City Business License Leads — New Business Lead Feed (API)

A multi-city business-license lead feed. This actor normalizes municipal open data from Chicago, Los Angeles, San Francisco, New York and Seattle into one clean schema with a "new since date" delta, a new-vs-renewal classification, an active-vs-dead licence flag, NAICS / industry tags, owner mailing address and phone where the city publishes one — purpose-built for local-services lead generation. Sourced from official municipal open-data portals (Socrata). Keyless, no anti-bot, no scraping.

What changed in v1.1 (read this before you compare row counts)

v1.1 is a correctness pass. Every number below was measured live on 2026-08-01 against the cities' own APIs, and every one of these produced a plausible wrong answer on a run that reported success:

• Renewals were being sold as new businesses. Chicago publishes an application_type column; v1.0 fetched it and threw it away. 915,200 of Chicago's 1,202,710 records (76.1%) are RENEW — existing businesses renewing, not new ones. On the 1,000 most recent records by issue date it was 721 of 1,000 (72.1%). Every row now carries application_type and record_kind, and onlyNewLicenses (on by default) filters renewals out server-side so you are not billed for them.
• Dead licences were being sold as leads. license_status was also fetched and discarded. Only 43,581 of New York's 69,885 records (62.4%) are Active — the rest are Expired, Surrendered, Failed to Renew, Revoked, Voided, Suspended, or explicitly Out of Business. Chicago carries 78,619 cancelled and 1,461 revoked licences. Every row now carries license_status and is_active_license, and onlyActiveLicenses (on by default) excludes the dead ones server-side.
• Los Angeles returned undated junk on a default run. Socrata's $order <col> DESC defaults to NULL FIRST, and LA has 3,569 records with no location_start_date — so a default LA page was filled with undated records and contained not one recent business. Ordering is now NULL LAST.
• A dead city vanished silently. v1.0 warned and continued, with nothing on any row saying a city was missing, so a five-city run where two cities 404'd produced a three-city dataset that looked complete. Every row now carries city_query_status, and a city that could not be measured emits an explicit row instead of disappearing.
• New York is stale and nothing said so. Its newest record is dated 2026-04-16 and zero records have been created since 2026-07-01. A "since last week" run legitimately returns nothing from New York — which is the city not publishing, not an absence of new businesses. source_stale, source_max_date and source_age_days now say which.
• Contact fields were being fetched and thrown away. Root cause: Socrata omits null columns from row JSON, so a schema read off a sample row under-reports the dataset. v1.1 reads each dataset's metadata instead. Recovered: LA mailing_address (337,260 records, 53.4%) and dba_name (201,059, 31.8%); SF mailing_address_1 (163,876, 44.9%), self_reported_naics_code (127,705, 35.0%) and lic_code_description; New York's unit/suite and address quality.
• Seattle dates were not normalized. license_start_date is a TEXT column in YYYYMMDD form, and v1.0 emitted "20261230" while every other city emitted "2026-12-30". Both the output and the since filter now handle it.

Set onlyNewLicenses: false and onlyActiveLicenses: false to restore the v1.0 firehose. Neither filter can ever drop a row from a city that does not publish the underlying column — those rows come back tagged record_kind: "unknown" and is_active_license: null.

Who it's for

• Local-services lead gen — insurance, POS/merchant services, payroll, business banking, and B2B SaaS that sell to brand-new businesses.
• Sales & RevOps teams building a fresh-leads pipeline of just-licensed companies.
• Data / list providers enriching company records with city, NAICS, mailing address and licence status.
• AI agents needing a "find newly licensed businesses in a city" tool.

What it does

Pulls business licences from each selected city, filters to the ones issued after your since date, classifies each as a new licence or a renewal, drops the dead ones, and returns a normalized record per licence. Run it on a schedule (e.g. daily with since = yesterday) and it becomes a recurring new-business alert you can route straight into your CRM.

Live cities

City	Source dataset	Records	New vs renewal	Licence status	Phone	Mailing address	NAICS
Chicago	Business Licenses (r5kz-chrr)	1,202,710	yes	yes	no	no	no
Los Angeles	Listing of Active Businesses (6rrh-rzua)	631,589	no	no	no	53.4%	yes
San Francisco	Registered Business Locations (g8m3-pdis)	364,774	no	no	no	44.9%	35.0%
New York City	Issued Licenses / DCWP (w7w3-xahh)	69,885	no	yes	72.5%	no	no
Seattle	Active Business License Tax Certificate (wnbq-64tb)	83,856	no	no	99.5%	no	yes

Percentages are the share of records that actually carry a value, measured live 2026-08-01. A blank cell means the city publishes no such column at all, so a null in that field is "not published", not "none on file" — publishes_phone, publishes_license_status and publishes_application_type state this on every row.

Coming soon (scaffolded, NOT wired): San Diego, Boston, Austin, Kansas City, Baton Rouge. Requesting one returns an explicit not_wired row rather than silently nothing.

