We publish our match rates. Including the bad ones.
Every property-data vendor claims records are “linked to parcels.” Almost none say at what rate, in which counties, or where it fails. This page does: per-county parcel match rates from a read-only production audit, the portfolio numbers, the weak counties with the reason for each, and live source health.
Match rate by market
Top markets by record volume. Method: APN = direct parcel-key join; crosswalk = official government key mapping; address = normalized exact match with ambiguity dropped.
| Market | Deeds / sales | Assessments | Method |
|---|---|---|---|
| Los Angeles, CA 12,354,743 of 12,512,363 assessment rows attached | — | 98.7% | APN |
| Cook, IL (Chicago) 2,494,748 deeds; 47.6M multi-year assessment rows | 93.7% | 97.3% | APN |
| New York City (ACRIS, 5 boroughs) ~20.9M of 22,448,704 deed rows; full-walk verified | 93% | see NY statewide | APN (BBL) + official DOF condo crosswalk |
| Florida statewide (all 67 counties) 10,305,441 of 10,511,205 assessments attached | 97.3% | 98.0% | APN (folio) |
| Massachusetts statewide (L3 sales) 2,128,592 of 2,141,022; residual verified unmatchable | 99.42% | matched at load | APN |
| Colorado Front Range (6 counties) 876,483 sales and 1,854,413 assessments, zero unmatched | 100% | 100% | APN |
| Wake, NC (Raleigh) 372,909 of 372,913 | 99.999% | — | APN |
| King, WA (Seattle) | 99.99% | matched at load | APN |
| San Francisco, CA | — | 98.4% | APN |
| Allegheny, PA (Pittsburgh) | 99.997% (events) | 100% (sample roll) | APN |
| Ohio metros (Cuyahoga, Hamilton, Montgomery, Lorain, Medina, Lake, Clermont) | 85–99% | 93–96% | APN |
| Connecticut statewide (CAMA) | 85.1% | 85.8% | APN |
| Vermont statewide (PTTR transfers) | 80.2% | — | APN |
| New York statewide roll (57 covered counties) 14,025,195 of 23,580,051 scanned; the shortfall is concentrated in 24 counties that withhold their parcel files from the state GIS program | see NYC | 59.5% | State SWIS/print-key crosswalk |
Where linkage is still thin — and why
| Area | What is affected | Why |
|---|---|---|
| Wayne, MI (Detroit) | 892,632 property events unlinked | Michigan's statewide parcel file excludes Wayne County; county file is on the acquisition list. |
| Clark, NV (Las Vegas) and Jackson, MO (Kansas City) | 1.79M permits unlinked | County parcel layers not yet loaded. |
| Nassau and Monroe, NY (plus 22 more NY counties) | Assessment rolls unlinked (~9M rows) | These counties withhold their parcel files from the state GIS program, so the state crosswalk has no target parcels. |
| Orleans Parish, LA and Jefferson, KY | ~279k property events unlinked | Parcels not yet loaded for the parish/county. |
| Tarrant and Bexar, TX | Fort Worth permits and San Antonio assessments queued | Parcel coordinates recovered 2026-07-30; matching is in the current work queue. |
Source health
754 registered sources as of 2026-07-30. A live per-source view is on the freshness page.
| Status | Sources | Meaning |
|---|---|---|
| Healthy | 598 | Pulled on schedule; recurring scrapers and feeds. |
| Frozen | 107 | One-shot bulk loads (state GDB/CSV drops) — frozen by design, refreshed when the state publishes a new vintage. |
| Degraded | 35 | Recurring source stale beyond its window; repairs in progress. |
| Dead | 14 | One vendor class of portal scrapers stale since June; replacement path underway. |
Method
A record attaches to a parcel by direct parcel-key join first (APN/BBL/folio against the unique parcel index), by an official government crosswalk where the county's key differs from the state layer, and only then by normalized-address exact match. Ambiguity is dropped, never guessed: an address matching multiple parcels does not attach. The full counting rules are on the methodology page.
Questions buyers ask
The share of records (deeds, assessments, events) that carry a parcel_id resolving to a row in the parcel table. Records attach by direct parcel-key join (APN/BBL/folio) first, official government crosswalks second, and normalized-address exact match last. Ambiguous matches are dropped, never guessed.
Because a buyer who discovers a gap after paying is worse than one who knew before. Every weak county listed has a specific, verifiable reason — usually a county that withholds its parcel file — and a remediation path.
The match-rate figures come from a read-only production audit dated 2026-07-30, with each figure labeled by its exact source: a production count, a named run log, or arithmetic over the two.