Businesses invest heavily in acquisition and then treat retention as something that either happens or does not.
Where churn concentrates
Disproportionately in the first weeks after purchase.
Which points at onboarding rather than at long-term satisfaction.
Cohort curves show this clearly and are rarely examined.
Expectation mismatch
Sales promising more than the product delivers.
Which produces customers who were never going to stay.
This shows as a churn problem and originates upstream.
Voluntary and involuntary churn
Customers who leave and customers whose payment fails.
Which require completely different remedies.
Failed payment recovery is frequently the cheapest retention improvement available.
Measuring retention
Logo retention, revenue retention and net revenue retention.
Which answer different questions and are frequently conflated.
Net revenue retention above one hundred percent means expansion exceeds churn.
Onboarding design
Getting a customer to their first meaningful outcome quickly.
Which is the strongest lever available for early churn.
Defining what that outcome is precedes designing for it.
Health signals
Usage patterns preceding cancellation.
Which can be identified from historical data.
Intervention based on them works where the intervention is useful rather than a retention offer.
Discounting to retain
Offering price reductions at cancellation.
Which retains some customers and teaches others to threaten cancellation.
Fixing the reason is slower and does not compound the problem.
Why retention outperforms acquisition
Retained customers cost nothing to acquire again and frequently spend more over time.
Which makes small retention improvements compound in a way acquisition improvements do not.
The widely quoted multiples on this are unreliable; the direction is not.
Segmenting churn
Different customer types leave for different reasons and at different rates.
Which averages conceal completely.
The segment churning fastest is frequently one that was never a good fit.
Exit surveys
Asking why customers left.
Which produces low response rates and useful patterns anyway.
The stated reason and the real reason frequently differ, and both are informative.
Product and retention
Most durable retention improvement comes from the product doing its job better.
Which is slower than any campaign and is the only thing that holds.
Where to start
Plot a cohort curve, find where the drop concentrates, and work on that period.
The role of price
Customers who joined on a discount behave differently from those who paid full price.
Which shows up clearly in cohort analysis.
Discount-acquired cohorts churn faster and should be measured separately.
Support quality
Response time and resolution rate.
Which correlate with retention in most measured settings.
Support is treated as a cost centre and functions partly as a retention channel.
Habit formation
Products used regularly are cancelled less.
Which is why usage frequency is the most watched early signal.
Increasing natural use frequency is harder than it sounds and worth attempting.
Contract length
Annual commitments defer the churn decision rather than removing it.
Which flatters monthly numbers and concentrates risk at renewal.
Renewal rates are the honest measure for annual contracts.
Expansion revenue
Existing customers buying more.
Which can offset churn entirely in some models.
A practical summary
Measure by cohort, look at the first weeks, fix the product reason rather than the symptom.
Winning customers back
Re-engaging those who left.
Which converts at rates well above cold acquisition for some businesses.
It works best where the reason for leaving has actually been addressed.
Communication frequency
Contact that is useful against contact that is noise.
Which is judged by the recipient rather than the sender.
Unsubscribe rates are a direct measure and are usually ignored.
Loyalty programmes
Rewards for repeat purchase.
Which retain customers who would have stayed and cost margin doing so.
Incrementality applies here as much as to acquisition spending.
Organisational ownership
Who is accountable for retention.
Which in many businesses is nobody in particular.
Acquisition has a budget and an owner; retention frequently has neither.
The starting point
Pick one cohort, follow it for six months, and find out where people actually leave.
Benchmarks
Published retention rates vary enormously by model and by definition.
Which makes them weak comparators.
Your own trend is what tells you whether anything is improving.
The one-line summary
Retention is mostly an onboarding and expectation problem, and it is measured by cohort or not at all.
A closing caution
None of this is prescriptive. Businesses differ by sector, by scale and by stage, and practices that work well in one context fail in another for reasons that are not always visible from outside.
What is consistent is that the businesses handling these questions well tend to have written something down, measured it in a defined way, and reviewed it on a schedule rather than when a problem forces the issue.
Where a decision carries legal, tax or employment consequences, professional advice specific to your jurisdiction is worth the cost, and this article is general description rather than advice.
One more thing worth saying
Most of what is written about running a business is written by people selling something, which shapes what gets emphasised and what gets left out.
The unglamorous parts, keeping records, reading the contract, updating the forecast, rarely feature because nobody can sell them. They are also the parts that most reliably separate businesses that survive from businesses that do not.
Whatever you take from this, take the habit of writing the number down and looking at it again next month.