Most growth plans add customers at the top, lose them at the bottom, and then wonder why paid budgets keep rising. Retention improves every other number at once: lifetime value, payback, the return on each acquisition channel, and the most customers your base can ever hold. The maths below uses hypothetical numbers. After it comes a way to find where retention leaks, and the order to fix things in.
About that “five times cheaper” claim
You have probably read that keeping a customer is five times cheaper than winning a new one. We could not trace it to original data. The 2014 Harvard Business Review article that usually gets cited gives a range of “five to 25 times”, and itself says the figure depends on “which study you believe” and on the industry. No single study backs the number.
The better-known finding in that article comes from Bain & Company research, which HBR summarises as “increasing customer retention rates by 5% increases profits by 25% to 95%”. The original Bain brief from 2001 describes it in the context of financial services. It is decades old and specific to one sector. Use it as a direction, not a forecast for your business.
You don’t need anyone’s multiple anyway. Working out what retention is worth to you takes an afternoon.
The maths: what a small change in churn does
Take a hypothetical subscription business. All numbers are illustrative.
- Average revenue per account: ₹2,000 a month
- Gross margin: 80%, so ₹1,600 of gross profit per account per month
- Customer acquisition cost (CAC): ₹12,000 per new customer
- New customers: 100 a month
Now compare monthly churn of 5% with monthly churn of 3%.
Lifetime value
In the simple model, average customer lifetime is 1 ÷ monthly churn.
| 5% monthly churn | 3% monthly churn | |
|---|---|---|
| Average lifetime | 20 months | 33.3 months |
| Lifetime gross profit (LTV) | ₹32,000 | ₹53,333 |
| LTV ÷ CAC | 2.7 | 4.4 |
| Customers still active after 12 months | 54% | 69% |
Two points off churn and LTV goes up by two-thirds, with pricing, ads and sales untouched. The simple formula overstates LTV when churn is very low, so cap the assumed lifetime at a sensible horizon, say 3 to 5 years.
Payback on a cohort
The textbook payback formula (CAC ÷ monthly gross profit) gives 7.5 months in both cases because it ignores churn. Real cohorts shrink while you are trying to earn back the CAC. Counting the gross profit a cohort of 100 customers actually earns, month by month, as some of them leave:
- At 5% churn, the cohort earns back its ₹12,00,000 of acquisition cost in about 9.2 months.
- At 3% churn, it takes about 8.4 months.
Shorter payback means cash comes back sooner and can fund the next cohort. For a company short of cash, that can decide whether it grows from revenue or has to raise money.
The ceiling on your customer base
Founders miss this effect more than any other. Add 100 customers a month, lose a fixed percentage, and your customer base approaches a ceiling of new customers ÷ churn.
| 5% monthly churn | 3% monthly churn | |
|---|---|---|
| Active customers after 12 months | about 919 | about 1,021 |
| Active customers after 24 months | about 1,416 | about 1,729 |
| Long-run ceiling | 2,000 | about 3,333 |
Reaching 1,729 customers by month 24 at 5% churn would take about 122 new customers a month instead of 100. At ₹12,000 each, that is roughly ₹2,64,000 of extra acquisition spend every month to keep pace with the lower-churn version of your own business. CAC usually rises as you push spend harder, so the real gap is wider.
Expansion makes it compound
So far, retained customers pay the same every month. Many don’t. Customers who stay often add seats, move up a plan or use more. Net dollar retention (NDR) captures this: recurring revenue from a cohort a year later, including expansion and minus churn and downgrades, divided by what it was at the start. Slack’s 2019 S-1 reported NDR of 143% as of 31 January 2019, so its paid customer base from a year earlier was spending 43% more. Above 100%, the existing base grows before you acquire anyone. The net dollar retention lesson shows how to calculate it.
If you do not sell subscriptions
E-commerce, D2C brands, marketplaces and apps work the same way, with repeat purchase in place of renewal. Swap “monthly churn” for “share of customers who buy again within a set window”, such as 90 days.
A first order that earns ₹400 of contribution after product cost, shipping and payment fees, against ₹600 to acquire the customer, loses ₹200 per first-time buyer on day one. If 30% of them buy again within 90 days at the same contribution, the cohort recovers ₹120 per original customer. At 40% it recovers ₹160. Repeat purchase decides whether paid acquisition makes money at all, so measure it by cohort exactly as you would subscription retention.
Diagnose where retention leaks
“Churn is high” is a symptom. You need to know when people leave, who leaves and why, and four checks in order get you there.
