Last-Minute Pricing in Shoulder Season Versus Peak

The same empty night means two different things in April and in August. Running one playbook across both is how shoulder season gets written off before it starts.

Last updated: September 26, 2026

"Last minute" is a phrase, not a fixed number of days. In peak season it might mean the final three nights before arrival. In shoulder season it stretches out to a few weeks. The reason is simple: what's left unsold in each season is a different kind of inventory, and the two need different treatment rather than a shared discount.

Key Takeaways

• The window changes length. Peak leaves days unsold; shoulder leaves weeks.
• A shoulder night rarely comes back. Thin demand means an unsold night is a lost night, not a delayed one.
• A peak night usually sells. So a late cut mostly teaches next season's guests to wait.
• Use non-price levers first in peak. Release the dates, shorten the minimum stay, widen the arrival window.
• Set both ladders once. Decide the season's rules before the season, not during it.

Why the last-minute window is a different length in every season

Demand arrives at different speeds depending on the season, and the window is simply where that demand runs out. In peak, the calendar around your dates fills early because there are more guests chasing fewer beds. By the time you're a week out, what's left is usually a genuine gap rather than a price problem. In shoulder, the same date a month out can still be wide open across the whole neighbourhood.

That difference matters because it changes what an unsold night means. A night unsold in peak is often a night that was never going to sell — the wrong arrival day, a gap between two bookings, a date that fell outside someone's flight schedule. A night unsold in shoulder is a night that could have sold, and would have, if it had been priced for the demand that actually exists.

Hosts who use one rule all year end up with the worst of both. They discount too late in shoulder, when the booking window has already closed. They discount too early in peak, when the dates would have filled on their own. The fix isn't a better single rule. It's two rules, one per season, each written before the season starts.

Shoulder season: what an unsold night actually costs you

In shoulder season, an unsold night is close to a permanent loss. Demand is thin enough that a date which goes unbooked this week is unlikely to be rescued next week, because there simply aren't many guests looking. That changes the arithmetic in a useful way: recovering something beats protecting a rate you won't get.

So the shoulder lever is time, not depth. Release the dates earlier, rather than waiting until they're a week out, because the guests who book shoulder season are usually planning further ahead than peak guests. A modest cut applied three weeks out does more than a steep one applied three days out, and it leaves your peak rate untouched.

Widening the stay is the second lever. Shoulder guests often have more flexibility about arrival days, so opening up the midweek nights around a weekend can turn a two-night booking into a four-night one without any price movement at all. The nights you're trying to recover are usually the awkward ones, and awkward nights respond better to flexibility than to discounts.

Keep one boundary in view while you do this. A shoulder rate that stays live into peak is no longer a shoulder rate. Date-range rules exist so the low-season ladder switches off on a date you set in advance, rather than staying on because nobody remembered to turn it off.

Peak: selling the final nights without training guests to wait

Peak presents the opposite problem. The dates have value, the demand is thick, and a late discount works — which is exactly why it's dangerous. Every guest who learns that your final week always comes with a cut learns to wait for it, and that lesson survives into next season and the one after.

That's the trap. You get the night sold and you pay for it with the booking behaviour of every future guest who watched you do it. The cost doesn't appear in this month's numbers. It appears as a booking curve that arrives later every year, which is the same thing as a calendar that looks empty for longer each spring.

So in peak, lead with levers that aren't price. Release inventory that was being held back, if you were holding any. Shorten the minimum stay for the final dates so a two-night booking can take them. Widen the arrival window so a guest can start midweek rather than only on a Friday. And open the final nights to a channel you'd normally keep them off, if that's a lever your setup allows.

When you do move the rate, move it in a step rather than a slide, and move it once. A single, dated adjustment at a defined point reads as a yield decision. A rate that creeps down every day for a fortnight reads as a sale, and sales teach waiting. One more check while the final nights are live: make sure the rate you set is the rate the channel is actually advertising, and that the listing is bookable rather than paused. Where your calendar shows the source price and status for each connected channel in one place, that check takes a glance. The same view is open at localsbnb.com.

Setting both ladders once, so you are not deciding nightly

A ladder is just a set of rules tied to dates, and building two of them takes an afternoon. The point isn't precision. It's that the decision gets made in advance, when you're calm about it, instead of at nine at night when a gap has just appeared.

Start with the season boundaries. Pick the dates where shoulder becomes peak and peak becomes shoulder again, and write them down. If you have a few years of bookings, those boundaries are already in the data: the week when lead times shorten sharply and the calendar stops having holes is the week peak starts.

Then define the bands inside each season. For shoulder, that's usually a lead-time ladder: full rate until a set point, then a smaller step, then the floor. For peak, that's a length-of-stay ladder: the minimum stay that applies at the start of the season, and the shorter one that applies to the final nights. Keep each band tied to a date range so it switches without you.

The last piece is the floor. Both ladders should stop at the same number, and that number should come from your cost structure rather than from what the calendar looks like on a slow Tuesday. Once the floor is written down, the nightly decision mostly disappears — you're no longer choosing a price, you're reading one off a rule you already agreed with yourself.

FAQ

How do I know where shoulder season ends and peak begins?

Look at when your lead times collapse. During shoulder, bookings arrive weeks ahead; in peak, the same dates sell much closer to arrival or not at all. The crossover week is your boundary, and it's usually consistent from year to year.

Isn't it better to sell the last peak night at any price?

Not at any price. A final-night booking at your floor is a reasonable trade if the alternative is an empty night. A habit of cutting every year at the same point isn't a trade, because the cost lands on next season's booking pattern rather than on this month's.

Can one rate plan do both ladders?

Technically yes, and practically no. Two ladders have different triggers — one watches lead time, the other watches length of stay — and folding them into a single rule makes both harder to reason about when a season goes sideways. Keep them separate.

Two ladders, one for each season, do the work that a nightly decision can't. Write the season boundaries down, keep the floor the same on both, and check that the last-minute rate you set is the one actually live on each channel. LOCALSBNB runs at $4.5 per room per month on annual billing and $7 month-to-month. It holds every connected channel's rates on one calendar, so both ladders can be read from the same place. Details are at localsbnb.com.

---

This article is general guidance for hosts and isn't financial or platform policy advice; demand patterns, rates and platform rules vary by market and season, and the current terms of each platform prevail.