Occupancy and average daily rate data is the raw material of short-term rental underwriting, and it is also the most frequently misread data in the industry. This page covers how to read trend data correctly, what the common distortions are, and how to turn published market figures into an underwriting input rather than a headline.
What do occupancy and ADR figures actually measure?
Less than they appear to. Occupancy is booked nights divided by available nights, and the denominator is where the definition varies: some sources exclude owner-blocked nights, some exclude nights a listing was inactive, and some count only listings that were active for the full period. ADR is the average rate on booked nights only, which means a property that booked four nights at a high rate reports a high ADR. Neither figure is comparable across sources without knowing which denominator produced it.
Why is RevPAN the number that matters?
Because it is the only one that cannot be improved by moving the other. Raising rates lifts ADR while thinning the calendar; discounting fills the calendar while destroying rate. RevPAN, which is ADR multiplied by occupancy, captures the net effect. Any market trend reported as an ADR change without the corresponding occupancy change is describing half of a movement, and the missing half is frequently in the opposite direction.
What distorts market-level trend data?
| Distortion | Effect on the reported trend |
|---|---|
| Supply growth outpacing demand growth | Occupancy falls while total market revenue rises |
| New listings entering mid-period | Depresses average occupancy without any property performing worse |
| Mix shift toward larger properties | Raises market ADR with no rate increase anywhere |
| Inactive listings included or excluded | Moves occupancy several points in either direction |
| Calendar effects and holiday placement | Creates apparent year-over-year swings |
| Regulatory enforcement removing listings | Raises occupancy for survivors, reads as demand growth |
The first and last are the ones that mislead most often. A market where occupancy is falling while total revenue rises is a market absorbing supply, which is a very different underwriting environment from a market where demand is weakening, and the two produce identical occupancy headlines.
How do you separate supply effects from demand effects?
Look at three series together rather than one: active listing count, total market revenue, and RevPAN. Rising listings with rising total revenue and falling RevPAN is supply absorption, which compresses individual property returns while the market grows. Flat listings with falling revenue and falling RevPAN is demand contraction, which is materially worse. Falling listings with rising RevPAN usually indicates regulatory enforcement or operator attrition, which improves economics for compliant survivors and raises the regulatory risk on any new entrant.
Why is national data almost useless for underwriting?
Because short-term rental performance is submarket-level and national aggregates average across markets moving in opposite directions. A national occupancy figure blends urban markets constrained by regulation, mountain markets driven by snowfall, and coastal markets driven by weather and fuel prices. Nothing in that average describes any specific property. Underwriting requires submarket data at minimum, and within a submarket it requires a comp set matched on bedroom count, guest capacity, and the amenities that actually drive booking.
How should seasonality be read?
As the shape of the year rather than as a footnote to the annual figure. An annual occupancy of 60 percent can describe a property booked evenly across twelve months or one booked at 95 percent for four months and 15 percent for eight. Those are different assets with different reserve requirements, because debt service, insurance, and utilities arrive at the same rate in both cases. Build the model monthly, then annualize, and look specifically at how many consecutive months the property operates below breakeven.
What does a trailing twelve month figure hide?
Direction. A trailing figure is an average across a period in which performance may have been rising or falling throughout, and a market declining steadily can produce an attractive trailing number for a full year after the decline began. Compare the most recent quarter against the same quarter a year earlier rather than relying on the annual total, and where both are available, look at forward booking pace, which leads reported performance by months.
How do you turn market data into an underwriting input?
Four steps. Establish which definition of occupancy the source uses and recompute against a full 365-day denominator if necessary. Convert to RevPAN so the figure is comparable. Take the median of a matched comp set rather than the market average, since the distribution is right-skewed and the mean flatters. Then apply a first-year discount for a new operator inheriting a listing without reviews, calendar momentum, or search position. The result is an input. The published headline was never one.
Where does market data sit in the analysis?
At Step 1 of the ten-step analysis sequence, before the seller’s figures are examined, so that the independent estimate anchors the comparison rather than the other way around. Investment Grade STR labels every revenue input by source and flags seller-reported and listing-reported figures as unverified wherever they are carried forward, because a number whose provenance is lost stops looking uncertain by the time it reaches a conclusion.
What data sources are available, and how do they differ?
Short-term rental market data comes from a small number of providers who derive figures by observing listings and calendars on the major booking platforms rather than from reported transactions. That derivation method is the source of most of the variance between them: providers differ on how they infer a booking versus an owner block, how they treat listings that go inactive, whether they include or exclude cleaning fees in revenue, and how far back their historical depth runs. Two providers can report occupancy figures for the same submarket that differ by ten points without either being wrong, because they are measuring different things and calling both occupancy.
The practical implication for underwriting is that a figure should never be moved between sources mid-model. Pick one source, understand its definitions, and hold it constant across the comp set, the market view, and the trend analysis. Blending providers produces a number with no definition at all.
How much historical depth do you actually need?
Enough to contain at least one demand disruption, which in practice means several years rather than a trailing twelve months. A market that has only been observed through a strong period looks structurally sound in a way that cannot be verified. Depth matters most in markets with weather-dependent demand, where a single poor snow season or an active hurricane year reveals the floor, and in markets that have experienced a regulatory change, where the before-and-after tells you what enforcement actually did to supply and to surviving operators.
What should you do when the data and the seller disagree?
Underwrite the lower figure and document the gap explicitly rather than reconciling it away. A seller number above the comp-derived estimate can be legitimate: a genuinely superior operator, a property with an amenity the comp set lacks, or an established direct booking channel. Each of those is verifiable, and the burden of verification sits with the party making the claim. Where it verifies, raise the estimate and note why. Where it does not, the comp-derived figure stands and the difference is recorded as an unverified seller claim rather than discarded silently. See income verification.