Every rental underwrite rests on one number, and that number is rent. Get it wrong and everything downstream (cash flow, cash-on-cash return, the entire buy decision) is wrong with it. The problem is that rent is not a fact you can look up. It is an estimate, and the quality of the estimate depends entirely on the comparable rentals you build it from. Pull the wrong comps and you will confidently underwrite a rent your property will never see.
This is how to read rent comps like the estimate is your money, because it is.
What a rent comp is
A rent comp is a nearby property, similar to yours, that is currently rented or listed for rent, whose rent tells you what a tenant will actually pay for what you are offering. The keyword is actually. Not what the seller hopes, not what the listing description claims, not what the mortgage requires. What a real tenant is paying right now for a comparable unit.
Good comps share four traits with your property: location, size, type, and condition. The closer the match on all four, the less adjusting you have to do and the more trustworthy the number. Weak matches force big adjustments, and every adjustment is a place for error to creep in.
Where to pull comps
Use more than one source, because each has a blind spot:
- Active rental listings. Zillow, Apartments.com, Realtor, and local listing sites show what landlords are asking right now. The catch: asking rent is not achieved rent. A unit listed for 45 days at $1,500 is telling you $1,500 is too high, not that the market rent is $1,500.
- Recently rented listings. More valuable than active ones, because a unit that rented tells you a tenant agreed to that price. Some platforms show rented history; it is worth more than any active listing.
- Rent estimate models (AVMs). Automated valuation models return a rent range from an algorithm trained on comps. Good ones give a low, middle, and high. Use the low-to-middle portion, never the top of the range.
- HUD Fair Market Rents. A government dataset of typical rents by ZIP and bedroom count. Useful as a sanity ceiling, but it runs high relative to what ordinary units achieve, so treat it as an upper bound, not a target. We cover how investors use it in How Investors Use HUD Fair Market Rents.
No single source is the answer. The estimate you trust is the one where independent sources agree, and when they disagree, you underwrite with the lower figure.
The five traits to match, and how to adjust
You will rarely find a perfect comp, so you adjust. Match these five, in roughly this order of importance:
- Location. The single most important trait, and the one that does not adjust cleanly. A comp on the next street can be worth 20 percent more or less than yours because of a school boundary, a busy road, or a rougher block. Stay as tight geographically as you can. A pristine comp two miles away is often worse than a rough one next door.
- Bedrooms and bathrooms. Match bedroom count first; it drives rent more than square footage. Adjust for bathroom count second.
- Square footage. Adjust within the same bedroom count. A 1,100 and a 1,400 square foot three-bedroom rent differently, but less than a two-bed versus a three-bed does.
- Property type. A single-family house, a townhome, and a unit in a fourplex rent differently even at the same size, because tenants value the yard, the privacy, and the parking differently.
- Condition and finishes. Updated kitchen and flooring command a premium; dated or worn units discount. Photos in the comp listing are your best evidence here.
The goal is not a spreadsheet of precise dollar adjustments. It is a defensible range. Three to five genuinely comparable units that cluster around a number give you far more confidence than one perfect comp or ten loose ones.
One more filter worth applying: recency. A comp that rented eight months ago in a different season may not reflect today's market, because rents move and demand swings with the calendar. Weight recent rentals over older ones, and if the only comps you can find are stale, widen your search radius before you widen your time window. A slightly less local comp from last month usually beats a perfect-match comp from last year.
A worked read
Suppose you are estimating rent for a three-bedroom, two-bath single-family house of about 1,300 square feet, recently updated. You pull five comps:
| Comp | Beds/Baths | Sq ft | Status | Rent | Note |
|---|---|---|---|---|---|
| A | 3/2 | 1,280 | Rented | $1,650 | Nearly identical, same neighborhood |
| B | 3/2 | 1,400 | Rented | $1,725 | Larger, updated |
| C | 3/1.5 | 1,250 | Active 40 days | $1,800 | Overpriced, sitting |
| D | 3/2 | 1,320 | Rented | $1,600 | Dated finishes |
| E | 4/2 | 1,600 | Rented | $1,900 | Bigger, more beds |
Read this the way the market does. Comps A and D are your tightest matches and both rented, at $1,650 and $1,600. Comp B rented higher but is larger and updated, so it caps your upside, not your baseline. Comp C is the trap: its $1,800 asking price is seductive, but it has been sitting for 40 days, which is the market saying no. Comp E is a four-bedroom and simply not your unit.
The honest read is a range of roughly $1,600 to $1,700, and you underwrite at the conservative end, near $1,600 to $1,650. You do not underwrite at $1,800 because one overpriced listing exists. Using Comp C's number is exactly how investors talk themselves into deals that do not work.
The single most expensive habit in rent estimation is anchoring on the highest asking rent you can find. Asking rent that is not renting is not evidence of anything except a landlord who has not cut the price yet.
The traps that produce a fake number
A few specific mistakes account for most bad rent estimates:
- Inheriting the seller's number. The listing's claimed rent is a marketing figure. Rebuild it from comps yourself, every time. This is the core discipline in How to Estimate Rent Accurately Before You Buy.
- Using asking rents as achieved rents. Filter for rented units when you can, and discount active listings that have been sitting.
- Pricing multifamily by total bedrooms. A duplex is two units, each estimated separately and added. Feeding a duplex's combined bedroom count into a single-family rent model produces a number that is meaningfully wrong, because a four-bedroom house and two two-bedroom units do not rent for the same total.
- Ignoring days on market. How long comps sit is data. A market full of quick rentals supports your number; a market full of stale listings warns you the asking prices are fantasy. The same signal applies to for-sale listings, covered in Days on Market and Price Cuts.
- Cherry-picking to hit a target. If you already decided the deal works at $1,800 and then go looking for a comp that justifies it, you are not estimating rent. You are confirming a bias, and the tenant will not cooperate.
Why this is the number that matters most
Rent sits at the top of the underwrite, which means every error in it flows straight through to cash flow and return, magnified by leverage. Overstate rent by $150 a month and you have overstated annual income by $1,800, which on a leveraged deal can be the entire difference between positive and negative cash flow. That is why the discipline of reading comps carefully is not busywork. It is the highest-leverage ten minutes in the whole analysis.
This is exactly the step PadSweep automates and refuses to shortcut. Every listing gets an independent rent estimate built from comparable rentals and cross-checked against multiple sources, deliberately conservative, never inherited from the listing. You can see the resulting numbers on live listings on the markets page, or start a free trial and run it on your own market.