How to tell which clients are quietly leaving

Clients do not leave salons. They just do not come back, and there is no moment where that becomes visible. There is no cancellation, no complaint, no notice. Someone who came every four weeks for three years simply is not in the book in May, and by the time you notice in July they have a new routine somewhere else.

The reason this goes unnoticed is arithmetic, not attentiveness. A busy salon might see forty people a day. Nobody can hold four hundred individual visit rhythms in their head and notice which ones broke.

Why "six weeks" is the wrong rule

The common approach is a fixed window: flag anyone who has not been in for six or eight weeks. It is easy and it is mostly noise, because it is measuring against an average nobody actually follows.

A barber client on a two-week cycle is a month late by week six, well past the point where they have found someone else. A colour client on a ten-week cycle is perfectly on schedule at week six, and messaging them reads as either desperate or as though you do not know them. The same rule is simultaneously far too late for one and wrong for the other.

Measure each client against their own rhythm

The useful calculation is per client, not per salon:

• Take their own average interval. The mean gap between their visits, from their actual history. Ten weeks for the colour client, two for the barber client.
• Flag at a percentage over it, not a fixed number of days. Twenty-five percent past their own average is a reasonable starting point. That is roughly two and a half weeks late for the colour client and three or four days late for the barber client, which is correct in both cases.
• Require at least two visits before calculating anything. One visit gives you no interval at all, just a date. This matters more than it sounds like it does, and it is the subject of the next section.
• Sort by most overdue, and stop at ten. A list of ninety names is a list nobody works. The point is the conversation you will actually have this week.

The gap this method leaves

Interval maths structurally cannot see your worst problem: clients who came once and never returned. They have no interval, so they never appear on any overdue list, and they are usually the largest leak in the business.

They need separate, cruder handling: anyone with exactly one visit and no second booking after some flat window, say forty-five days. No calculation, just a count. It is worth keeping this as its own list, because the two groups need different conversations. A regular who has drifted knows you. A first-timer who did not come back does not.

What to actually send

The message that works is short, specific and does not mention a discount. "Hi Sam, noticed it has been about seven weeks, usually see you around five. Want me to hold your usual Thursday?" It works because it is obviously from a person who knows them, it names a real gap, and it ends with one thing to say yes to.

Discounting as the opening move trains regulars to wait for the offer, and converts a relationship problem into a price problem. Keep the discount in reserve for people who do not answer the first message.

None of this requires software to understand. It does require somewhere that records each visit, so the intervals exist to be calculated at all, which is the part a paper book cannot do.