Why do regulars stop coming back to Singapore cafés and restaurants?

Ask a Singapore operator why regulars leave and you get a list of things guests said. The guests who said something are a small minority: most simply stop appearing, and in a CBD unit with high footfall the absence is invisible unless something is counting visits per person.

Cause one — the routine broke. A large share of Singapore weekday loyalty is commute-shaped: the same MRT exit, the same 8:15am coffee, the same lunch window with the same colleagues. When the office moves floors, the team goes hybrid, or the line changes, the guest never decides to leave; the trigger that produced the visit stops firing. This is the most recoverable group because there is no grievance to overcome.

Cause two — the unreported bad visit. A slow order during the lunch crush, a drink made wrong, a queue that outlasted the break. Complaining is socially expensive and alternatives are one block away, so the guest quietly downgrades you from default to occasional. Recovery here needs acknowledgement, not a discount: a message that reads like a person noticing beats a voucher that reads like a mailing list.

Cause three — displacement. A new unit opened closer, or a delivery app made a competitor one tap easier. Nothing about you got worse; the cost of the alternative fell. What works is a reason specific to you — a saved usual order, a nearly complete reward, a held table at their usual time.

All three are legible in visit data before they are permanent. A guest who came every three days and is now at nine has told you something without saying anything. The recoverable window is roughly two of that guest's own cycles; inside it, a relevant message recovers a meaningful share (commonly quoted as 15–30%, which we treat as an illustrative range rather than a Singapore benchmark). Outside it, a substitute habit has formed. That is why cadence detection per guest — not a single 30-day rule — is the part worth automating.