US Building Permits Data – 93 Cities & Counties, Contractor & Construction Leads

Fresh building permit data from 93 US cities, counties and states in one clean table – with the work type, trade, residential/commercial class, contractor, ZIP and coordinates already filled in. Find roofing, window, solar, HVAC and remodel jobs the week they are permitted, see which contractors are busiest in your area, or track construction activity by ZIP.

What it does: queries 122 official government open-data feeds in 27 states + Washington, DC, and returns one normalized row per permit. Who it's for: contractors and installers looking for leads, building-material suppliers, real-estate investors and market researchers. Why: every city publishes permits differently; this Actor gives you one schema, one set of filters and an optional "only new / only changed since last run" feed, without logins or scraping private portals.

Measured, not promised (test run over permits issued 2026-10-26 to 2026-11-01, all sources): 33,811 permits from 81 jurisdictions that had activity that week; 96.4 % have a ZIP, 87.8 % coordinates, 91.3 % a specific work category, 79.7 % a residential/commercial class and 42.8 % a contractor name. The whole run took about 2–5 minutes (4.4 minutes on the Apify platform).

What one week looks like (real numbers)

Permits issued Oct 26 – Nov 1, 2026 in five big cities (run on Nov 3, 2026):

City	Permits	New construction	Roofing	HVAC	Solar	Demolition
Los Angeles	3,418	167	352	895	5	38
New York City	2,968	120	74	360	106	82
Austin	1,370	153	10	269	1	40
Seattle	887	38	2	244	0	12
Chicago	583	82	44	13	24	19
Total	9,226	560	482	1,781	136	191

59 % of those permits name a contractor, 99.9 % have a ZIP and 97 % have coordinates. At $5 per 1,000 permits, the whole week for these five cities costs about $46; a filtered list (for example only residential roofing in LA: 341 permits) costs about $1.70.

Sample output (CSV preview)

Real rows from that week (selected columns; the full record is shown further down):

city	permit_number	issue_date	category	property_class	description	zip	valuation	contractor_normalized
Chicago, IL	B200517115	2026-10-27	solar	residential	INSTALL SOLAR PANELS ON ROOF OF PRINCIPAL BUILDING…	60620	14006	Sunrun Installation Services
Seattle, WA	7164089-ME	2026-10-27	hvac	commercial	FURNISH AND INSTALL (3) NEW VAV BOXES AND ASSOCIATED DUCTWORK…	98101	48790	Macdonald Miller Facility Solutions
Seattle, WA	7169407-BP	2026-10-27	hvac	residential	BOILER REPLACEMENT	98108	15475	Carlson Mechanical
New York City, NY	B01325256-I1-GC	2026-10-26	new_construction	residential	PROPOSED NEW 7 STORY BUILDING WITH 26 RESIDENTIAL UNITS…	11230		Brookstone Developers

Coverage at a glance

AZ Eloy, Maricopa County, Mesa, Scottsdale, Tempe, Tucson · CA Atascadero, Campbell, Citrus Heights, Corona, Los Angeles, Marin County, Placentia, Sacramento County, San Diego County, San Francisco, San Jose · CO Boulder, Denver, Douglas County · DC Washington · FL Alachua County, Broward County, Miami, Miami-Dade County, Orlando, Port St. Lucie, Weston · ID Boise · IL Chicago, Cook County (demolition), Rockford, Wheaton · KS Overland Park · KY Louisville · LA Baton Rouge, New Orleans · MA Boston, Cambridge, Somerville, Worcester · MD Baltimore, Baltimore County, Montgomery County · MI Detroit · MN Minneapolis · MO Kansas City, Raymore · NC Brunswick County, Cary, Charlotte/Mecklenburg, Greensboro, Raleigh, Wake County · NJ statewide · NV Henderson · NY New York City · OH Cincinnati, Cleveland, Columbus · OR Bend, Portland, Salem · PA Philadelphia, Pittsburgh · SC Bluffton, Charleston · TN Memphis/Shelby County, Nashville · TX Austin, College Station, Collin County, Dallas, Fort Worth, San Antonio, San Marcos, Tyler · VA Arlington County, Charlottesville, Norfolk, Virginia Beach · WA Bellevue, Everett, Pierce County, Puyallup, Seattle, Tacoma. Full table with source IDs, licenses and freshness: see Supported sources below.

