How Many Comps Does a Legitimate STR Revenue Projection Actually Use?

Why there's no fixed "right" number

A dense urban short-term rental market might have hundreds of genuinely comparable listings within a tight radius. A remote rural market might have only a handful of short-term rentals within a reasonable distance, period. Requiring the same comp count in both cases doesn't make sense — the rural projection isn't inherently worse just because fewer comps exist; it's working with what the market actually offers.

It helps to think about comp count the same way you'd think about a survey sample size in any other context. A larger sample generally gives you more statistical confidence, but only if the individuals in it are actually representative of what you're trying to measure. A large sample of the wrong population is worse than a small sample of the right one, and a comp set works exactly the same way — raw size on its own isn't the goal.

This is the trade-off worth understanding, covered in more depth in /learn/what-comp-set-means-str-revenue-projection/: a wider radius or looser similarity filter can artificially inflate the comp count, but at the cost of relevance. More comps that are less similar to your property isn't obviously better than fewer comps that are more similar.

It's also worth recognizing that comp density itself can vary by season within the same market. A ski town might have plenty of comparable listings with winter performance data but relatively few with a clear summer track record, simply because fewer owners actively market the property outside the main season. That kind of seasonal thinness in the comp set is different from a market being generally thin, and it's worth asking about specifically if your property's off-season performance is part of what's being estimated.

What to actually evaluate instead of chasing a number

• Relevance over raw count: five tightly matched comps in your exact micro-market can be more informative than twenty pulled from a much broader, more varied area.
• Whether the radius had to be widened to reach a usable count: if a projection needed to stretch its geographic or similarity filters significantly to hit a reasonable comp count, that's worth knowing.
• Consistency among the comps used: a comp set where individual properties' performance varies wildly from each other suggests more underlying uncertainty than one where comps cluster more closely.
• Transparency: a projection that discloses its comp count and filters at all is generally more trustworthy than one that only shows a final number with no visibility into the process.

How this affects a genuinely rural or unusual property

If your property sits in a market with very few comparable short-term rentals nearby, expect any projection — automated or appraiser-driven — to lean more heavily on judgment and a wider search radius than it would in a denser market. That's not a flaw in the process; it's an honest reflection of a market with less available comparable data.

In these thinner markets, it's often worth putting extra weight on an appraiser's in-person opinion relative to an automated tool, precisely because a human evaluator can incorporate local knowledge — a nearby attraction opening, a road improving access, a specific reason demand is shifting — that a purely comp-based model has no way to account for when the comp pool itself is small.

It's also reasonable, in a thin market, to ask more than one source the same question rather than relying on a single projection. If a market-projection tool, an appraiser, and any available nearby rent rolls all point in a broadly similar direction despite each drawing on a small and imperfect comp pool, that convergence across independent methods is meaningful evidence in its own right — arguably more meaningful than a single source with a large but less relevant comp count would be.