Most revenue estimates for a short-term rental purchase are built backwards. Someone lands on a number first — a projection tool's annual figure, a seller's claimed gross, something a neighbor said over a fence — and then goes looking for support. By the time the offer goes in, the estimate has stopped being a forecast and started being a justification.
A useful estimate runs the other way. You pick a comp set deliberately, you adjust it down to the specific house you are buying, and you finish with a range you would still be willing to close on if the bottom of that range turned out to be the truth. The work takes an afternoon. It is the highest-return afternoon in the whole deal.
The estimate is not there to tell you the property will work. It is there to tell you what has to be true for the property to work, so you can go check those things before you wire money.
That reframing changes what a good output looks like. A single annual number is close to useless, because it hides every assumption inside it. A range with the drivers named is useful, because each driver is something you can independently verify, argue with, or walk away over. If your estimate says the deal needs a certain nightly rate in peak season, you can go look at whether comparable homes are actually holding that rate in peak season. If it says the deal needs the shoulder months to carry a certain occupancy, you can look at whether anything in that submarket fills shoulder months at all.
Everything downstream depends on which listings you decide are comparable. Get this wrong and no amount of spreadsheet discipline saves you.
A comp set is not "short-term rentals in the city." City-level averages blend beachfront against interior, four-bedroom against studio, professionally run against dormant. The average of those is a number that describes no property you could actually buy.
Pick eight to fifteen listings that genuinely substitute for the house you are underwriting. Same guest-count capacity, same walkability or drive time to whatever brings people to the area, same general condition tier, same parking situation. Then read them properly: look at their calendars over a full year, their review pace, their photo quality, whether they appear actively managed or abandoned. A listing with six reviews in three years is not a comp for a property you intend to run well. Neither is a listing three streets over that happens to have ocean views yours does not.
This is the part of underwriting that does not travel. Knowing that one side of a street in Deerfield Beach books differently than the other, or that a Durham house near a particular corridor draws a different guest than one two miles out, comes from working the market. It is also the specific reason a national management company cannot underwrite your deal well, no matter how much data sits behind their platform. They have the averages. The averages are not where the money is.
Once the comp set is honest, four inputs do nearly all the work.
| Input | What to pull | The mistake to avoid |
|---|---|---|
| Achieved nightly rate | What comps actually booked at, by season, net of length-of-stay and early-booking discounts | Using listed or asking rates, which are what hosts hoped for, not what they got |
| Paid occupancy | Nights booked divided by nights genuinely available for booking | Counting owner blocks, maintenance holds, or a closed calendar as vacancy, which flatters or crushes the number depending on which way you err |
| Seasonal shape | Month-by-month distribution across a full twelve months, not an annual average | Underwriting an annual figure that is really three strong months carrying nine weak ones |
| Stay-length mix | Typical booked stay length in the comp set | Ignoring that turnover frequency drives both operating load and how many nights you can realistically sell |
Notice that all four are properties of the comp set, not of the market. You are not estimating what short-term rentals earn. You are estimating what this house, run competently, in this specific spot, is likely to earn.
Just as important is what does not go into the model.
Comps tell you what the neighborhood supports. Adjustments tell you where your specific property sits inside that.
Work through the differences deliberately and assign each one a direction and a rough magnitude: bedroom and bathroom count against the comp set, sleeping capacity, whether there is a pool where the strong comps have one, parking, outdoor space, layout for groups, condition and furnishing tier, and how the property photographs. Then write down what you intend to spend to close the gaps you can close. A house that needs a full furnishing package to reach the comp set's tier does not earn the comp set's revenue on day one, and the gap between those two states is real money you have to fund.
Be equally honest about the gaps you cannot close. You can furnish your way to better photographs. You cannot furnish your way to a different street.
Three cases, not one number.
The base case is the comp set's middle performance, adjusted for your property and discounted for year one. The upside case assumes the property stabilizes at the stronger end of the set after it has a review history. The downside case is the one that matters: assume a softer rate environment, a slower ramp, and one significant disruption during the year — a repair that takes the property offline for a stretch, or a soft season that does not recover.
Then ask the only question the estimate exists to answer: if the downside case is what happens, is this still a deal you want? If the answer depends on the upside case landing, you are not underwriting an investment, you are buying a lottery ticket with a mortgage attached.
Three failure modes account for most of the damage we see in numbers owners bring us.
Annual averages hiding seasonal reality. An annual figure spread evenly across twelve months describes almost no real market. Model the months.
Assuming competent operation without accounting for it. Comp revenue reflects listings that are actively priced, actively maintained, and actively responded to. That performance is the output of someone doing work continuously. Short-term rental is a business, not passive income. It can be passive for the owner, but only because someone else is doing the work, and that arrangement is a line in the model rather than an assumption you get for free.
Estimating once and never revisiting. The estimate you build before closing is a hypothesis. Once the property is live, real booking data tells you within a season whether the comp set was right. Owners who go back and compare learn something they can use on the next purchase. Owners who file the spreadsheet away repeat the same error.
This is a top-line exercise. It tells you what the property is likely to earn, not what you will keep. Operating costs, financing, capital reserves, and the rest of the diligence a purchase requires are separate tracks, and none of them should be estimated from a revenue model. Keep them separate so a weak number in one does not get quietly offset by an optimistic number in another.
Choose comps that genuinely substitute for your property. Pull achieved rate, paid occupancy, seasonal shape, and stay-length mix from those comps rather than from the market. Throw out the seller's claim, the tool's headline number, and the one standout listing. Adjust for the house you are actually buying, discount the first year, and build a range you can live at the bottom of.
We run this exercise in The Triangle and across Southeast Florida for owners deciding whether a specific house is worth buying, and we will tell you when the answer is no. If you have a property under consideration in Raleigh, Durham, Chapel Hill, or anywhere from Miami up through Lantana, send it to us for a free revenue estimate and we will show you the comp set behind the number.