How to use it

1. Click Try for free (or Start). The form is prefilled with 5 big cities, the last 7 days and 100 results – that run takes under a minute.
2. Pick your cities (or "All supported cities") and a date window – leave dates empty for the last 7 days.
3. Optionally add filters: work category (e.g. roofing), trade, residential/commercial, job value, ZIP codes or a radius around your shop.
4. Set Max results to cap your cost, then run. Download the results as Excel, CSV or JSON, or connect them to Google Sheets, Zapier, Make or your CRM.
5. For a lead feed, schedule it daily or weekly with Only new permits switched on so each run returns only permits you haven't seen.

Why this Actor

• Breadth: 93 jurisdictions, including big cities (NYC, LA, Chicago, Dallas, Seattle, Philadelphia…), fast-growing suburbs/counties and statewide New Jersey. Every source passed a freshness check (newest record within 30 days) on 2026-11-03.
• Only new permits, or status changes (incremental modes): schedule it daily or weekly and get only permits you have not seen before (new), only permits whose status changed since the last run (changes, e.g. "In review" → "Issued", with previous_status and changed_at), or both (new_and_changes). The first run just records what exists (outputs and charges nothing).
• Enrichment on every record: category (new construction, addition, alteration/remodel, roofing, windows/doors, siding, HVAC, electrical, plumbing, solar, demolition, pool, fence, sign, other), trades[] + primary trade, property_class (residential / commercial / mixed) with property_class_basis (which signal decided it: occupancy field, permit type, description, address unit, zoning, contractor or valuation), is_new_construction, status_std, cleaned valuation, and contractor_normalized (in the test week "SUNRUN INSTALLATION SVC" in NYC, "SUNRUN INSTALLATION SERVICES INC" in Chicago and "Sunrun Installation Services" in Rockford all become Sunrun Installation Services).
• ZIP + coordinates filled in: a missing ZIP is taken from the address or derived offline from the coordinates (US Census ZCTA boundaries, 96.4 % agreement with the ZIP the city publishes, tested on 18,409 permits). Missing coordinates are geocoded with the free US Census Bureau geocoder (exact matches only). zip_source / geo_source tell you where each value came from.
• Real filters: category, trade, status, residential/commercial, valuation range, ZIP list (or 3-digit prefixes), radius around a point, states, keywords, contractor-present.
• Reliable: per-source timeouts and retries; a source that fails is skipped (the rest of the run continues) and reported in RUN_STATS with error / stale flags.
• Analyst extras: per-city SUMMARY (JSON + CSV) with counts by category/status, total and average valuation and top contractors (grouped by normalized name, with the sources and name variants each covers); optional summary rows in the dataset; CSV-friendly flat output.

Use cases

Who	What they pull	Suggested input
Roofers, window/siding and remodel contractors	New residential exterior jobs near the shop, every morning	categories: ["roofing","windows_doors","siding"], propertyClass: "residential", radius 30 mi, incremental: true
Solar, HVAC, electrical and plumbing installers	Trade permits pulled by competitors, by ZIP	trades: ["solar"] (or hvac, electrical, plumbing), zipCodes
Building-material suppliers and distributors	Which contractors are most active where, for sales prospecting	onlyWithContractor: true, then SUMMARY.top_contractors
Real estate investors and developers	New-construction and demolition activity by ZIP, as a leading indicator	newConstructionOnly: true or categories: ["demolition"]
Market researchers, analysts and journalists	Permit counts and valuations by city, category and month	cities: ["all"], monthly dateFrom/dateTo, SUMMARY.csv
Proptech and data teams	A normalized feed instead of 116 different city APIs	API or scheduled runs into your warehouse, flatOutput: true

For AI assistants

Set "mode": "agent" to use every filter and get a machine-readable coverage record. Permits come first in the dataset; the last item is a record_type: "coverage" record (not charged), also saved in the key-value store as COVERAGE. Each queried jurisdiction gets a status: completed (ran, returned ≥1 permit), no_match (ran, 0 permits matched), or failed (source error or timeout, see error). Jurisdictions not queried are listed under skipped with a reason. Counts are only what the official sources returned; nothing is estimated.

