All articles

· 6 min read

Do Blocked Dates Change Your 7-Day Average?

Owner blocks, maintenance holds and unbooked gaps are not periods of customer use. What the 7-day average measures and what it ignores.

An open monthly planner

General information based on our understanding of the rules, not tax advice, and not a guarantee of accuracy. We are not CPAs. Speak to your own tax professional before acting on it.

Generally, no. The average period of customer use in Treas. Reg. §1.469-1T(e)(3) counts days a paying customer had the property, divided by the number of those rental periods, and a night you held for yourself, for a contractor, or for nothing at all is not customer use. It therefore enters neither side of the fraction. That cuts both ways: blocking dates cannot rescue an average that long bookings have pushed past seven days, and a half-empty calendar does not damage an average that short stays have kept low.

What the regulation measures

The short-term rental exception most owners are relying on sits in Treas. Reg. §1.469-1T(e)(3)(ii)(A): an activity is generally not a rental activity if the average period of customer use of the property is seven days or less. The computation is described at §1.469-1T(e)(3)(iii), and it is a ratio of two counts taken over the tax year:

Average period of customer use = total days of customer use ÷ number of periods of customer use

Both counts are about customers. The denominator is the number of rental periods, not the number of nights on the calendar, not the number of weeks in the year, and not the number of nights the listing was available. Nothing in the computation references availability, occupancy rate, or whether the property sat idle. IRS Publication 925 describes the rental-activity rules in general terms and does not add an availability concept either.

This is why the test behaves unlike the metrics a host is used to watching. Occupancy is a business measure with a denominator of available nights. The average period of customer use has a denominator of stays. The two numbers move independently, and improving one does nothing to the other. The seven-day average stay guide works through the definition in more depth.

Customer use, not calendar occupancy

The practical rule that falls out of the definition: if no customer had the property, there is nothing to count.

Nights that are generally not customer use include:

  • Nights the owner, family or friends stayed without paying a market rent
  • Nights blocked for cleaning, turnover or a maintenance window
  • Nights blocked while the property was between seasons or taken off the market
  • Nights that were simply never booked

Nights that generally are customer use include every paid stay, whatever its source: platform bookings, direct bookings, repeat guests who paid by transfer, an insurance or relocation placement, a stay arranged through a local agency. Completeness matters more than the platform the booking came through, because a stay left out of the record is a stay left out of both counts.

One caution about the first list. Personal-use nights do not enter this calculation, but they are not irrelevant to your return. Personal use has its own consequences under other provisions, including the vacation-home rules, and heavy personal use can matter even when the seven-day average is comfortable. Excluding a night here is not the same as it having no tax consequence anywhere.

Owner blocks and maintenance holds

Owners sometimes reason that a long block makes the property look less like a rental and more like a personal asset, or conversely that a stretch of blocked dates will dilute a long booking's effect on the average. Neither follows from the computation. A ninety-night winter block and a ninety-night winter lease produce opposite results: the block is invisible to the calculation, and the lease is one period of ninety days that lands squarely in both counts.

The arithmetic is unforgiving about that asymmetry. A property with ten four-night stays averages four days whether the rest of the year was fully blocked or fully available. Add one thirty-night stay and the average becomes 6.4 days across eleven periods. Add a sixty-night stay instead and it is 9.1 days, and the property is generally a rental activity for the year regardless of how many hours were logged. Volume of short stays is the only thing that cushions a long one, which is the point of the companion piece on how one long booking can break the average.

Cancellations and partial refunds

Here the regulation is thinner than owners would like, and it is worth being candid about that. Section 1.469-1T(e)(3)(iii) speaks of periods of customer use without addressing bookings that collapsed partway through, were cancelled before arrival under a policy that retained part of the payment, or were cut short with a partial refund.

Practitioners handle these differently, and the sensible reading usually tracks what actually happened rather than what was booked. A reservation cancelled before any stay occurred involved no customer use of the property even if money was retained. A guest who checked in for a fourteen-night booking and left after four used the property for four days. A stay that straddles 31 December raises a separate counting question that practitioners also resolve in more than one way.

What you can control is the record. Log what was booked, what was actually occupied, what was refunded and when, and keep the platform or contract documentation alongside it. Then the question goes to your CPA or EA with the facts attached rather than as a judgment you made silently on a spreadsheet months earlier.

Why the platform figure may not be the tax figure

A booking platform's "average length of stay" statistic is a business metric built for a different purpose. It may be computed over a trailing window rather than the tax year, may measure in days rather than nights, may exclude direct and off-platform bookings entirely, and may or may not treat a cancelled reservation as a stay. Any one of those differences moves the number.

STR Tracker does not compute the average for you; it keeps your participation hours with contemporaneous timestamps and reminds you to check the average-stay figure at the source, which is the reservation report on your booking platform plus whatever direct bookings live outside it. The figure that belongs in a tax analysis is one you assembled from the complete list of stays, not one a dashboard produced from part of it.

Checking it before year-end, not after

The average is a per-year, per-activity calculation, so it closes on 31 December and nothing after that changes it. The failure owners describe is finding out in March.

Two numbers per property, reviewed monthly, are enough: stays so far and nights those stays covered. From those, the running average follows, and so does the more useful figure — how long one more booking could run before the average crosses seven. That is the number that answers an October inquiry for a three-week stay while there is still a decision to make. How the seven-day test fits with material participation covers why both conditions have to hold in the same year, and why hours alone do not settle it.

What to write down

For each property, for each stay: check-in and check-out dates, nights actually occupied, the booking source, and whether anything was cancelled or refunded. Record blocked dates separately with a reason — personal use, maintenance, off-season — because that separation is what keeps blocked nights out of the customer-use counts while still documenting why the calendar looked empty. Participation hours stay in their own log under Treas. Reg. §1.469-5T; the two records answer different questions and should not be merged into one column.

7-day ruleSTR loopholerecord-keeping

Keep the record while it is still fresh

STR Tracker logs hours in seconds, keeps taxpayer and spouse separate, records other participants’ hours, and exports a CPA-ready PDF for any property and period. 14 days free, no card.

Prefer a spreadsheet? Take the free template

The columns a material-participation log needs, with a worked example.

Download the template