A regular who visited your restaurant twice a week for six months just stopped coming. You didn't notice for three weeks. By then, they've found a new spot. They're not angry, they're not complaining. They're just gone. This is churn, and it happens silently in every restaurant.
The 7-14 day window
Research on customer retention shows that the most effective intervention happens within 7-14 days of a behavior change. After that, recovery rates drop significantly. A customer who missed their usual weekly visit is recoverable in week two. By week four, they've established a new routine that doesn't include you.
This means you need two things: the ability to detect the change, and a system to act on it fast. Both require knowing your customers individually, not just as aggregate transaction counts.
Why generic offers don't work
Sending "20% off your next visit" to all customers is wasteful. Your champions don't need it and might feel cheapened. Your dormant customers won't respond to it because 20% isn't enough to change a decision they've already made. Your at-risk customers might respond, but a generic offer feels like spam, not recognition.
Effective recovery is personal. "We noticed you haven't been in this week. Your next coffee is on us." That's specific to their behavior, acknowledges the relationship, and costs you AED 15 instead of giving away 20% on a full meal.
Three types of churn
Gradual fade: Frequency slowly declines over weeks. Weekly becomes biweekly becomes monthly becomes never. This is the most common and the hardest to notice without data. It's also the most recoverable if caught early.
Sudden stop: A regular just disappears. Often caused by a bad experience, a life change (moved, changed jobs), or a new competitor opening nearby. Recovery depends on the cause. If it was a bad experience and you reach out within a week, recovery rates are high.
Seasonal drift: Visit patterns shift with weather, holidays, or school schedules. This isn't true churn, but it looks the same in the data. Understanding seasonal patterns prevents panic and wasted recovery offers.
The cost of doing nothing
A regular worth AED 1,440/year walks away. You spend AED 100+ acquiring a replacement. The replacement has a 70% chance of never returning. The math is clear: investing AED 15-30 in recovering an at-risk customer is an order of magnitude cheaper than replacing them.
The first step isn't buying software or running a campaign. It's deciding that you want to know when a regular stops coming. Once you make that decision, the tools to act on it become obvious.