Example input:

{ "mode": "agent", "cities": ["seattle", "denver", "nowhere_city", "chicago"], "states": ["WA", "CO"],
  "trades": ["solar"], "dateFrom": "2026-10-01", "dateTo": "2026-11-01", "maxItems": 25 }

Coverage item from that run (shortened, real test output):

{ "record_type": "coverage", "mode": "agent", "run_status": "completed", "total_results": 25, "limit_reached": true,
  "jurisdictions_queried": 2, "completed_count": 1, "no_match_count": 1, "failed_count": 0,
  "queried": [
    { "source_id": "seattle", "state": "WA", "status": "completed", "result_count": 25, "truncated_by_limit": true, "error": null },
    { "source_id": "denver", "state": "CO", "status": "no_match", "result_count": 0, "truncated_by_limit": false, "error": null } ],
  "skipped": [
    { "source_id": "nowhere_city", "reason": "unknown jurisdiction id (not one of the supported sources)" },
    { "source_id": "chicago", "reason": "not in selected states (WA, CO)" } ] }

If the input itself is invalid (e.g. dateFrom after dateTo), agent mode still writes a coverage record with run_status: "failed" and the error, then the run fails. Human mode (the default, and any input without mode) behaves exactly as before.

Input examples

Weekly lead list: residential exterior work within 40 miles of Chicago, $5k–$100k

{
  "cities": ["all"],
  "lastNDays": 7,
  "categories": ["roofing", "windows_doors", "siding"],
  "propertyClass": "residential",
  "minValuation": 5000,
  "maxValuation": 100000,
  "centerLat": "41.8781", "centerLon": "-87.6298", "radiusMiles": 40
}

Scheduled monitoring – only new permits each day

{
  "cities": ["austin", "dallas", "san_antonio"],
  "lastNDays": 10,
  "incremental": true,
  "incrementalStateKey": "texas-daily",
  "trades": ["hvac", "electrical"]
}

Create a schedule (e.g. every morning) with this input. Run 1 records the current permits and outputs nothing. Every later run outputs only permits that were not returned before. Use a look-back (lastNDays) a bit longer than the schedule interval, because some cities post permits a few days late; already-seen permits are never returned twice. In our test, run 1 (Oct 26–29) recorded 2,974 permits and output 0; run 2 (Oct 26–Nov 1) output only the 776 new permits (all dated Oct 30–Nov 1); run 3 with the same input output 0.

Daily status-change feed (permits that moved from review to issued, finaled, expired…)

{
  "cities": ["kansas_city_mo", "charlottesville"],
  "lastNDays": 60,
  "incremental": true,
  "incrementalMode": "changes",
  "incrementalStateKey": "status-watch"
}

Run 1 stores each permit's current status and outputs nothing. Later runs output only permits whose status differs from the stored one, with change_type: "status_changed", previous_status, previous_status_std and changed_at (the run date the change was detected). new_and_changes combines both feeds (change_type is new or status_changed). Use a long look-back for changes, since a status can change weeks after issue. Older incremental state from v1.2 keeps working.

By ZIP, flat CSV output with per-city summary rows

{ "cities": ["chicago", "los_angeles"], "dateFrom": "2026-10-01", "dateTo": "2026-10-31", "zipCodes": ["606", "90026"], "flatOutput": true, "pushSummaryRecords": true }

If you leave out the dates, the run covers the last 7 days. cities: ["all"] queries every source; with states you can limit it to some states.

