CHURN METRICS

Customer acquisition vs retention cost: how to compare them

How to calculate customer retention cost next to CAC, why the 5 to 25 times rule is not your number, and how to compare both per dollar of MRR you keep.

Published 8 min read

Customer acquisition vs retention cost: how to compare them

Customer acquisition cost (CAC) is what you spend to win one new customer. Customer retention cost is what you spend to keep the customers you already have: retention tools, the team’s time and the discounts you give. To compare them fairly, divide each by the revenue it actually brings or keeps: the cost of $1 of new MRR next to the cost of $1 of MRR that would otherwise have left. The widely quoted “5 to 25 times cheaper to retain” is a range across studies and industries, not a subscription benchmark. Your own billing data gives a better answer in an afternoon.

This guide covers what goes into each cost, how to count discounts at their real cost in Stripe, Chargebee, Recurly and Paddle, how to measure the revenue retention really saves, and when acquisition is the better place to spend.

What is the difference between acquisition cost and retention cost?

Acquisition cost (CAC) Retention cost
What it pays for Winning a customer who has never paid you Keeping a customer who already pays you
Usual costs Ads, sales and marketing salaries and commissions, agencies, tools Retention and dunning tools, support and success time, discounts and free months given
Divided by New customers won in the period Customers or MRR kept in the period
Where the data lives Your P&L and marketing tools Your P&L, plus your billing tool for discounts and recovered payments

Recurly’s glossary defines CAC as total sales and marketing spend over a period divided by the new customers acquired in that period, and says it “rolls up every cost that goes into acquiring customers, not just media spend”. Paddle groups the costs into three: salaries, tools and spend. The LTV:CAC ratio guide covers CAC in detail, and the CAC calculator does the division for you.

Retention cost has no single official definition. Build it the same way: everything on your P&L that exists to keep current customers, over the same period.

How do you calculate customer retention cost?

Add up four kinds of cost for the period:

  1. Tools. Dunning, cancel flow, win-back and customer success software, plus any fee your payment provider charges for retries or card updates.
  2. People. The share of support and customer success time spent on keeping accounts: saving cancellations, chasing failed payments, renewal calls. If one person splits time between onboarding new customers and saving old ones, split their cost the same way.
  3. Offers. Discounts, free months and extended trials that customers actually redeemed, at their full cost (see the next section).
  4. Win-back campaigns. Emails, discounts and any paid ads aimed at customers who already cancelled.

Then divide. Per customer: retention cost ÷ average number of active customers in the period. Per dollar: retention cost ÷ MRR kept that would otherwise have left. The per-dollar version is the one you can compare with acquisition.

How do you count discounts at their real cost?

A discount costs what customers stop paying, for as long as it lasts. All four billing tools let you set that length, and the length changes the cost more than the percentage does:

Tool Duration options
Stripe once (first invoice only, the default), repeating for a number of months, forever
Chargebee One Time, Forever, Limited Period
Recurly Single use, forever, limited time (days, weeks or months), a number of billing periods
Paddle One-time by default, or recurring forever or for a number of billing periods

Stripe’s example: a 50% coupon with duration_in_months=4 applies to every invoice in those four months. On a $100 monthly plan that costs $200, the price of two full months. A forever coupon has no end, so its cost grows with every renewal. Count it each month as long as the subscription runs. The same applies to the retention coupon Stripe’s customer portal can offer when a customer tries to cancel: it is a normal coupon, with the duration you gave it.

Count only redeemed offers, not offers shown. And count the offer for its full duration in the period you gave it, so a long discount does not look cheap in the month it starts.

Where does the “5 to 25 times” figure come from?

It is usually traced to a 2014 Harvard Business Review article by Amy Gallo: “Depending on which study you believe, and what industry you’re in, acquiring a new customer is anywhere from five to 25 times more expensive than retaining an existing one.” The sentence itself gives the warning: a wide range, depending on study and industry.

Three reasons it does not carry over to a subscription business as is:

  • Part of your churn costs almost nothing to prevent. A renewal that fails because a card expired is not a decision to leave. Recovering it takes retries and an email, so its cost per dollar is far lower than a sales-led save. See involuntary churn.
  • Part of what you “save” would have stayed anyway. A customer who accepts a discount in a cancel flow might not have left without it. Counting every accepted offer as saved revenue makes retention look cheaper than it is.
  • CAC varies by channel. Organic signups and paid ads can differ widely. One blended ratio hides both.

How do you compare them per dollar of MRR?

Use the same period and the same horizon on both sides, then compare the cost of each dollar of MRR still active 90 days later.

