Airbnb Cancellation Policy Strategy: The Real Revenue Math

A three-night peak weekend at $280 a night cancels ten days out. Under Moderate, you are paid nothing. Under Firm, you are paid $420 and the calendar reopens. Most hosts read that gap and immediately switch to Firm, which is usually the wrong conclusion, because the $420 is only half of the calculation. The other half is how many bookings you never received because your terms scared the guest into a competitor’s listing.

Airbnb rebuilt its cancellation system on 1 October 2025 and migrated most listings without hosts doing anything. Nearly every guide published since then explains what each tier pays. Almost none of them explain how to decide, which requires two numbers you already own and probably have never calculated: your cancellation rate by lead time, and your rebooking probability on the dates that cancel. This article walks through both, with the arithmetic shown.

What Actually Changed on 1 October 2025

Three things changed at once, and the third one is the reason so much older advice is now wrong.

Strict was retired as a selectable policy. Listings that were on Strict were moved to Firm unless the host actively opted out before the deadline. New listings cannot choose Strict at all. If your dashboard still reads Strict today, you are a grandfathered exception, not the norm.

A new tier called Limited was introduced, sitting between Moderate and Firm with a 14-day full-refund window. It exists mainly to catch ex-Strict hosts who found Firm’s 30-day window too loose for their market.

And every standard policy now carries a universal 24-hour cancellation window. A guest who books at least seven days before check-in can cancel within 24 hours of booking for a full refund, and you cannot switch that off. It applies on top of Firm exactly as it applies on top of Flexible. Bookings made inside seven days of check-in do not get the window, which protects short-lead inventory.

The operational consequence of that third change is small but real. No reservation is fully committed on its first day. If you are the kind of operator who declines a competing inquiry the moment a booking lands, wait out the window first.

Airbnb’s stated reason for the restructure was revenue, not simplification. Its own figure is that hosts who moved from Strict to Firm earned roughly 10% more on average, and it has said that more than 40% of guests rate free cancellation among their top considerations when choosing a stay. Read that carefully. Airbnb is telling you that looser terms won more bookings than they lost to cancellations. That is a claim about demand elasticity, and whether it holds on your listing depends entirely on your market.

The Four Policies, Read as Revenue Instruments

Forget the names for a moment. Each policy is a bet on how far in advance you need certainty. The full-refund window is the length of that bet.

Flexible and Moderate

These are demand instruments. They buy you visibility and conversion at the cost of carrying essentially all the cancellation risk yourself. They are also the only two tiers on which Airbnb’s Reserve Now, Pay Later option appears, so choosing Firm removes your listing from a payment feature that exists specifically to convert hesitant bookers on higher-value trips.

Moderate is the sensible default for an urban listing with a short booking window and deep replacement demand. If a cancelled Tuesday refills in a day, the payout terms barely matter.

Limited and Firm

These are protection instruments. They are worth having when a cancelled night is genuinely hard to replace: event weekends, ski weeks, coastal peak summer, anywhere the guest who wanted those dates booked them four months ago and is not coming back.

Firm’s 30-day window is aggressive for most markets. If your average booking lands 25 days before arrival, a Firm policy means the majority of your bookings sit permanently inside the paid-cancellation zone, which sounds excellent until you notice how many guests filtered you out before booking at all. Limited is the more defensible middle for operators who want a buffer without paying for it in conversion.

Why the Payout Table Is Not the Decision

Every guide on this topic ends at the table above. That is where the analysis should start.

The payout table tells you what happens after a cancellation. It says nothing about how often cancellations happen, nothing about what you recover by reselling the night, and nothing about the bookings a tighter policy costs you upstream. A policy decision made on the payout table alone optimises for the wrong event: it treats the cancellation as the outcome, when the cancellation is one branch of a distribution.

The number that actually matters is expected revenue per booking, and it has three inputs:

• Cancellation probability by lead time. Not your overall cancellation rate. The rate within each window that matters: beyond 30 days, 14 to 30 days, 7 to 14 days, inside 7 days.
• Policy payout. What you keep when a cancellation lands in each of those bands.
• Recapture. The probability you resell the night, multiplied by the rate you resell it at, which is almost never the original rate.

Set against those three, a tighter policy has a fourth cost that is invisible in your own data: the bookings that never arrived. You cannot measure it directly. You can only bound it, which the worked example below does.

How to Measure Your Own Rebooking Probability

This is the input nobody supplies, because nobody can supply it for you. It is listing-specific and season-specific, and it is the difference between a policy chosen on evidence and one chosen on temperament.

Pulling the numbers from your own calendar

Export twelve months of reservation history. For every cancelled reservation, record the check-in date, the cancellation date, the lead time in days, the subtotal, and whether those nights were subsequently sold. Then answer two questions per cancellation:

• Did the night resell at all before check-in?
• If it resold, at what percentage of the original nightly rate?

