Real Estate Comps by Address: What One Line of Text Can and Can’t Get You
undefined
Short answer
An address alone reliably gets you location, public property characteristics and a comparable-sales search. It does not reliably get you condition, renovations, the correct unit in a multi-unit building, or a confirmed sale price in non-disclosure states. Address-only lookups fail in three specific places — geocoding, record matching and unit resolution — and the failures are usually silent. Check the returned property facts against what you know before trusting the comps built on them.
"Enter an address, get comps" is now the standard interface for every valuation tool, ours included. It is a good interface. It also hides three separate lookups, each of which can fail quietly and produce a confident, wrong answer.
Knowing where the failures are is most of what separates someone who uses these tools well from someone who gets burned by one.
What actually happens when you type an address
Three steps, in order, each depending on the last being right.
- Geocoding. The text becomes a latitude and longitude.
- Record matching. The address is matched against a property database to retrieve characteristics — square footage, beds, baths, year built, lot size.
- Comparable search. A search runs around that coordinate for recent sales matching those characteristics.
Step three is the visible one. Steps one and two are where things go wrong.
Failure 1 — Geocoding lands on the wrong building
Geocoders return a match quality alongside the coordinate, and the levels are not equivalent:
| Match quality | What it means | Safe for comps? |
|---|---|---|
| Rooftop | The specific building was identified | Yes |
| Range interpolated | Estimated along a street segment | Usually — can be a lot or two off |
| Geometric centre | Centre of a block or parcel | Marginal |
| Approximate | Neighbourhood or ZIP-level guess | No |
An interpolated match assumes house numbers are evenly spaced along a street. On a cul-de-sac, a flag lot, a corner lot with an address on the cross street, or a road renumbered after a subdivision, that assumption breaks and the pin lands next door.
The visible symptom is a map or photo showing a building that is not yours. If you see that, do not trust the comps underneath it — the search ran from the wrong point.
Failure 2 — The record matched, but to the wrong record
Address text is messier than it looks. The same house can be written as 1200 W 6th St, 1200 West Sixth Street, 1200 W. 6th Street Apt 4, or with a directional the post office recognises and the county does not.
Matching engines normalise all of that, and mostly succeed. When they fail, the failure modes are:
- Dropped unit numbers — the most common, and the most damaging in condos.
- New construction — the address exists in the postal system before it exists in tax records, so a brand-new house returns nothing, or returns the vacant lot it replaced.
- Recently renumbered or renamed streets — annexation and subdivision plats do this routinely.
- Split or merged parcels — one record where there are now two properties, or vice versa.
The check is quick and worth doing every time: does the square footage the tool returned match what you know? If a tool tells you your 2,400 sq ft house is 1,650 sq ft, it matched a different record, and every comp and every adjustment downstream is now built on the wrong subject.
Failure 3 — Multi-unit buildings
Fifty condos share one street address. Unit 12B on the eighth floor with a view and Unit 2A facing the car park are not the same asset, and in many buildings the spread between them is 20–30%.
If a tool returns "comps for 400 Main Street" without resolving which unit, it has answered a different question than the one you asked. This is the single biggest weakness of address-only lookup, and it is why condo comps need their own rules.
What an address genuinely can't tell anyone
Set the failures aside. Even with perfect geocoding and a perfect record match, some things are simply not encoded in an address:
- Condition. Whether the kitchen is from 2024 or 1987.
- Unpermitted work. A converted garage often does not appear in the record but does affect value.
- What the property faces. A rail line, a commercial lot, a school pickup queue every afternoon.
- Interior layout quality. Two 1,800 sq ft houses with the same bed count can live very differently.
- Deferred maintenance. Roof age, foundation movement, systems at end of life.
Every address-driven valuation implicitly assumes average condition. That assumption is fine for a screening pass and wrong for a listing price.
The four-point check before you trust the output
Thirty seconds, and it catches nearly every silent failure:
- Is the map pin on your building? Not your street — your building.
- Is the square footage right, within a rounding error?
- Are beds, baths and year built right?
- If it is a unit, is it your unit?
If all four pass, the comps below are worth reading. If any fail, fix the subject before looking at anything else.
Why the price may still be missing
One more thing an address cannot overcome. In non-disclosure states — Texas among them — sale prices are not recorded publicly. A tool can resolve your address perfectly, find genuinely comparable properties, and still have no verified price for any of them unless it is licensed to MLS data.
That is a data-rights problem, not a lookup problem, and no amount of address precision solves it. It is worth asking any comps provider which of the two situations you are in.
The bottom line
Address-based comps are the right interface. One line of text really can resolve a location, a property record and a comparable set, and for a screening pass that is genuinely enough.
What it cannot do is verify itself. Geocoding can land next door, record matching can grab the wrong parcel, and a unit number can vanish — all without an error message. Run the four-point check, and treat the condition assumption as the open question it is.
Frequently asked questions
Can I get real estate comps with just an address?
Yes. An address is enough to geocode a location, match a public property record and run a comparable-sales search around it. What an address cannot convey is condition, renovation history or anything a person would have to look at the property to know, so address-only comps are a baseline that assumes average condition.
Why did a comps tool return the wrong property for my address?
Almost always geocoding or record matching. Geocoders interpolate along a street segment when they lack a rooftop-level match, which can land on a neighbouring parcel. Record matching fails when the address is formatted differently in the source data — a missing unit number, a road versus rd abbreviation, or a recently renumbered street.
Do comps by address work for condos and apartments?
Less reliably. Many units share one street address, so a lookup that ignores the unit number may return the building rather than your unit, or a unit on a different floor with a different view. Always confirm the tool resolved the specific unit, not just the building.
How accurate are automated comps from an address?
Accurate on retrieval and arithmetic, weaker on selection and condition. Automated selection reliably finds recent nearby sales of similar size, and reliably misses that one of them backs a highway or sold with a failed foundation. Treat the output as a screened candidate list that still needs human review.
What information should I check before trusting address-based comps?
Four things: that the returned square footage matches what you know, that the bed and bath counts are right, that the year built is right, and that the map pin is on your building rather than the one next door. If any of those four are wrong, every comp and adjustment downstream inherits the error.
Sources
- Geocoding match-quality terminology (rooftop, range-interpolated, geometric centre, approximate) follows the Google Geocoding API’s published location_type classification, which is the standard vocabulary for this distinction.
- Non-disclosure states, where sale prices are not recorded publicly, commonly listed as Alaska, Idaho, Kansas, Louisiana, Mississippi, Missouri (some counties), Montana, New Mexico, North Dakota, Texas, Utah and Wyoming.
- Fannie Mae Selling Guide B4-1.3-08 (effective 06/04/2025) requires distance to a comparable to be reported in miles with a directional indicator, measured as a straight line — the reason straight-line distance is the industry convention even where it is a poor proxy for comparability.