How to Get County Parcel Data: Portals, Records Requests and Bulk Delivery

County parcel data, meaning the boundaries, owners, assessed values and characteristics of every taxable piece of land, is produced by each county's assessor and GIS office and is public in most of the country. The free route is the county's own GIS or open-data portal, which in many counties offers a downloadable parcel layer, or a public records request to the assessor for the assessment roll. Some states aggregate their counties into a statewide parcel layer. The paid route is a dataset that has already done the collecting and normalizing across the more than 3,000 counties and county equivalents; PropRaven publishes 191.3M parcels, 110.0M of them mapped, at the 2026-08-30 epoch, with a coverage page for every county under parcel data. This guide walks through the county route first, then explains what the files contain, what tends to go wrong, and where an aggregated dataset earns its place.

The steps

1. Decide which layers you actually need. "Parcel data" is several datasets from several offices. Boundaries come from the GIS office. Owner, mailing address, assessed values and characteristics come from the assessor's roll (sometimes called the CAMA extract). Deeds live with the recorder. Tax bills and delinquency live with the treasurer or tax collector. Permits live with the building department. Knowing which you need tells you which door to knock on.

2. Check the county GIS or open-data portal. Search for the county name plus "GIS data download" or "open data." Look for a parcel layer, typically as a shapefile, file geodatabase or GeoJSON, and an assessment table as CSV or Excel, joined on the parcel number. Read the metadata page before downloading: it states the vintage, the coordinate system, the update cadence and any license terms.

3. Look for a statewide program. A number of states publish a statewide parcel layer assembled from their counties, usually through a state GIS office, sometimes with a standardized schema. It is worth checking before assembling counties yourself.

4. If it is not posted, make a public records request. Write to the assessor (or the GIS office), name the dataset, the fields and the format you want, and ask about fees. Practice varies widely: some counties send a file for free within days, some charge a nominal media fee, and some charge substantial amounts or license the data with restrictions on commercial redistribution. Some withhold owner names from bulk extracts even though they show them one record at a time.

5. Join and normalize. Match the GIS layer to the assessment table on the parcel number after stripping punctuation and preserving leading zeros. Reproject the geometry, which is usually in a state plane coordinate system in feet, to latitude and longitude if you plan to combine counties. Decide how to handle parcels with multiple polygons, condominium stacks, and rows that appear more than once.

6. Plan for refresh. Assessor rolls update on the county's own schedule, from monthly to annually, and GIS layers often lag the roll. A one-time download ages quickly; note the vintage and decide how often you will repeat the pull.

7. Or take the aggregated route. If the job spans several counties, or needs deeds, permits and hazard data joined to the parcel, a national dataset that has already done steps 2 through 6 is usually the cheaper path. The last section covers what that looks like on PropRaven.

What the files actually contain

Every county's schema is its own, but the common core is consistent.

• Geometry: the parcel polygon in the county's coordinate system, with a parcel-number attribute to join on. Some counties publish only centroids; some publish polygons for part of the county.
• Identity: the parcel number, the legal description (often truncated), lot size, and zoning or land-use codes, though zoning frequently comes from the planning department rather than the assessor.
• Situs address: the property's own address, which is blank for many vacant parcels.
• Owner and mailing address: the owner name as recorded and the tax-bill mailing address.
• Assessment: land, improvement and total assessed values, exemptions and taxable value for the current year, and sometimes prior years.
• Characteristics: property class, year built, building area, unit counts.
• Last transfer: last sale date and, where the state discloses it, the price.

Usually not in the assessor extract: deed history beyond the last transfer, building permits, flood zone or other hazard layers, and the owner's other properties. Each of those is a separate source and a separate join.

Common pitfalls

• Coordinate systems. A layer in state plane feet dropped onto a lat/lon map lands in the ocean. Check the projection file and reproject.
• Parcel-number formatting. The GIS layer and the roll often format the same number differently. Normalize both sides before joining, and never let a spreadsheet turn the field into an integer.
• Polygons without attributes, attributes without polygons. Some records exist in one file and not the other. Report the join rate rather than assuming it is 100%.
• Duplicate rows. A parcel can appear more than once when a county publishes multiple sources or when a property spans records. Row counts overstate parcel counts; PropRaven's own parcel data index explains why its county record counts sum to more than its distinct-parcel total.
• Redacted and placeholder owners. Some feeds withhold owner names, and many carry placeholders such as "Current Owner" or "Not Available." Treat those as blanks.
• Zero assessed values. Several counties publish an assessed value of zero for tax-exempt parcels (government, religious, nonprofit). Filter, do not average, before you compute anything.
• Non-disclosure states. Where the sale price is not public, the field is blank by law, not by error.
• License terms. Some counties restrict commercial use or redistribution. Read the terms before building a product on the file.

Where PropRaven fits

PropRaven collects the assessor, recorder and permit records itself and publishes them as one national schema, refreshed on a weekly canonical cadence. Each county page under parcel data shows the county's parcel record count, its rank in the state, and a coverage table with the share of records carrying a situs address, a usable coordinate, an owner name, an assessed value above zero and a parcel polygon, each beside the state average, along with the serving epoch the figures were measured at. A field that is empty for a county is empty because the county source does not publish it, and the page says so rather than hiding it. That table is the honest first stop before you size a job in any county.

For a filtered list now, parcel search narrows by state, property type, assessed value and year built and exports the first 1,000 rows to CSV. The REST API export endpoint returns up to 10,000 rows with a key, and the Free plan gives 1,000 lookups a month with no card; SDKs ship for Python, TypeScript and Go, and the hosted MCP server exposes parcel_search and market_coverage to Claude, Cursor and other agents. Whole-county, whole-state and national tables are delivered under a data license as a Snowflake Marketplace share, Parquet files or CSV, compared on the bulk parcel data page.

If you need one county's polygons with every attribute in the county's own schema, the county GIS download is often the right answer and it is free. PropRaven's value is the normalization across counties and the joins to deeds, permits and hazard data that the assessor extract does not carry.