Group the results into the four lead-time bands that map to the policy windows. On a portfolio with any volume, patterns show up quickly: weekday city nights resell at high rates and near-full price, peak weekends resell at low rates and only after a real discount. If you have fewer than about thirty cancellations across the year, pool two years or accept that you are working with a directional estimate rather than a measurement. Directional is still better than the payout table.

The recapture rate nobody calculates

Rebooking probability is only half of recapture. The other half is the price you get. A night freed up ten days before arrival goes back on sale into a thinner, more price-sensitive pool of shoppers, and the rate you achieve reflects that. In practice, recovered peak nights tend to sell below the rate they originally sold at, because the guests who valued them most bought them months ago.

So recapture value is the product of two fractions, not one:

Recapture = probability the night resells × achieved rate as a share of the original rate

A 40% rebooking probability at 85% of the original rate gives a recapture of 34%, not 40%. That distinction moves the policy decision more often than people expect. If you are not already tracking how recovered nights price against their original rate, that measurement belongs in the same workflow as your booking pace tracking, since both run off the same pickup data.

Worked Example: One Peak Weekend, Four Policies

Take a single listing. Three-night peak weekend, ADR $280, booking subtotal $840. The guest cancels ten days before check-in. From twelve months of history, this host has measured a 40% rebooking probability at that lead time on peak dates, at 85% of the original rate.

Recapture value = 0.40 × $840 × 0.85 = $285.60

Limited and Firm are worth $420 more than Moderate on this single cancellation. Now annualise it honestly.

Say the listing takes 60 bookings a year and 15% of them cancel, so nine cancellations. Suppose three of those nine land in the 7 to 30 day band where Firm and Limited pay 50% and Moderate pays nothing. The annual protection value is 3 × $420 = $1,260.

That is the entire upside. Against it, the average booking on this listing is worth $840. If moving from Moderate to Firm costs you more than one and a half bookings across the year, the tighter policy is a net loss. On 60 bookings, one and a half is a 2.5% conversion drag, which is well inside the range a visible policy change can produce.

This is why Airbnb’s 10% figure for Strict-to-Firm migrants is credible rather than surprising. Protection value is bounded and small. Conversion effects are unbounded and compound across every search impression. Tightening terms is a good trade only when the protection is large, which means only on dates that genuinely do not refill.

The late cancellation that pays twice

There is one case where the tighter policies are unambiguously strong, and almost nobody states it plainly.

Under Limited and Firm, a cancellation inside seven days of check-in pays you 100% of the booking. The calendar also reopens, so those nights go back on sale. If you resell even one of the three nights at a discounted $200, you finish at $1,040 on a reservation that was worth $840. The late cancellation is revenue-positive.

Under Moderate, the same cancellation four days out pays you one night plus 50% of the remainder, roughly $560, and you still get the resale upside. Under Flexible it pays nothing. The spread between tiers is at its widest precisely in the window where recapture is hardest, which is the structural argument for protecting peak dates and nothing else.

Non-Refundable Rates: Price the Option, Do Not Just Offer It

Airbnb lets you offer a non-refundable rate alongside your standard one. The guest takes a discount, typically 10%, and gives up refund rights entirely. Airbnb has said hosts in its pilot earned about 5% more on average after adding the option.

Treat that 10% as what it is: the price you are paying to eliminate refund risk. Whether it is a good price depends on how much refund risk you actually carry, and that is calculable.

Under Firm, your expected refund leakage per booking is the cancellation probability multiplied by the average share you refund across the timing distribution. If 12% of bookings cancel and you refund an average of 60% of the subtotal across those cancellations, your leakage is 0.12 × 0.60, or 7.2% of revenue.

Paying a flat 10% discount to remove 7.2% of leakage is a losing trade on that arithmetic alone. It only turns positive if the non-refundable rate brings bookings you would not otherwise have won, which is exactly what Airbnb’s 5% figure implies, or if cash-flow certainty is worth something to you independently. Both can be true. Neither is automatic.

The practical version: run the non-refundable option on dates where your leakage calculation exceeds the discount, which is almost always peak and event dates on a loose base policy. On a listing already running Firm through peak season, adding a 10% non-refundable rate on the same dates is usually paying twice for the same protection. Getting the interaction right between rate structure and cancellation terms is the same problem as getting rate structure right in general, which is why it belongs inside your dynamic pricing strategy rather than being managed as a separate setting.

The same logic transfers to Booking.com, where non-refundable rate plans work as a distribution lever as much as a risk one, and where a single rate plan makes you invisible to filtered searches. If Booking.com is a meaningful share of your mix, that structure deserves its own review alongside your Booking.com listing setup.

