Most referral programs are run on intuition. "Let's offer $10 to the referrer and $10 to the new customer." Two months later, the CFO runs the numbers and shuts it down because the unit economics are upside down.
This post is the napkin math for designing a referral program that actually works. If you remember three numbers you will never blow up your economics again.
The three numbers
- CAC — your current blended customer acquisition cost. What does a new customer cost you today through ads, promotions, etc.?
- LTV — the average revenue a customer generates for you over their lifetime.
- Margin — what fraction of LTV is actual profit after costs.
A referral program has two payouts: the referrer bonus and the referee bonus. The total combined payout must be less than LTV × margin. That's it. That's the constraint.
A worked example
Say you run a coffee shop. Your average customer visits 18 times a year, spends $6 per visit, and you have a 65% gross margin on coffee.
- LTV per year: 18 × $6 = $108
- Margin: $108 × 0.65 = $70.20
So the absolute maximum combined payout on a successful referral is $70.20. Above that, you're losing money on growth.
But you don't want to spend your whole margin on acquisition. A sane rule of thumb is: spend at most 30% of margin on referral incentives. That gives you:
- Max combined referral payout: $70.20 × 0.30 = $21.06
Split that how you want. Classic 50/50: $10 to referrer, $10 to referee. Or weighted towards the referrer: $15 to referrer, $5 to referee. The point is: you now have a number that tells you when you're overspending.
Why the 5× multiplier works
In Pounds, Pro members earn 5× points on referrals. You might think this breaks the math — but it doesn't, because the points themselves cost the shop almost nothing to issue; they only cost when they're redeemed.
The real cost of a 5× multiplier is the redemption rate. If a referred customer redeems $10 worth of rewards, the shop pays $10. But because the multiplier creates stronger referral behavior, the number of referrals per customer typically goes up 2.5–3×.
Net effect: you pay more per successful referral, but you get many more of them, and the LTV ÷ referral-cost ratio improves.
The spreadsheet exercise
Before you launch a program, sit down with a spreadsheet and model these three scenarios:
- 10% of customers refer one person each (typical free program)
- 20% of customers refer 2 people each (typical with a 2–3× multiplier)
- 30% of customers refer 3 people each (typical with a 5× multiplier + social proof)
Compute cost per successful referral at each scenario. Compute total new customers at each scenario. You'll find that #3 almost always has the best unit economics — which is exactly why we built Pounds Pro around it.
One last thing
The single biggest mistake we see is programs that measure "referrals sent" instead of "referrals converted". They are not the same number. Always measure on completed first bookings, not on codes handed out.