The single most expensive number in a deal you do not own yet is the rent. Every ratio you run, the cap rate, the cash-on-cash, the debt-service coverage, all of it sits on top of one assumption about what the units will actually lease for. Overstate the rent by $150 a month on a duplex and you have overstated annual income by $3,600, which can flip a deal from cash-flowing to bleeding without changing a single other line. New investors do not lose money on cap rate math. They lose it on a rent number they took from a seller, a listing site, or an automated estimate and never checked.
This page is about the rent number before you buy: how to build your own comparable-rent set for a property you are underwriting, why automated estimates run hot, and how to adjust comps so the figure you drop into your worksheet is one you can defend. If you already own the unit and are deciding what to charge a sitting tenant, that is a different job with different inputs. Here the question is narrower and higher-stakes: what will this specific unit rent for once it is yours, and is the rent everyone keeps quoting you real.
Rent comps, defined for a buyer
A rent comp is a unit similar enough to the one you are buying that its rent is evidence for what yours will fetch. The closer the match on the things tenants pay for, the better the evidence. The variables that move rent the most, roughly in order, are location down to the neighborhood and school zone, bedroom count, bathroom count, square footage, and then condition and amenities like in-unit laundry, parking, and central air. A three-bedroom on the same street is worth more as a comp than a two-bedroom in a nicer suburb two zip codes over, even if the second one feels closer in price.
There are two kinds of comps and they answer different questions. Active listings are units on the market right now, and they tell you what landlords are asking. Closed or signed leases tell you what tenants actually agreed to pay, which is the number you want, because asking rents include the optimism and the units that sit unrented for weeks. Asking rents are easy to find and signed rents are not, so most buyers end up with a set of active listings and a discount applied to them. That is fine as long as you know that is what you are doing.
Where to actually pull comps
You can build a workable comp set from free, public listings in an afternoon. The point is to gather raw rents you can see and verify, not to outsource the number to a black box.
- Active rental listing sites. The large rental marketplaces show current asking rents with photos, bedroom and bathroom counts, and square footage. Filter to the subject's bed and bath count within a tight radius and save the listings that genuinely match. Photos matter: they are how you judge condition, which the asking rent alone hides.
- A local agent or property manager. Someone who leases units in the submarket can often tell you what comparable units actually signed for, not only what they listed for. Signed rents are the gold standard and this is usually the only way a buyer gets them before closing. A fifteen-minute call can be worth more than a day on listing sites.
- Property records and the current lease. When you are under contract, ask the seller for the in-place lease and a rent ledger showing what the current tenant has actually paid. That is the one unarguable rent in the whole exercise, and it also tells you whether a tenant is sitting well under market.
- Automated rent estimators. Tools that spit out a single number are useful for a first cut and for sanity-checking your own work, but they are an input, not an answer. The next section is about why.
Why automated estimates run hot
Automated rent estimates have two structural problems that push the number up. First, most of them are trained on asking rents from listings, and asking rents sit above signed rents because they include every unit that listed high and then cut the price or sat empty. The gap is real money: depending on the market and how tight it is, signed rents commonly land a few percent to north of ten percent below the asking rents the models learned from. Second, the models cannot see condition. A unit with original 1990s kitchens and a unit that was newly renovated to the studs read as the same three-bedroom to an algorithm, so the tired unit gets credit it has not earned.
That does not make the tools useless. It makes them a ceiling, not a target. Take the estimate as the top of a plausible range, then move down from it based on the actual condition of the unit you are buying and the signed rents you can find. One study after another on deal-analysis failures lands in the same place: the most common reason a new investor's first deal underperforms the spreadsheet is rent that came in below the assumption. Building from real comps instead of a single estimate is how you avoid being that statistic.
When a seller or a turnkey provider quotes a rent, find out whether it is in-place rent or pro-forma rent. In-place means a tenant is paying it today, which you can verify with the lease. Pro-forma means the rent the unit is supposed to achieve after some improvement or turnover, which it may never have achieved. The two get blended in marketing on purpose. Make them separate them for you.
Adjusting comps to your unit
Raw comps are never identical to your unit, so you adjust. The method is the same one appraisers use for sale prices, applied to rent: start with the comp's rent and add or subtract for each feature that differs, ending at an adjusted rent that estimates what your unit would command. Say you are buying a two-bedroom, one-bath unit with no in-unit laundry, and your best comp is a two-bedroom, one-bath three blocks away renting for $1,500 that has a renovated kitchen and a washer-dryer in the unit. If in-unit laundry is worth roughly $50 a month in that market and the renovated kitchen another $75, you subtract $125, giving an adjusted comp of $1,375 for your unit. Do that across three to five comps and you have a defensible range instead of a guess.
Two discipline rules keep this honest. Adjust toward your unit, not toward the rent you want, which means writing the adjustments down before you total them so you cannot quietly nudge the answer. And throw out any comp you have to adjust by more than about fifteen percent, because at that point it is a different kind of unit and is no longer telling you much. Three tight comps beat ten loose ones every time.
Turning the comp into an underwriting number
The rent you settle on is the gross rent at the top of your worksheet, and from there the same number flows into every metric you use to judge the deal. Before you trust the seller's spreadsheet at all, read it against your own figures: a listing's pro forma rent is a claim, not a fact, and the whole point of building comps is to replace that claim with something you verified. Once your rent number is solid, feed it into a full rental property analysis and a cash-on-cash calculator so you can see what the deal returns at the rent you can actually defend, not the rent the listing assumed.
It also pays to be conservative on the way in. Underwrite the deal at a rent at or slightly below the middle of your comp range, not the top, and apply a real vacancy allowance so a month or two of turnover does not blow up your first-year numbers. Rent comps are the foundation, and like any foundation they only do their job when they sit below the optimism, not on top of it. This discipline is also where a rental property portfolio gets built one verified deal at a time.
From comps to your own numbers
The comp you build before you buy is a borrowed number about a unit you do not own yet. The moment you close, you start generating something better, which is your own signed rents on your own units in that submarket. I self-manage my own small portfolio from two time zones away, and a few years in, my single best comp set is not a listing site, it is what my own units actually rent for and what they renew at. That history is what turns every future lease decision from a guess into a comparison.
That is the gap rents.ai is built to close after the purchase: it tracks the actual rent on each of your units and flags upcoming lease renewals, so each renewal starts from your real numbers across the portfolio instead of a fresh search. It will not pull comps or estimate rent for a property you are still underwriting, though. Market rent is a value you enter per unit, so the pre-purchase comp work on this page stays yours to do. Borrowed comps get you to the closing table. Your own rent roll is what tells you whether the next one is priced right.