Output example (real record from a test run)

{
  "record_type": "permit",
  "source_id": "brunswick_county_nc",
  "city": "Brunswick County",
  "state": "NC",
  "permit_number": "693780-1006",
  "issue_date": "2026-10-30",
  "application_date": null,
  "status": "Active",
  "permit_type": "Commercial - ELECTRICAL",
  "work_class": "Trades",
  "description": "Elect per code/pool",
  "address": "7310 WILBUR MCKEITHAN WAY NE",
  "zip": "28451",
  "lat": 34.18943706,
  "lon": -78.08322261,
  "valuation": 349351,
  "contractor_name": "REEVES ELECTRIC",
  "contractor_license": null,
  "source_url": "https://www.brunswickcountync.gov/",
  "data_source": "ArcGIS: https://services1.arcgis.com/W6gamXPYQeLXrdAd/arcgis/rest/services/Permit_Locations/FeatureServer/0",
  "date_basis": "issued",
  "category": "electrical",
  "trades": ["electrical", "pool"],
  "trade": "electrical",
  "property_class": "commercial",
  "property_class_basis": "permit_type",
  "is_new_construction": false,
  "status_std": "issued",
  "contractor_normalized": "Reeves Electric",
  "zip_source": "source",
  "geo_source": "source"
}

Other fields: distance_miles (radius searches), raw (optional, sanitized source row), and in change modes change_type, previous_status, previous_status_std, changed_at. Dataset views: Permits, Contractor leads, Locations, Status changes. Download as JSON, CSV, Excel or XML.

Key-value store outputs: SUMMARY / SUMMARY.csv (per city), RUN_STATS (per-source scanned / matched / pushed counts, newest date seen, stale, error, geocoder use, incremental state) and contractor stats (unique raw vs normalized names, multi-source contractors, top contractors).

Supported sources (93)

Permit counts are from the local test run over issue dates 2026-10-26..2026-11-01 (33,811 permits) and change week to week. A 0 means the source published nothing with an issue date in that week (it was still fresh within 30 days; Tyler, TX publishes about monthly).