1. Plot retention curves by cohort
Group customers by the week or month they started and plot the share still active over time. The shape tells you more than any single number. Amplitude’s retention documentation explains the main ways to measure return, such as returning on a specific day versus on or after it. Pick one definition and stick with it. The retention curves and churn lesson and the cohort analysis lesson go through the method step by step.
| Curve shape | Likely leak | Where to look |
|---|---|---|
| Steep drop in the first days or weeks, then flat | Activation | Sign-up to first value, onboarding |
| Gradual slope that never flattens | Ongoing value or habit | Core use case, frequency, fit of segment |
| Flat, then a sharp drop at renewal or month 12 | Price or perceived value at renewal | Annual renewals, procurement, competitor offers |
| Newer cohorts worse than older ones | Acquisition mix or product change | Channels, campaigns, recent releases |
| Newer cohorts better than older ones | Something is working | Find it and do more of it |
2. Split voluntary from involuntary churn
Some customers never choose to leave. Their card expires or a payment fails. That is involuntary churn, and it needs a different fix. Check your billing data for subscriptions cancelled after failed payment attempts. Stripe, for example, documents a recommended default of 8 retry attempts within 2 weeks in its Smart Retries guide. The same page notes that Stripe does not automatically retry payments on India-issued cards. If you bill Indian customers, you need your own reminder and mandate flow.
3. Segment the churned
Compare customers who left with those who stayed, by plan, company size, acquisition channel, use case and country. A 5% churn rate often hides one segment at 2% and another at 12%. Sometimes the fix isn’t in the product at all. Stop acquiring the segment that never retains.
4. Ask, but weight behaviour more
Cancellation surveys and exit calls give you the reason people state. Usage data shows what they did. When the two disagree, “too expensive” often means “I did not get enough value to justify the price”. Read both.
What to fix first
Fixes early in the lifecycle multiply everything after them, so the sequence goes like this.
- Involuntary churn. Usually the fastest win: retry logic, card-expiry reminders, a clear update-payment page and dunning emails, with no product redesign. See failed payment recovery.
- Activation. If the curve drops steeply early, define the moment a new user gets value and get more people there sooner. Start with defining your activation metric. Every later improvement then acts on a bigger base of activated users.
- The segment that never retains. Move acquisition spend away from the channels and audiences whose cohorts churn fastest. Retention improves because different people arrive.
- Ongoing value and habit. If the curve never flattens, work on the core use case, the reason to come back every week. This is slower and usually needs product work.
- Renewal and win-back. Prepare customers for renewal before the date, and run a structured win-back campaign for inactive users. Win-back is worth doing, but it recovers a fraction of what better activation would have kept.
How to make the case internally
Retention work competes with acquisition for budget, and acquisition has cleaner dashboards. Put the trade in front of people.
Rebuild the ceiling table above with your own numbers and show what a one-point cut in monthly churn is worth in customers and revenue at months 12 and 24. Then say it in acquisition terms: “a one-point cut in churn is worth the same as X extra customers a month, which would cost ₹Y in ads.”
Agree one retention metric the whole team reports against, such as month-3 retention for new cohorts or NDR for the base. And review cohort curves monthly in the same meeting as the acquisition numbers, so nobody celebrates a record sign-up month that has churned out by the next one.
A worksheet for your own numbers
| Pull this | Put it here | It tells you |
|---|---|---|
| Monthly churn and CAC for the last 6 months | The LTV, payback and ceiling tables above, with your real figures | What a one-point cut in churn is worth |
| Retention curves for your last 6 monthly cohorts | The curve-shape table | Which leak you have |
| Share of churn that followed a failed payment | Priority 1 in the fix list | Whether retries and reminders come first |
| Churn by acquisition channel | Priority 3 in the fix list | Which segment retains worst |
| One chosen fix and a target | The next cohort, not the whole base | Whether the fix worked |
Questions people ask
Is it really five times cheaper to retain a customer than to acquire one?
That figure is widely repeated but rarely traced to original data. A 2014 Harvard Business Review article gives a range of five to 25 times and says it depends on which study you believe. Rather than rely on a generic multiple, calculate your own: compare your cost per acquired customer with the cost of the retention work you plan to do.
How much does reducing churn increase LTV?
In the simple model, average customer lifetime is one divided by the churn rate, so lifetime value scales with it. Cutting monthly churn from 5% to 3% raises average lifetime from 20 to about 33 months, an increase of roughly two-thirds, with no change in price or acquisition cost.
Where should I start if churn is high?
Plot retention curves by sign-up cohort and find where the curve drops most. A steep drop in the first week or month usually means an activation problem. A curve that never flattens points to weak ongoing value. Separately, check how much churn comes from failed payments, which is often the quickest to fix.
What is the difference between logo churn and revenue churn?
Logo churn counts customers lost. Revenue churn counts recurring revenue lost, and net revenue measures such as net dollar retention also add expansion from customers who stay. A business can lose customers yet grow revenue from its existing base if expansion outweighs losses.
