separately.
- Translate the percentage into average customer lifetime: 1 divided by monthly churn rate, in months.
- If accounts can expand, track net revenue retention. Above 100 percent means your base grows on its own.
- Use the numbers to decide where the bigger fire is: the leaky bucket or the empty top.
Find Out Why People Actually Leave
You cannot fix churn from a spreadsheet alone. Numbers tell you that people leave and roughly when. They do not tell you why, and the why is where the fix lives. The most common mistake is to guess the reason, build a feature against the guess, and watch churn stay flat because the real cause was something else entirely.
Run a cancel survey at the moment of cancellation, when the reason is fresh and honest. Keep it to one or two questions: why are you leaving, and what would have made you stay. Make it optional and short. Even a modest response rate gives you a ranked list of real reasons instead of your assumptions. Categorize the answers and you will usually find that a handful of causes account for most of the churn.
Then go deeper than the survey. Email or call recently churned customers and ask what happened. People are surprisingly willing to tell you, especially if you make clear you are not trying to win them back, just to learn. Listen for the difference between people who never got value (an onboarding and activation problem) and people who got value and left anyway (a value-delivery or pricing problem). These are different diseases with different cures.
- Add a one or two question cancel survey at the moment of cancellation.
- Categorize the answers so you can see which few reasons drive most of the loss.
- Interview recently churned customers. Make clear you want to learn, not to sell.
- Separate 'never got value' churn from 'got value and still left' churn. They need different fixes.
Read Cohort Retention to See the Pattern
A single churn number hides the story. Cohort retention groups customers by when they joined and tracks how many remain over time. This reveals the shape of your churn, and the shape tells you where the problem is. Plot the percentage of each monthly cohort still active in month one, month two, month three, and so on.
Look at where the curve drops. If most churn happens in the first month, you have an activation problem: people sign up, never reach the point where the product is useful, and leave. That is fixable in onboarding. If the curve declines steadily over many months, you have an ongoing value problem: the product works but does not stay worth paying for. If the curve flattens after the early drop, that flat line is your retained core, the people who genuinely need the product.
Cohorts also show whether your changes are working. When you improve onboarding, newer cohorts should retain better than older ones at the same age. Comparing cohorts over time is how you prove a fix actually moved the needle rather than fooling yourself with a noisy overall average. If newer cohorts are not retaining better, your fix did not address the real cause.
- Build a cohort retention table: customers grouped by join month, tracked month over month.
- Early steep drop equals an activation and onboarding problem.
- Slow steady decline equals an ongoing value-delivery problem.
- Compare new cohorts to old ones to confirm a fix actually improved retention.
Fix Onboarding and the Path to Value
Most early churn is decided in the first session. If a new user does not reach the aha moment, the point where the product visibly solves their problem, they churn before they ever experience the value you built. The single highest-impact retention work is shortening and clarifying the path from signup to that first real win.
Map the steps a new user takes from signup to value and find where they fall off. Cut every step that is not strictly necessary to reach the first win. Replace blank-screen empty states with a clear next action or a template. If activation depends on a setup task users avoid, do it for them or make it trivial. The goal is that a new user gets a concrete result fast, before motivation fades.
Onboarding is not a tour of features. It is a guided path to one outcome that matters to the user. Tie it to a measurable activation event, the specific action that correlates with people sticking around. Then drive as many new users to that event as fast as you can. Improving activation is usually the biggest single lever on churn, because a user who never activated was never really a customer.
- Identify your aha moment: the action after which users reliably stick.
- Map signup to first value and delete every step that is not essential.
- Replace empty states with a clear next action, a template, or done-for-you setup.
- Pick one activation metric and relentlessly push new users to hit it fast.
Keep Delivering Value After the Honeymoon
Activated customers still churn if the product stops being worth its price. This is the slow-decline churn that cohort curves reveal, and it comes from value fading: the user solved the immediate problem and no longer needs you, the product became part of the furniture and the value is invisible, or a competitor offers more. Fixing it means keeping the value present and growing over the relationship.
Build habit and deepening use, not just a one-time result. Surface the value the customer is getting so it does not become invisible. A periodic summary of what they accomplished, a nudge toward a feature that solves their next problem, a reason to come back this week. Customers who use more of the product and embed it in their workflow churn far less than those who touch one feature once.
Pay attention to leading indicators rather than waiting for the cancellation. Declining logins, dropping usage, an unanswered support ticket. These predict churn weeks before it happens, which gives you time to reach out. A short, human check-in with a customer whose usage is sliding often saves the account, and always teaches you something about why value is slipping.
- Watch leading indicators (login frequency, core-action usage) that predict churn before it happens.
- Make the value visible: remind customers what they accomplished and what to do next.
- Drive deeper adoption. Customers embedded in more of the product churn less.
- Reach out to at-risk accounts while there is still time, and learn from every save and loss.