ID	City / county	Portal	License / terms	Permits issued 2026-10-26..11-01	Newest issue date (health check 2026-11-03)
eloy	Eloy, AZ	ArcGIS	City of Eloy open GIS data	1	2026-10-28
maricopa_county	Maricopa County (unincorporated), AZ	ArcGIS	Maricopa County GIS open data terms	162	2026-10-30
mesa	Mesa, AZ	Socrata	City of Mesa Open Data terms	89	2026-11-01
scottsdale	Scottsdale, AZ	ArcGIS	City of Scottsdale open data	5	2026-10-30
tempe	Tempe, AZ	ArcGIS	City of Tempe Open Data (CC BY 4.0)	64	2026-10-30
tucson	Tucson, AZ	ArcGIS	City of Tucson open data	174	2026-11-01
atascadero	Atascadero, CA	ArcGIS	City of Atascadero open GIS data	23	2026-10-30
campbell	Campbell, CA	ArcGIS	City of Campbell open GIS data	45	2026-10-30
citrus_heights	Citrus Heights, CA	ArcGIS	City of Citrus Heights open data	131	2026-11-02
corona	Corona, CA	Socrata	City of Corona CorStat open data	0	2026-10-15
los_angeles_county_rebuild	Los Angeles County (fire rebuild), CA	ArcGIS	LA County Public Works open GIS data (EPIC-LA)	54	2026-11-02
los_angeles	Los Angeles, CA	Socrata	LA Open Data Terms of Use	3418	2026-10-31
marin_county	Marin County, CA	Socrata	Open Data Commons ODbL	75	2026-11-01
placentia	Placentia, CA	ArcGIS	City of Placentia open data	0	2026-10-21
sacramento_county	Sacramento County (unincorporated), CA	ArcGIS	Sacramento County open data	572	2026-10-31
san_diego_county	San Diego County (unincorporated), CA	Socrata	County of San Diego Open Data Terms	188	2026-11-01
san_francisco	San Francisco, CA	Socrata	DataSF Open Data (PDDL)	334	2026-10-30
san_jose	San Jose, CA	CKAN	Creative Commons Zero (CC0)	504	2026-10-31
boulder	Boulder, CO	ArcGIS	City of Boulder Open Data terms	179	2026-10-30
denver	Denver, CO	ArcGIS	Denver Open Data (CC BY 3.0)	140	2026-10-31
douglas_county_co	Douglas County, CO	ArcGIS	Douglas County open data	202	2026-10-30
washington_dc	Washington, DC	ArcGIS	Open Data DC (CC BY 4.0)	841	2026-10-29
alachua_county	Alachua County (unincorporated), FL	ArcGIS	Alachua County open data	205	2026-11-02
broward_county	Broward County (unincorporated), FL	ArcGIS	Broward County GIS open data	10	2026-10-29
miami	Miami, FL	ArcGIS	City of Miami Open Data terms	624	2026-11-01
miami_dade_county	Miami-Dade County, FL	ArcGIS	Miami-Dade County Open Data terms	1451	2026-11-01
orlando	Orlando, FL	Socrata	City of Orlando Open Data Terms	586	2026-11-02
port_st_lucie	Port St. Lucie, FL	ArcGIS	City of Port St. Lucie open GIS data	491	2026-10-31
weston	Weston, FL	ArcGIS	City of Weston open GIS data	72	2026-10-30
atlanta	Atlanta, GA	ArcGIS	City of Atlanta open GIS data (Accela building permits)	212	2026-10-29
boise	Boise (high-impact, new housing & demolition permits), ID	ArcGIS	City of Boise open data	2	2026-10-30
chicago	Chicago, IL	Socrata	City of Chicago Data Portal Terms of Use	583	2026-11-01
cook_county_demolition	Cook County (suburban asbestos & demolition permits), IL	Socrata	Public Domain (Cook County Open Data)	100	2026-11-02
rockford_il	Rockford, IL	Socrata	Illinois EDP open data (City of Rockford)	230	2026-10-31
st_charles_il	St. Charles, IL	ArcGIS	City of St. Charles, IL open GIS data	47	2026-11-02
wheaton	Wheaton, IL	ArcGIS	City of Wheaton open data	0	2026-10-22
overland_park	Overland Park, KS	ArcGIS	City of Overland Park open data	65	2026-10-29
louisville	Louisville (Metro), KY	ArcGIS	Louisville Metro Open Data terms	333	2026-11-02
baton_rouge	Baton Rouge (East Baton Rouge Parish), LA	Socrata	Public Domain	159	2026-10-30
new_orleans	New Orleans, LA	Socrata	CC0 1.0 (Public Domain Dedication)	321	2026-11-02
boston	Boston, MA	CKAN	Open Data Commons PDDL	819	2026-11-02
cambridge_ma	Cambridge, MA	Socrata	ODC Public Domain Dedication and License	122	2026-10-31