Seasonal Policies: Stop Running One Setting All Year

Airbnb now supports date-specific cancellation policies, set through custom settings on the desktop calendar. Select the dates, open custom settings, choose the policy, save. It rolled out from Airbnb’s late-2025 release and has been reaching markets progressively since.

This removes the compromise that made the whole decision hard. You no longer have to pick one setting that is too tight for February and too loose for July. Two rules govern how it behaves in practice. A booking that spans two policy zones takes the policy attached to the check-in date, and existing bookings keep whatever policy applied when they were made. Changing a policy does not reach backwards.

A calendar that works

Set peak-date policies 60 to 90 days before the booking window opens for those dates, not after. A policy applied once the reservations are already on the books protects nothing.

• Low season and shoulder: Moderate, or Flexible if you are still building review volume. Conversion is the constraint, not risk. This is the same period where slow season pricing should be doing most of the work.
• Standard peak weekends: Limited. The 14-day window catches the cancellations that are hardest to refill without excluding you from Reserve Now, Pay Later for most of the year.
• Event dates and holiday weeks: Firm, set well ahead of the booking window, optionally with a non-refundable rate alongside.
• Gap and orphan nights: keep these loose regardless of season. A one-night gap that cancels is not a protection problem, it is a pricing problem.

A new listing is a separate case. Until you have review volume, conversion beats protection on every date including peak ones, which is part of the wider argument for running loose on everything during a listing’s first few months.

What a Host Cancellation Actually Costs

Everything above concerns guest cancellations. Host cancellations run on a different and much more expensive set of rules, published in Airbnb’s Host Cancellation Policy.

A $50 minimum applies. The fee is calculated on base rate plus cleaning fee plus any pet fees, excluding taxes and guest fees, and it is withheld from your next payouts. You also forfeit the payout for the cancelled reservation itself.

The fee is rarely the worst part. Airbnb blocks the listing’s calendar for the affected dates, so you cannot resell nights you just freed. Host cancellations count against Superhost eligibility, and repeated avoidable cancellations put the listing or the account at risk. Waivers exist for documented emergencies and events covered under Airbnb’s disruption policies, and Airbnb decides those case by case on evidence.

The full terms, including how a host can be found responsible for a cancellation they did not initiate, such as a double-booking or a materially inaccurate listing, are set out in Airbnb’s cancellation policy documentation. In revenue terms the conclusion is simple. A host cancellation is the single most expensive event in the system, and calendar hygiene across channels is worth more than any policy tier you can select.

The 30-Minute Cancellation Policy Audit

Work through this once a quarter, or immediately if you have not looked at your policy since September 2025.

• Check what policy each listing is actually on today. If any still says Strict, you are grandfathered and should decide deliberately whether to stay there.
• Export twelve months of reservations and calculate the cancellation rate within each lead-time band: beyond 30 days, 14 to 30 days, 7 to 14 days, inside 7 days.
• For every cancelled reservation, check whether the nights resold and at what percentage of the original rate. Compute recapture as probability multiplied by achieved rate share.
• Split the result between peak and non-peak dates. They will not match, and that difference is the case for seasonal policies.
• Run the expected-value comparison on a typical booking, exactly as in the worked example above.
• Convert the annual protection value into bookings. If the tighter policy has to cost you fewer than two bookings a year to break even, treat it with suspicion.
• Compare your refund leakage percentage against the non-refundable discount before adding one. If leakage is below the discount, skip it.
• Set peak-date policies 60 to 90 days before those dates open for booking.
• Confirm calendar sync across every channel. The cheapest cancellation is the host cancellation that never happens.
• Diary the next review for the quarter before your next peak season, not during it.

If you are also running Vrbo or Booking.com, set each channel’s terms on its own merits. The tiers are not equivalent across platforms and inheriting Airbnb’s structure elsewhere is a guess, not a decision. Where those channels represent real volume, the policy question is part of the broader revenue management work rather than a settings task.

Frequently Asked Questions

Conclusion

The cancellation policy question has a clean answer, and it is not a tier name. Measure how often your bookings cancel inside each lead-time band, measure how much of a cancelled night you actually recover once you account for the discount it takes to resell it, and compare the annual protection value against the bookings a tighter policy costs you. On most listings the protection turns out to be worth one or two bookings a year, which is why the loose tiers win more often than instinct suggests, and why the tight tiers should be aimed at specific dates rather than the whole calendar.

The one case where tight terms are clearly worth it is the peak weekend that cancels inside seven days. Under Limited or Firm you keep the full booking and the nights go back on sale, and that is the only branch of the distribution where a policy tier pays for itself outright.

If your policy has not been reviewed since the October 2025 restructure, it is a migration default rather than a decision. Get in touch and we will run the numbers on your portfolio and set policy by date band before your next peak season opens for booking.