What does losing one regular customer actually cost?

Losing a regular never feels like an event, which is why it is never costed. A single missing Tuesday coffee is a $4 non-event. The cost only becomes visible when you extend it over the period the customer would have kept coming, and then add the two things that go with the loss.

Start with the direct number. A customer visiting weekly at a $4 average ticket produces about $208 a year in revenue. At a typical 60% gross margin that is roughly $125 of contribution — real money that pays rent and wages. A twice-weekly customer is double that; a $12-ticket lunch regular visiting three times a week is over $1,800 a year in revenue.

Add replacement. Losing them does not simply remove $208; it obliges you to acquire a substitute, and acquisition costs four to seven times recovery. In practice, replacing that customer costs somewhere between $8 and $25 in marketing and introductory discounts — and there is a 70–80% chance the replacement never becomes a regular at all, so the true expected cost of replacement is several attempts, not one.

Add the referral chain. Regulars bring colleagues, hold meetings at your tables and answer the 'where should we go' question in your favour. This value is invisible in every reporting system and is not small: in most venues a meaningful share of new customers arrive because an existing regular brought them. Losing the regular removes the node, not just the transactions.

Aggregate it. A small cafe losing twenty regulars over a year — which is an unremarkable, entirely unnoticed rate — has lost roughly $4,000–8,000 in direct revenue plus several thousand more in replacement spend and forgone referrals. That is a five-figure hole in a small business, and nothing in the P&L flagged it, because acquisition kept the top line flat.

Now the other side of the ledger. Recovering a drifting regular costs a message and, at most, a named item worth a couple of dollars. Even at a modest recovery rate, the arithmetic is not close: intervening on twenty at-risk regulars costs less than replacing three of them. That asymmetry — cheap to keep, expensive to replace — is the entire economic case for cadence detection, and it is why 'we're too small for retention software' usually means 'we have not costed the losses'.