Compare acquisition and retention cost
  1. 01 Pick one period and one horizon The same quarter on both sides, MRR still active after 90 days
  2. 02 Add up acquisition spend Ads, sales and marketing salaries and commissions, tools
  3. 03 Add up retention spend Tools, team time, discounts redeemed at their full cost
  4. 04 Measure the MRR truly kept A holdout shows what would have stayed anyway
  5. 05 Compare cost per $1 still active Cost of new MRR next to cost of MRR kept

An illustrative quarter, with made-up figures to show the method:

Acquisition Retention
Spend in the quarter $60,000 (ads, salaries, tools) $6,300 ($1,500 tools, $3,000 team time, $1,800 discounts)
MRR won or kept, as reported $6,000 new MRR from 120 customers $3,000: $1,800 of failed payments recovered, $1,200 of cancellations saved
Cost per $1, as reported $10.00 $2.10
MRR that is truly incremental $6,000 $2,100 (holdout: half the saves and $300 of recoveries would have happened anyway)
Still active after 90 days $5,400 (90%) $1,680 (80%)
Cost per $1 still active $11.11 $3.75

The $1,800 of discounts is the cancellation saves’ 50% off for 3 months: $600 a month for 3 months, counted in full. In this example, retention is about three times cheaper per dollar kept, not 5 to 25 times. And the reported figure, $2.10, flatters retention by almost half.

How to get the truly incremental line:

  • Cancel-flow saves. Show no offer to a small random share of customers who start cancelling, and see how many of them stay. That share would have stayed without the offer. The glossary entry on retention offers covers the offers themselves.
  • Recovered payments. Some customers would have updated their card on their own. Look at how many failed payments were paid without any email or retry, before you turned recovery on, and subtract that share.
  • Win-backs. A customer who comes back after cancelling is a reactivation, not a new customer. Count them on the retention side, and keep them out of CAC.

When should you spend on acquisition instead?

Retention spend has a ceiling: you cannot save more MRR than you are about to lose. If $9,000 of MRR churned last quarter, no retention budget keeps more than $9,000. Acquisition has no such ceiling. So:

  • Fix the cheap part first. Failed payments are often the cheapest to fix: there is no decision to reverse, only a payment to complete. Then cancel flows. Then win-back, which works on customers already gone.
  • Stop when the marginal save gets expensive. Bigger discounts and more calls save less each time. Once the cost per $1 kept reaches the cost per $1 of new MRR, more retention spend no longer beats acquisition.
  • Do not retain customers who were never a fit. If the customers who leave share a profile (a segment, a channel, a plan), the better fix is upstream: stop acquiring them.
  • Early on, acquisition usually wins. With 50 customers, even perfect retention keeps little revenue. The comparison matters more as the base grows.

What mistakes should you avoid?

  • Leaving salaries out of one side. Both sides must include people, or the side without them looks cheap.
  • Counting offers shown instead of offers redeemed, or counting only the first month of a repeating or forever discount.
  • Crediting every accepted offer as saved revenue, with no holdout to measure what would have stayed anyway.
  • Counting reactivations as new customers. It inflates acquisition and hides what win-back brings.
  • Comparing different periods. Spend in March can convert in May. Use a trailing quarter on both sides if your sales cycle is long.
  • Stopping at day 1. A save that churns the next month kept almost nothing. Compare MRR still active after the same delay on both sides.

How NoChurn helps

NoChurn records each failed payment as a recovery that ends recovered or lost, and each cancel flow session as saved or cancelled. That gives you the counts for the retention side of the table. It works with Stripe, Chargebee, Recurly, Paddle and other billing tools, or with any billing through the API.

FAQ

What is customer retention cost?

The money you spend over a period to keep current customers: retention tools, the share of support and success time spent saving accounts, discounts and free months redeemed, and win-back campaigns. Divide it by active customers for a cost per customer, or by MRR kept for a cost per dollar.

Is it really 5 to 25 times cheaper to retain a customer?

Not as a rule you can apply to your business. The figure, as quoted in Harvard Business Review in 2014, is a range that depends on the study and the industry. Measure your own cost per dollar of MRR kept, with a holdout group, and compare it with your cost per dollar of new MRR.

Should discounts count as a retention cost?

Yes, at their full cost: the revenue customers did not pay, for the whole length of the discount. A 50% discount for four months on a $100 plan costs $200. A forever discount keeps costing as long as the subscription runs.

Do win-back campaigns count as acquisition or retention?

Retention. The customer has paid you before. Count reactivations apart from new customers, or they lower your apparent CAC and hide what win-back brings.

Sources

All accessed 22 November 2026, except the Harvard Business Review article, checked through search results the same day.

PROMPT

Read https://www.nochurn.sh/llms.txt,
then follow https://docs.nochurn.sh/quickstart/
to add NoChurn to this app. Use a test key
(nc_test_…) until I say we go live.

Stop leaking revenue.

  • Free until $300 MRR
  • No card needed
  • Live in 10 minutes
  • No commission