data.sfgov.org logo
data.sfgov.org

sf-business-registry-lookup

Installation

Adds this website's skill for your agents

 

Summary

Search San Francisco's registered business registry (DataSF g8m3-pdis) by DBA name or owner via the public Socrata SODA API and return each location's legal/owner name, DBA, address, NAICS code, status, and registration date.

FIG. 01
FIG. 02
FIG. 03
FIG. 04
SKILL.md
174 lines

SF Registered Business Registry Lookup

Purpose

Look up businesses in San Francisco's official tax registry by DBA name or owner/legal name and return each matching location's registry record: legal/owner name, DBA, full address, NAICS code + description, status (active / closed / administratively closed), and registration (business start) date. Backed by the City & County of San Francisco "Registered Business Locations - San Francisco" open dataset (g8m3-pdis, ~362,000 rows, refreshed daily). Read-only — this only queries public open data; it never writes or registers anything.

When to Use

  • Verify whether a business is registered with the SF Treasurer & Tax Collector and whether it's currently active.
  • Resolve a DBA / trade name to its registered legal owner (and vice-versa).
  • Pull a business's NAICS classification, registered address(es), and registration date for due-diligence, enrichment, or compliance checks.
  • Enumerate all locations of a multi-location business (each location is a separate row sharing one Business Account Number).
  • Anywhere you'd otherwise scrape the DataSF data-grid UI — the SODA API is faster, cheaper, and structurally reliable.

Workflow

data.sfgov.org is a Socrata open-data portal. The data-grid web UI is a thin, JS-heavy client over a public SODA (Socrata Open Data API) endpoint. Query the API directly — it needs no auth, no API token/app-token, no cookies, no residential proxy, and no anti-bot stealth (a bare GET returns 200). The browser grid (documented as a fallback at the end) loads 46+ XHRs and is meaningfully slower for no benefit.

Base endpoint: https://data.sfgov.org/resource/g8m3-pdis.json

  1. Pick a search mode and build a SoQL query:

    • DBA name (substring, case-insensitive) — the most common case: 
      GET /resource/g8m3-pdis.json
   ?$where=upper(dba_name) like '%PHILZ COFFEE%'
   &$limit=200
    • Owner / legal name (substring, case-insensitive): 
      ?$where=upper(ownership_name) like '%STARBUCKS%'&$limit=200
    • Free-text across all columns (use when you don't know if the term is a DBA or owner): ?$q=Philz Coffee&$limit=200. This is what the grid's search box does.
    • Exact DBA match (fast equality, case-sensitive): ?dba_name=Philz Coffee as a simple query param.

    URL-encode the whole $where value. SoQL string literals use single quotes; escape an embedded apostrophe by doubling it (O''Reilly).

  2. Trim the payload with $select (optional but recommended) to just the fields you need:

    &$select=certificate_number,ownership_name,dba_name,full_business_address,city,state,business_zip,naic_code,naic_code_description,dba_start_date,dba_end_date,location_end_date,administratively_closed

    A $select referencing a column that doesn't exist returns HTTP 400 (a JSON error object, not an array) — use the exact fieldNames in Expected Output below.

  3. Filter to active records (optional). A row is active when it has no end date and is not administratively closed:

    &$where=... AND dba_end_date IS NULL AND location_end_date IS NULL AND administratively_closed IS NULL

  4. Parse the JSON array. Each element is one business location. Map fields:

    • ownership_name → registered owner / legal name
    • dba_name → DBA / trade name
    • full_business_address + city + state + business_zip → address
    • naic_code + naic_code_description → NAICS
    • dba_start_date → registration date (ISO YYYY-MM-DDT00:00:00.000; take the date part)
    • status: administratively_closed present → administratively_closed; else dba_end_date or location_end_date present → closed; else active

  5. Handle multiplicity. A single business commonly returns many rows — one per location and per registration period — all sharing the same certificate_number (Business Account Number). De-duplicate on certificate_number if you want one row per business, or keep all rows to enumerate locations.

  6. Paginate if $limit is hit: append &$offset=N (default page size is 1000). For exact counts, query ?$select=count(*)&$where=....

Browser fallback

Only if the API is unreachable. Open https://data.sfgov.org/Economy-and-Community/Registered-Business-Locations-San-Francisco/g8m3-pdis/about_data, click the "Data" tab (the /data URL itself redirects to the About tab), wait for the grid to render, type the query into the grid's Search box (top-right) and press Enter. The grid does full-text matching (equivalent to $q) and shows a result count ("1 to 27 of 27"). Read values from the row cells. Note: the grid is heavy JS and accessibility snapshots can be flaky mid-render — wait for the row count to settle before extracting. Under the hood the grid calls the internal https://data.sfgov.org/api/v3/views/g8m3-pdis/query.json SoQL endpoint; the public /resource/g8m3-pdis.json path above is the stable equivalent.