Example input

{
  "cities": ["chicago", "seattle"],
  "since": "2026-07-01",
  "onlyNewLicenses": true,
  "onlyActiveLicenses": true,
  "naics": "72",
  "maxResults": 1000
}

Input

Field	Type	Default	Notes
cities	string[]	all live	chicago, los-angeles, san-francisco, new-york, seattle
since	string	—	YYYY-MM-DD; only licences issued/started after this date
name	string	—	Business-name fragment, case-insensitive
licenseType	string	—	Licence category fragment (skipped where a city has no type column)
naics	string	—	NAICS prefix (LA, Seattle, and SF self-reported)
onlyNewLicenses	boolean	true	Exclude renewals where the city publishes an application type
onlyActiveLicenses	boolean	true	Exclude expired/revoked/void licences where the city publishes a status
maxResults	integer	1000	Per-city cap on rows returned
socrataAppToken	string (secret)	—	Optional free Socrata token to raise rate limits

Output fields

The lead

city, business_name, dba, license_type, license_description, naics, issue_date, start_date, expiration_date, end_date, address, city_name, state, zip, ward_or_district, latitude, longitude, phone, record_id, mailing_address, mailing_city, mailing_state, mailing_zip, unit_or_suite, address_quality, neighborhood, police_district, community_area, supervisor_district, business_corridor, ownership_type, source_url, retrieved_at.

Lead quality — is this actually a new, live business?

application_type, record_kind (new_license / renewal / change_of_location / other / unknown), license_status, license_status_normalized, is_active_license, business_closed, closed_reason.

City-specific detail

business_activity_id, license_code, application_created_date, payment_date, status_change_date, conditional_approval, special_service_area, parking_tax, transient_occupancy_tax, account_number, site_number, bin, bbl, community_board, census_tract, license_detail, location_description.

Per-city data quality (on every row)

source_city_slug, source_dataset, city_query_status, city_source_error, records_available, records_fetched, is_complete, source_max_date, source_age_days, source_stale, publishes_application_type, publishes_license_status, publishes_phone, data_completeness_note.

Reading null correctly

null means not published by this city. false means published and negative. They are different claims:

• record_kind: "unknown" means the city publishes no new-vs-renewal signal. It is not a claim that the record is new. Only Chicago publishes one today.
• is_active_license: null means the city publishes no status (LA, SF, Seattle). false means the city says the licence is expired, revoked, surrendered or void.
• business_closed: null means no closure signal was published — never a verified "still trading".
• phone: null in Chicago, LA or SF means the dataset has no phone column at all. publishes_phone is false for those cities.
• records_available is the source's own count of records matching your filters. When records_fetched is lower, is_complete is false — raise maxResults.

Fields that populate only on certain inputs, with an input that populates each (all verified live 2026-08-01):

Field(s)	Populating input
city_source_error, city_query_status other than ok	cities: ["atlantis", "boston", "chicago"] — an unknown slug and a scaffolded one
data_completeness_note	any run where a city is partial, stale or empty — e.g. cities: ["new-york"], since: "2026-07-01"
source_stale, source_age_days	cities: ["new-york"] (true today; the dataset is frozen at 2026-04-16)
unit_or_suite (21%), address_quality (69%), bin, bbl (56%), community_board, census_tract, license_detail	cities: ["new-york"], onlyActiveLicenses: false — New York-only columns
is_active_license: false, business_closed, closed_reason	cities: ["new-york"], onlyActiveLicenses: false (4 of the 300 most recent) or cities: ["san-francisco"], onlyActiveLicenses: false
end_date	cities: ["san-francisco"], onlyActiveLicenses: false (2% of recent rows; 238,700 of 364,774 dataset-wide)
record_kind: "renewal", application_type	cities: ["chicago"], onlyNewLicenses: false — 205 of the 300 most recent Chicago records
status_change_date	cities: ["chicago"], onlyActiveLicenses: false — populated on 80,094 of 1,202,710 Chicago records (6.7%), i.e. licences whose status has changed
ownership_type	cities: ["seattle"]
location_description	cities: ["los-angeles"]
mailing_address (80–100% of recent rows)	cities: ["los-angeles", "san-francisco"]
phone	cities: ["seattle"] (100% of recent rows) or ["new-york"] (64%)

What happens when a source breaks

Before any row is written, each selected city is checked live: the dataset still resolves and still holds at least a floor number of records; a negative control (a business name that cannot exist) must match exactly zero records, which catches Socrata silently ignoring the $where; an unfiltered read must return a real record; and every column the mapper depends on must still be present in the dataset's own metadata. A city that fails is excluded loudly and emits an unavailable row. If every city fails, the run fails and nothing is billed.

Use as an MCP tool

This actor works as a clean, chainable MCP tool for AI agents via mcp.apify.com. Every input and output field is described, and record_kind, is_active_license, city_query_status and source_stale let an agent tell a real lead from an unverified one before writing it into a CRM.

FAQ

Why did my row count drop after v1.1? Because onlyNewLicenses and onlyActiveLicenses default to on, and roughly three-quarters of the Chicago dataset is renewals while about 38% of the New York dataset is dead licences. Set both to false for the old behaviour.

Why does New York return nothing for a recent since date? The city's dataset has not been updated since 2026-04-16. source_stale, source_max_date and source_age_days tell you this on every row, and the log says it too.

Which cities give me a phone number? New York (72.5% of records) and Seattle (99.5%). Chicago, LA and SF publish no phone column at all. LA and SF publish an owner mailing address instead (53.4% and 44.9%).

Is this a company database? No — it is a licence feed. A licence is evidence a business registered with a city; it is a screening and prospecting tool, not a verified company record.

How is it priced? Pay-per-event: a small fee per dataset row. A failed run emits no rows.

Related actors

Pairs with the other lead-gen feeds in this catalogue — liquor-licence new openings, childcare provider leads, realtor licence rosters and the city/state business-registry monitors.