Every restaurant has a Sarah. Most of them lose her without ever knowing her name.
She has booked four times in three months. Always the corner table. Mildly gluten-intolerant, not allergic, she just prefers to avoid it. A New Zealand Sauvignon Blanc every visit, no prompting. On her last booking the steak came back slightly overdone. She said nothing. She left half of it and asked for the bill.
Your server clocked all of it. Your system kept none of it.
So Sarah doesn't rebook, and nobody hears the silence. Ninety days later a report quietly reclassifies her as "lapsed," and you spend about £100 acquiring a stranger to fill the seat she used to choose. You didn't lose Sarah on the night the steak was wrong. You lost her the morning nobody remembered it had been.
Run the other version. Your server speaks a five-second note into a handheld after clearing the table: steak returned, half eaten, no complaint raised. The system flags it overnight. Next morning a manager sends two lines: "Hi Sarah, I noticed the ribeye wasn't quite right last time. The next one's on me." She rebooks within the week. She tells two friends why.
“Same guest. Same steak. Two different endings, and the only variable is whether anyone wrote it down.
The operators proving it aren't theorising.
In Good Company grew its actionable guest database by 400%, not through a platform migration, but by training staff to tag preferences consistently. Notice what neither of them did. They didn't buy a bigger CRM. They started treating what their people already see as an asset worth keeping.
That intelligence is sitting in your building tonight: the waiter who remembers the wine, the host who reads the room, the manager who can feel a regular cooling off. Most of it clocks out at close and never comes back.
So before you sign off the next acquisition campaign, answer a smaller question first: how many Sarahs left last month? In most restaurants, nobody can name a single one. That, not the ad budget, is the problem worth fixing.