Site-Specific Gotchas

  • No proxy / no stealth needed. The SODA API returned 200 from a bare (no-proxy) request. Don't waste a residential proxy or --verified session on it. An app token is optional and only raises rate limits.
  • ownership_name is the owner and the legal name. This dataset has no separate "legal entity name" column distinct from the owner — ownership_name (e.g. "Philz Coffee Inc", "Greg Matt Inc") serves as both. dba_name is the trade name (e.g. "Philz Coffee", "Philz Coffee Truck").
  • Multiple rows per business is normal. Each location is its own row; one business (one certificate_number) can have dozens of rows. "Philz Coffee" returns 27 rows across owners "Philz Coffee Inc" and "Greg Matt Inc" (a franchise/truck operator). De-dupe on certificate_number if you need one record per business.
  • Status is derived from three columns, not one. administratively_closed holds the literal string "***Administratively Closed" (note the *** prefix) for ~42,775 rows, or is absent/NULL for the ~319,170 others. Per the dataset docs, "Administratively Closed" means the business hasn't filed/communicated with the Tax Collector for 3 years (or was flagged closed by another City dept). Separately, dba_end_date marks the business as ended and location_end_date marks a specific location closed (the business may still be active elsewhere). There is no single boolean "active" field — compute it.
  • Don't filter administratively_closed != 'Yes'. The value is never "Yes"; it's "***Administratively Closed" or NULL. Filter with administratively_closed IS NULL for active-only.
  • $select of a non-existent column 400s. A generated query that selects street_address (wrong) instead of full_business_address (correct) returns an HTTP 400 JSON error object, which then fails Array.isArray(). Use the exact field names. Other easy-to-get-wrong names: registration date is dba_start_date (not business_start_date); ZIP is business_zip (not source_zipcode/zip_code).
  • naic_code_description can be empty for some rows even when naic_code is present. Don't assume both are populated.
  • Exact-match dba_name= query param is case-sensitive; SoQL $where ... like is not (when wrapped in upper()). Prefer the upper(...) like '%...%' form for user-supplied queries.
  • $q full-text matches across all columns, so it can return owner-name hits when you intended a DBA search (and vice-versa). For precise DBA-only or owner-only results, use a column-scoped $where.
  • Dates are floating timestamps returned as YYYY-MM-DDT00:00:00.000 (no timezone). Split on T for the calendar date.
  • /data URL redirects to the About tab. In the browser fallback you must click the "Data" tab to reach the grid; landing on /data shows the dataset description, not rows.
  • Dataset scope is broader than "San Francisco". Despite the title, full_business_address/city can be outside SF (e.g. "Palo Alto", "Seattle") — these are SF-registered taxpayers whose physical/mailing location is elsewhere. Filter on city if you need SF-only locations.

Expected Output

A list of matched business locations. Example for a DBA search dba_name like '%PHILZ COFFEE%':

{
  "query": { "mode": "dba_name", "term": "Philz Coffee" },
  "total_matches": 27,
  "results": [
    {
      "certificate_number": "0415140",
      "legal_name": "Philz Coffee Inc",
      "dba_name": "Philz Coffee",
      "address": "549 Castro St, San Francisco, CA 94114",
      "naics_code": "722515",
      "naics_description": "Limited-service restaurants",
      "status": "active",
      "registration_date": "2007-04-01"
    },
    {
      "certificate_number": "0484044",
      "legal_name": "Greg Matt Inc",
      "dba_name": "Philz Coffee Truck",
      "address": "500 Marina Blvd, San Francisco, CA 94123",
      "naics_code": "722330",
      "naics_description": "Mobile food services",
      "status": "active",
      "registration_date": "2013-10-18"
    }
  ]
}

A closed / administratively-closed location:

{
  "certificate_number": "0123456",
  "legal_name": "Example Holdings LLC",
  "dba_name": "Old Corner Store",
  "address": "100 Market St, San Francisco, CA 94105",
  "naics_code": "445110",
  "naics_description": "Supermarkets and other grocery (except convenience) stores",
  "status": "administratively_closed",
  "registration_date": "2009-06-15",
  "dba_end_date": null,
  "location_end_date": "2018-03-07"
}

No matches:

{
  "query": { "mode": "dba_name", "term": "Nonexistent Biz Xyz" },
  "total_matches": 0,
  "results": []
}