What is the optimal win-back voucher amount for an F&B venue?

There is a narrow band where win-back offers work. Below roughly 10% of ticket value, the offer does not clear the mental threshold that makes someone change their afternoon plans. Above roughly 20%, two bad things happen at once: the margin on the recovered visit disappears, and — more damaging — the customer learns that going quiet produces a discount. That is a habit you pay for repeatedly.

Percentages versus items. At identical cost, a named free item consistently outperforms a percentage discount by around 1.3–1.6× in conversion. The reason is that customers value the item at menu price and the discount at cash saved, and menu price is always the bigger number in their head. 'A free pastry with your next coffee' and '15% off your next order' can cost the venue the same $1.20; only one of them sounds like a gift.

Expiry. Fourteen days is the default that survives testing. Seven days is too tight for anyone whose normal cycle is fortnightly — you burn the offer on people who were going to come back anyway and miss the ones who needed the nudge. Thirty days removes urgency entirely; redemption clusters in the last 48 hours regardless of window length, so a longer window mostly just delays the outcome.

Uniform offers waste money. A customer who visited eleven times and then went quiet is worth a meaningfully larger gesture than someone who came once. The practical rule: tier the offer into three bands — a small item for one-or-two-visit lapsers, the standard 10–15% or free item for established regulars, and a personal, non-discount gesture (a reserved table, an off-menu item, a note from the owner) for the top decile, where a discount actually cheapens the relationship.

Targeting matters more than the amount. The most common mistake is blasting the offer to everyone who has not visited in 30 days. Thirty days means nothing without context: for a daily commuter it is a crisis, for a weekend brunch customer it is a normal gap. Trigger on deviation from each customer's own cadence, and the same voucher will convert two to three times better because it lands on people who genuinely drifted rather than people who were on schedule.

Measure recovered value, not redemption rate. A 25% offer will always redeem better than a 12% offer — that is not the question. The question is contribution margin from recovered customers over the following 90 days, minus the cost of every voucher issued, including the ones redeemed by customers who would have returned anyway. Run a holdout of 10% of the eligible list and the honest number appears within a quarter.