Bulk parcel data and its coverage limits
Published historical coverage: 191.3M distinct parcel IDs and 110.0M mapped locations at the 2026-08-30 published epoch. The ID count uses (state_fips, parcel_id); the mapped count measures distinct drawable locations. These are not a current inventory or a count of unique physical parcels.
Compare delivery documentation below, then confirm the release, fields and jurisdictions required for your use case.
Delivery paths
Warehouse shares, Parquet releases and API exports can differ in version, projection and serving rules. This page does not establish byte parity or current availability across them.
| Path | Format | Scope | Freshness evidence |
|---|---|---|---|
| Snowflake Marketplace share | Warehouse share | Listing-specific tables and access; availability and freshness are not checked by this page. | Confirm the listing's current schema, source periods and refresh evidence before relying on it. |
| Parquet on Cloudflare R2 | Versioned Parquet files and release manifests | Use the selected dataset's published manifest for its states, files and physical row counts. Current release availability is not checked here. | Verify manifest and seal, file hashes and producer input vintage. A version stamp alone does not date every field. |
| CSV export from search | Filtered CSV projection | Columns and row limits depend on the export route. A CSV is a projection, not the full catalog. | Export time dates the response; source records can have different and older vintages. |
Fields and provenance
The historical catalog v=2026-08-28 lists 620 columns across 51 jurisdictions (states and DC). It was generated 2026-09-01T20:43:28Z from statistics describing 289,856,342 physical serving rows. Column presence does not mean every parcel has a value.
The catalog includes source-record fields, derived values and historical enrichments. It does not establish that every field came from a county assessor, recorder or permit record, or that every producer is current. Names such as owner_name and total_assessed_value do not guarantee a current owner or current tax year.
Coverage measures population of values, not correctness, producer lineage or delivery eligibility. API serving gates may withhold fields present in a historical export. Review dated fill estimates and the schema for the specific delivery surface before sizing a job.
Refresh and release evidence
A scheduled cadence is a target, not evidence that a refresh completed. A publication or export timestamp also does not establish the source vintage of every field. Ask for the selected release's producer receipts, source periods, completeness checks and known limitations. The operations preview is illustrative and does not report live job status.
Licensing and evaluation
Confirm scope, delivery access, refresh obligations and pricing for the requested license with the team. Documentation and historical coverage alone do not establish what a current entitlement will deliver.
Frequently asked questions
- Can I download all U.S. parcel data at once?
- Bulk access is licensed. Confirm the requested states, dataset version, schema and delivery path with the licensing team; this page does not verify current warehouse or object-store availability.
- How many parcels are included?
- The historical published count is 191.3M distinct parcel IDs and 110.0M mapped locations at the 2026-08-30 published epoch. Distinct IDs, mapped locations and physical rows measure different populations; none should be substituted for another.
- Does the catalog count mean every field is populated?
- No. A catalog lists fields. Coverage, producer vintage, field correctness and the fields delivered by a particular endpoint must be assessed separately.
- Are warehouse, Parquet and API records identical?
- No cross-surface byte parity is established here. Releases, projections, derived fields and serving gates can differ. Compare the exact release and schema you intend to use.