Which Nights Should You Give Up?

Every article about night caps stops at the number. Ninety in London, thirty in Amsterdam, seventy in Copenhagen. Useful, as far as it goes — but if you let more than that, knowing the number is the easy half. The hard half arrives the moment you accept it: you are going to close part of your year, and nobody tells you which part.

That decision is worth real money, and most hosts make it the worst way possible.

The default answer is the expensive one

Faced with “stop at 90”, the two instincts are both bad.

The first is to do nothing and stop when you hit the limit. That means you give up whatever falls at the end — November, December, the first quiet weeks of next year if the window is a rolling one. It sounds harmless. It is the only strategy that guarantees you never choose at all; you simply let the calendar choose for you, in the order bookings happened to arrive.

The second is to close the nearest thing. You are over, you want it fixed today, so you block the next two months. Those are the months with deposits already taken and the highest chance of an enquiry. You have paid peak prices for a problem you could have solved in February.

A cap is not a ban. It is a budget. And a budget spent well is not the same as a budget spent first.

It is the same question your rate card already answers

If you set seasonal rates, you have already decided which parts of your year are worth more. That is all the information the decision needs.

Rank the year by what each night is actually worth to you, then work upward from the bottom until you are under the limit. The nights you give up are the ones you were least likely to sell well anyway — and the ones you keep are the ones that pay for the year.

Here is a one-bedroom listing on demo data. It is forecast to let 179 nights, so under a 90-night cap it must close 89 of them.

Example data from an Airflow demo account, not a real customer. Average nightly value by month for one listing. Orange is the 89 nights a 90-night cap closes when it works upward from the cheapest.

Month	Nightly value	Nights	Closed by the plan	Given up
January	95	31	yes	2,945
February	100	28	yes	2,800
March	115	31
April	140	30
May	165	31
June	190	30
July	230	31
August	235	31
September	185	30
October	150	31
November	110	30	yes	3,300
December	170	31
Total closed		89		9,045

Eighty-nine nights, taken from the bottom of the rate card, cost 9,045.

The same eighty-nine nights taken from the top — August, July and most of June — cost 19,545. Identical compliance. Identical number of nights. Ten and a half thousand difference, decided entirely by which end of the year you start from.

That is the whole argument, and it is why a night cap belongs beside your pricing rather than on a sticky note. If you have never looked at your year this way, Occupancy Was 78%. So Why Was It a Bad Year? is the companion piece — the same ranking, used for rates instead of limits.

Close runs, not scattered nights

There is a second mistake underneath the first, and it is subtler: closing the right number of nights in the wrong shape.

If you rank every night individually and close the cheapest 89 one at a time, you will end up with single closed days dotted through the shoulder season. Every one of those creates a one-night gap — and a one-night gap is unsellable under any minimum stay you are likely to have. You have not given up 89 nights. You have given up 89 nights plus every orphaned night either side of them.

So the nights have to come out in contiguous runs: whole weeks and whole months, not a sieve. Airflow plans them that way, grouping the closures into a handful of ranges you can look at and sanity-check in one glance. In testing, a five-bedroom listing that needed 183 nights closed got them back as three ranges — three decisions, not a hundred and eighty-three.

The honest part

Two things about this are worth saying plainly, because they change what the numbers above mean for you.

Availability caps and occupancy caps are different problems. Cape Town’s draft rule counts nights your property was offered. London, Amsterdam and Paris count nights it was let. Under an availability rule the calendar is the thing you manage. Under an occupancy rule the bookings are — and planning against bookings means planning against what might sell, not what did. We go into the distinction properly in A Night Cap Is an Accounting Problem, Not a Calendar Problem.

For a let-nights rule, the plan is deliberately cautious. It treats every open night as exposure, so the guarantee it gives you is “even if everything sold, you stay under”. If your January rarely sells anyway, that closes more nights than you strictly needed — you were never going to let them. The upside is that you cannot be surprised in November. The trade is yours to make, and it is why the feature proposes rather than acts.

And the figures in the table above are demo data for one invented listing. Your own ranking will look nothing like it. That is rather the point: the ranking has to be yours, built from your own seasonal rates and your own booking history, or it is just somebody else’s calendar.

What the plan needs from you

Three things, and it will tell you if any are missing:

Your seasonal rates. No rate card, no ranking. The year is just 365 identical nights and the plan has nothing to sort by.
Your booking history. Past bookings are what make the forecast real rather than theoretical. Import them before you need the answer. With a year or more of history, each night is weighed by its rate and by how often that month has actually sold, so the nights given up are the least valuable ones, not just the lowest-priced.
Every channel connected. This is the one that bites. A channel Airflow cannot see is a channel whose bookings are missing from the count, and a gauge fed by half your calendar will under-report. If your iCal feed is not connected, the number is optimistic and the plan is too small.

Bedroom counts matter too wherever the rule is measured in room-nights rather than nights — a room-night rule cannot be calculated at all until the listing says how many bedrooms it has.

Do it early or do not bother

One last thing, and it is the reason to act on this in the first quarter rather than the fourth: only future nights can change the outcome. A night already recorded as offered or let is counted for the rest of the window. There is no retrospective fix. A plan drawn up in February can close January’s equivalent next year; a plan drawn up in November can only close December, at December’s prices.

That asymmetry is the whole case for watching the number before it is a problem. The cheapest nights to give up are always the ones furthest away.

What to do this week

Create an account — start free; a card is needed at sign-up.
Import your booking history, by iCal link or CSV, so the forecast is built on your real pattern rather than today’s calendar.
Turn on the letting limit for the property and pick your city’s rule. It is four settings — what you count, per night or per room-night, over which window, up to what limit — and the presets fill them in.
Check your seasonal rates are current, because that is the rate card the plan ranks against.

Then ask it which nights to close. It will show you the ranges and what they cost, and it will not close anything until you say so. You can see where this sits in the wider product on the reporting and dashboard page.

Checked 2026-10-02: London’s 90 nights per calendar year is set by section 25 of the Greater London Council (General Powers) Act 1973, as amended by section 44 of the Deregulation Act 2015, and carries a second condition about council-tax liability — see the Greater London Authority’s guidance and GOV.UK. Rules change; check your own city before relying on a number in any article, including this one.

Start free →