The Regular You Would Rather Not Have

He is in three times a week. He spends reasonably. He also occupies a member of staff for most of the night, complains about something every visit, and two tables have moved away from him this month.

The till says he is a good customer.

What the till cannot see

Three costs, none of which appear against his name:

Staff time. If one person is effectively assigned to him, that is a large share of a small team, and it is being paid for out of everyone else's service.

Adjacent guests. ⚠️ The tables that move, shorten their visit, or do not return are the largest cost and the completely invisible one. They do not complain; they just leave.

The team. People do not stay in jobs where one guest makes every shift worse. The retention cost lands months later and is never connected back.

Working out what he is actually worth

Not precisely — directionally, which is enough.

His annual spend. The number people already know.

Staff hours consumed. Rough. If it is an hour a visit, three visits a week, that is a meaningful fraction of a wage.

Adjacent effects. ⚠️ Ask the floor: "who moves away, and how often?" They know exactly, and nobody has asked.

In many cases the arithmetic reverses, and the reversal is uncomfortable precisely because he is friendly, familiar, and has been coming for years.

The options between tolerating and refusing

The debate is usually framed as keep or ban. There is a large middle:

Change who serves him. Rotate deliberately, so the cost is shared and no single person carries it. ⚠️ This alone resolves a lot of cases, because much of the problem is one person's shift being ruined rather than the guest's behaviour.

Change where he sits. Distance from other tables removes the adjacent cost without removing him.

Set a specific limit. Not a general disapproval — one concrete boundary, stated once, calmly. Frequently more effective than anyone expects, because nobody has ever actually said it.

Change what is comped. If part of the cost is habitual complaint-and-comp, ending that quietly changes the incentive.

The conversation

If it comes to it, one principle: behaviour, not character.

"We cannot have shouting in the room" is actionable and defensible. "You are difficult" is neither, and it turns a specific issue into a personal one that ends badly.

⚠️ What you may do — refusing service, asking someone to leave, banning — is governed by law and by your licence, and getting the process wrong creates a different and larger problem. Confirm the ground with your counsel before you need it, not during.

The trap on the other side

Do not do this to anyone who is merely demanding.

A guest who asks for a lot, spends a lot, and is a bit tiring is a good customer being a bit tiring. ⚠️ The test is the adjacent cost and the staff cost, not how much anyone enjoys serving them — otherwise this becomes a mechanism for removing guests the team finds inconvenient, which is a different and worse thing.

Three to hold

Staff hours consumed per visit, roughly. The number nobody estimates.

Adjacent guests affected, per month. Ask the floor; they know.

Complaints and comps against the same guest, rolling. The pattern that distinguishes a bad night from a method.

Where the record has to sit

If incidents live in memory and spend lives in the till, one side of the arithmetic is always missing — and the answer defaults to the side that is written down, which is spend.

tasteck keeps notes and visits against the guest, so a pattern across months is visible rather than something three people half-remember differently.

He is a good customer by the only number anyone measures. That is not the same as being worth it.