How to Run Comparable Sales (Comps) on a Property

A comparable sale — usually just called a "comp" — is a property that has actually closed escrow, is located near the subject property, and is similar enough in size, condition, age, and features that its sale price offers a reliable clue about what the subject property is worth. Comps are the foundation of almost every real estate valuation method used by investors, appraisers, and agents alike, whether the goal is estimating current market value, projecting an After Repair Value (covered in our ARV guide), or pricing a raw land parcel.

The discipline of running comps is really the discipline of finding the closest available proxies for a property that, by definition, hasn't sold yet — and then correcting for every way the proxies differ from the real thing. Done carelessly, a comp analysis can be bent to support almost any number a person wants to arrive at. Done correctly, it's one of the more objective tools in real estate.

Three filters matter most when selecting comps. Recency: sales within the last three to six months are generally reliable; anything older risks reflecting a market that has since shifted, especially in Northern California markets where interest-rate changes have moved buyer behavior quickly over the past few years. Proximity: comps within a half-mile to one mile are strongest, since even within the same city, pricing can vary meaningfully by school boundary, HOA, or proximity to a busy road — a comp two miles away in a different neighborhood may look similar on paper but sell in a completely different price band.

Similarity covers the physical characteristics: square footage within roughly 15% to 20% of the subject property, the same bed and bath count where possible, comparable lot size, similar age or era of construction, and — critically — similar condition. A recently sold home that had a full remodel is not a good comp for a subject property in original 1975 condition unless you're specifically trying to establish an After Repair Value rather than current as-is value. Active listings and pending sales should never be treated as comps for a final valuation; only closed, recorded sales reflect what a buyer was actually willing and able to pay.

No two homes are identical, so the next step is adjusting each comp's sale price to account for differences from the subject property. This is done with a paired-data approach where possible — looking at what the local market actually pays for an extra bedroom, an additional garage bay, a larger lot, or a pool, based on how similar homes with and without that feature have sold. In the absence of enough local paired data, appraisers and investors fall back on standard rule-of-thumb adjustments, though these should be treated as a starting point rather than a precise science.

Adjustments run in a specific direction that trips people up at first: if a comp has a feature the subject property lacks, you subtract value from the comp (because the comp sold for more partly due to that feature, and the subject doesn't have it). If the comp lacks a feature the subject property has, you add value to the comp. The adjusted comp prices should converge toward a consistent range for the subject property; if adjusted values are scattered widely, that's a signal the comps weren't similar enough to begin with.

Say the subject property is a 1,800-square-foot, three-bedroom, two-bath home on a 0.25-acre lot in unincorporated Placer County, in good but unrenovated condition. Comp A sold for $460,000: same square footage and bed/bath count, but has a two-car garage versus the subject's one-car garage (adjust down $8,000) and a slightly larger 0.3-acre lot (adjust down $5,000), landing at an adjusted $447,000. Comp B sold for $445,000: 100 square feet smaller (adjust up $6,000) and updated kitchen the subject lacks (adjust down $12,000), landing at an adjusted $439,000. Comp C sold for $475,000: same size and lot, but fully remodeled throughout (adjust down $25,000 to account for the renovation gap), landing at an adjusted $450,000.

The three adjusted values — $447,000, $439,000, and $450,000 — cluster in a reasonably tight $11,000 range, giving a defensible estimate of roughly $445,000 for the subject property in its current condition. If instead the three adjusted values had come out to $390,000, $460,000, and $510,000, that spread would signal the comps weren't actually comparable enough, and better ones should be found before trusting the number.

The most common error is cherry-picking comps that support a number decided in advance rather than letting the best available data drive the estimate — whether that's a seller hoping for a high number or a buyer angling for a low one. A close second is using active or pending listings instead of closed sales; a listing price reflects what a seller hopes to get, not what a buyer has actually paid. Ignoring outliers is another frequent mistake — a distressed or off-market sale between related parties, or a rushed foreclosure sale, can print a price well below true market value and shouldn't be weighted the same as an arm's-length sale on the open market.

Failing to adjust for condition is especially common when a seller is comparing their unrenovated home to a neighbor's recently updated sale and assuming the prices should match. And relying on too few comps — sometimes just one — leaves no way to sanity-check whether the estimate is reliable or a fluke.

Running comps on vacant land follows the same core logic but with different variables. Instead of square footage and bed count, land comps are typically compared on a per-acre or per-buildable-lot basis, adjusted for access (paved road frontage versus a shared easement), utility availability (public water and sewer versus well and septic), topography and usable acreage versus steep or unbuildable terrain, and zoning. A five-acre parcel with confirmed perc-test results and existing utility access is not comparable to a superficially similar five-acre parcel that has never been tested and may face site constraints — the sale price gap between the two can be substantial even though they look alike on a map. Our land appreciation guide covers how these land-specific value drivers have trended across the region over time.