The Regular You Lost Without Noticing

A gym member cancels. A subscriber unsubscribes. A phone contract terminates. In each case a moment exists where the business is told.

Night venues get no such moment. A regular who came every second Thursday for a year simply stops coming. There is no event, no notification, no line in a report. There is only an absence — and absences are invisible until someone goes looking for them.

By the time the room notices, the loss is months old and the guest has a new place.

Why "lapsed after 90 days" is usually wrong

Most systems that track this at all use a fixed cutoff — often ninety days. It is a reasonable default and it is wrong for almost every specific room, for a simple reason:

Guests set their own interval, and the intervals differ by an order of magnitude.

A weekly regular      …absent 3 weeks is already unusual
A monthly regular     …absent 3 weeks is normal
A quarterly entertainer of clients …absent 3 months is normal

A single 90-day line treats all three identically. It calls the weekly regular healthy for two months after they actually left, and it calls the quarterly guest lapsed while they are still perfectly on schedule.

Both errors cost money, in opposite directions. The first loses you the guest. The second wastes your contact on someone who was coming anyway — and worse, it makes your win-back campaigns look effective when they were only catching people who had never gone.

The number that fixes it: each guest's own interval

Instead of one line for everyone, compute the interval per guest.

For each returning guest:
  intervals = the gaps in days between their consecutive visits
  typical   = the median of their own intervals
  overdue   = days since last visit ÷ typical

overdue is the number to sort by. A guest at 1.0 is exactly on schedule. At 2.0 they have gone twice as long as they ever normally do. At 3.0, something has changed.

This works because it compares each guest against themselves rather than against a room-wide average. The weekly regular and the quarterly guest both hit 2.0 at the point that is unusual for them.

It needs at least three visits

A guest with one visit has no interval. A guest with two has one interval, which could be anything. Three visits gives you two intervals and a median that means something.

So the population splits naturally, and the split is useful:

1 visit   …not a regular yet. A different problem — this is conversion, not retention
2 visits  …forming. Watch the second gap; it predicts whether a third comes
3+ visits …a regular with a rhythm. overdue is computable and meaningful

Do not mix these three groups in one number. A "retention rate" that includes first-timers is mostly measuring acquisition volume.

What to do with an overdue regular

The temptation is a discount. Resist it for one cycle and try the cheaper thing first.

A regular who has drifted is usually not price-sensitive — they are absent. Something changed: their schedule, their group, the staff member they came for, or a single bad night nobody logged. A discount answers a question they did not ask.

What tends to work better is specific and small:

• The staff member they usually saw, contacting them directly rather than the venue contacting them
• A concrete reason with a date attached — a night, an event, a return, something that is actually happening
• If they have an unfinished bottle in keep, the bottle itself is the reason, and it needs no offer at all

Measure whichever you choose:

Win-back rate = overdue guests who returned within one interval
                ÷ overdue guests contacted

And keep a holdout. Leave a random slice of overdue guests uncontacted for one cycle. Without it you cannot tell your campaign from the guests who were going to come back anyway — and in this business a meaningful share always do.

The concentration risk nobody prices

While computing intervals, compute one more thing:

Concentration = share of last 12 months' revenue from the top N guests

In a relationship-driven room this number is often much higher than the operator expects. That is not automatically bad — high-value regulars are the business model — but it is a risk with a size, and knowing the size lets you decide deliberately.

The version that matters more:

Staff concentration = share of revenue from guests who come for one specific person

If a large share of your revenue is attached to three staff members, a resignation is a revenue event, not just an HR one. Knowing this in advance is the difference between planning and reacting.

What this needs from your records

Everything above comes from one table:

guest_id, visit_date, spend, served_by

Four columns. No system required to start — a spreadsheet exported from your POS or booking log will produce every number in this guide.

The one prerequisite is a guest identifier that survives across visits. In a cash-heavy room that means a phone number captured at booking, consistently. Without it there are no intervals, because there is no way to know that tonight's guest is the same person as last month's.

The system we built

tasteck is a booking and analytics system for night venues, and the guest record is what it is built around rather than a module attached later.

What it computes: guest records that persist across visits, repeat interval, new-versus-returning share, cohort behaviour, revenue by acquisition channel through to lifetime value, staff scheduling and shift records, dispatch and driver status for rooms that run cars, and a reservation screen wired to inbound calls so the identifier is captured at the moment the phone rings.

One honest note. Our customer analytics have used ninety days as the boundary for an active guest, and we are in the middle of making that configurable per venue — for exactly the reason written at the top of this guide. The right window is the one that matches your room, not the one we picked.

Built by operators

tasteck was built by people who ran venues in this industry for sixteen years and grew the business from ¥200 million to ¥1.2 billion a year — six-fold, by attacking it with systems rather than by selling harder. Every competitor in this space is a software company's interpretation of the business. This one is the business's own, which is why the guest record sits at the centre instead of the till.

The output no one else produces

Retention protects the guests you have. The other half is what you pay to get more, and there tasteck outputs something nothing else in the nightlife category does:

The maximum you can spend on each marketing channel next month, in your currency.

Not a chart — an amount, per channel, that you take into the renewal conversation. It comes from the lifetime value of the guests each source actually delivered, with your target margin applied.

And it connects directly to intervals. Lifetime value is built out of repeat behaviour, so the moment you can measure intervals properly, the channel ceiling becomes accurate too. The same guest record answers both questions.

Ask it from ChatGPT

tasteck connects to ChatGPT over MCP: ask your numbers as a question and the answer comes back in the chat — who is overdue, which sources produce repeat guests, how concentrated last year's revenue was.

Measured against every vendor listed on Japan's principal nightlife-industry directory, this is the first implementation of it in the category, and the same interface is callable from anywhere rather than being tied to one assistant.

Multi-language is built in, the operating surface itself, with your language set put in place during onboarding.

From $34 a month for up to two venues. Thirty days free on every plan, cancel any time. → Pricing

Start this week

1. Export guest_id, visit_date, spend, served_by for the last twelve months.
2. Compute each guest's own median interval, using only guests with three or more visits.
3. Sort by overdue (days since last visit ÷ their own interval). Look at everyone above 2.0.
4. Contact a slice, hold back a slice. Without the holdout you will not know what your campaign did.
5. Compute staff concentration once. If the answer is uncomfortable, it is better known than not.

The channel-ceiling calculation is open on our site with nothing to sign up for: Ad budget calculator. Nothing is transmitted anywhere — use it and close the tab.

Read next

• Guest CRM for Rooms Where Everyone Pays Cash
• No-Shows: The Money That Never Walks In
• Four Questions Your POS Cannot Answer

---

On benchmarks. No target figures for repeat interval, win-back rate, or concentration appear here. We do not have a dataset broad enough to publish them, and they vary enormously by format and price band. The whole point of the method is that it uses each guest's own history rather than someone else's average.