Methodology

Permit records are pulled from each jurisdiction's official open-data portal — primarily Socrata (SODA API) and ArcGIS REST services, with a browser-based fallback for portals without a clean API. We cache ingested records and serve from our database; we do not re-query a portal on every page view. Every record links back to its source and shows when it was retrieved.

Portals update on very different cadences — some daily, some monthly, and some lag a year or more. We surface a “data current as of” date for every market and flag jurisdictions that have gone quiet past their expected cadence. We would rather show you that a market is stale than present old data as if it were live.

Jurisdictions describe permits inconsistently. We normalize each permit's type and description into a consistent taxonomy (new construction, addition, ADU, roofing, solar, HVAC, electrical, plumbing, demolition, and more) using keyword rules, and we honor native flags where a portal provides them (for example, an explicit ADU indicator). These classifications are computed analysis, not official designations.

Neighborhood activity levels, trends (month-over-month and year-over-year), project-type mix, and “hotspot” grades are computed from the underlying public permits over a trailing window. They are an analytical lens to help you read an area quickly — not an official measure or a prediction.

Contractor pages aggregate public permit activity — where a contractor has pulled permits, what they build, and at what volume. This is public business information, not an endorsement, rating, or recommendation.

We limit output to public business and project fields. We deliberately exclude individual owners' or applicants' personal names and other personal information, even when a portal includes them.

Found something wrong or out of date? Permit data ultimately lives with the jurisdiction — but we want our copy to be accurate. Contact us and we'll review and re-sync the source.