somerville	Somerville, MA	Socrata	Open Data Commons ODbL	121	2026-10-29
worcester	Worcester, MA	ArcGIS	City of Worcester open data	378	2026-10-31
baltimore_county_md	Baltimore County, MD	ArcGIS	Baltimore County Government open GIS data	193	2026-11-02
baltimore	Baltimore, MD	ArcGIS	Open Baltimore terms	650	2026-11-01
montgomery_county_md	Montgomery County, MD	Socrata	Public Domain	325	2026-10-28
detroit	Detroit, MI	ArcGIS	Detroit Open Data Portal terms	485	2026-11-01
minneapolis	Minneapolis, MN	ArcGIS	Open Minneapolis terms	945	2026-11-01
kansas_city_mo	Kansas City, MO	Socrata	City of Kansas City, MO Open Data terms of use	876	2026-11-02
raymore	Raymore, MO	ArcGIS	City of Raymore open data	54	2026-10-30
brunswick_county_nc	Brunswick County, NC	ArcGIS	Brunswick County open data	810	2026-10-30
cary	Cary, NC	OpenDataSoft	CC0 1.0 Universal	185	2026-10-30
charlotte	Charlotte / Mecklenburg County, NC	ArcGIS	Mecklenburg County open GIS data	754	2026-11-01
greensboro	Greensboro, NC	ArcGIS	City of Greensboro open GIS data	77	2026-11-02
raleigh	Raleigh, NC	ArcGIS	City of Raleigh Open Data terms	146	2026-10-29
wake_county	Wake County (unincorporated & contract towns), NC	ArcGIS	Wake County open data	109	2026-10-30
new_jersey	New Jersey (statewide, municipality level), NJ	Socrata	Public Domain (NJ DCA)	0	2026-10-19
henderson	Henderson, NV	ArcGIS	City of Henderson open data	38	2026-10-29
new_york_city	New York City, NY	Socrata	NYC Open Data Terms of Use	2968	2026-10-30
cincinnati	Cincinnati, OH	Socrata	Public Domain	201	2026-10-30
cleveland	Cleveland, OH	ArcGIS	City of Cleveland Open Data terms	299	2026-10-31
columbus	Columbus, OH	ArcGIS	City of Columbus Open Data terms	898	2026-10-31
bend	Bend, OR	ArcGIS	City of Bend Open Data terms	109	2026-11-01
jackson_county_or	Jackson County, OR	ArcGIS	Jackson County, OR open GIS data	76	2026-11-02
portland_or	Portland, OR	ArcGIS	City of Portland open data (PortlandMaps)	2	2026-10-28
salem	Salem, OR	ArcGIS	City of Salem open data	24	2026-10-30
philadelphia	Philadelphia, PA	CARTO	OpenDataPhilly / City of Philadelphia terms	964	2026-11-02
pittsburgh	Pittsburgh, PA	CKAN	WPRDC (CC BY 4.0 / ODC)	175	2026-10-30
bluffton_sc	Bluffton, SC	ArcGIS	Town of Bluffton open GIS data	23	2026-10-30
charleston_sc	Charleston (new construction only), SC	ArcGIS	City of Charleston open data	0	2026-10-22
memphis	Memphis / Shelby County, TN	ArcGIS	Memphis Data Hub terms	154	2026-10-30
nashville	Nashville (Metro Davidson County), TN	ArcGIS	Nashville Open Data terms	208	2026-11-01
austin	Austin, TX	Socrata	Public Domain U.S. Government	1370	2026-11-01
college_station	College Station, TX	Socrata	City of College Station Open Data Terms	117	2026-10-30
collin_county	Collin County (Collin CAD, all cities), TX	Socrata	Texas Open Data Portal terms	13	2026-10-31
dallas	Dallas, TX	Socrata	Dallas OpenData Terms of Use	1499	2026-11-01
fort_worth	Fort Worth, TX	ArcGIS	City of Fort Worth open data terms	1109	2026-11-02
harris_county_tx	Harris County (unincorporated), TX	ArcGIS	Harris County Engineering open GIS data (ePermits)	696	2026-11-03
san_antonio	San Antonio, TX	CKAN	City of San Antonio Open Data terms	1300	2026-10-30
san_marcos	San Marcos, TX	ArcGIS	City of San Marcos open data	63	2026-10-31
tyler_tx	Tyler, TX	ArcGIS	City of Tyler open GIS data	0	2026-10-09
arlington_county_va	Arlington County, VA	Arlington Open Data API	Arlington County Open Data policy	304	2026-11-01
charlottesville	Charlottesville, VA	ArcGIS	City of Charlottesville open data	82	2026-11-01
norfolk	Norfolk, VA	Socrata	Norfolk Open Data Terms of Use	268	2026-11-01
virginia_beach	Virginia Beach, VA	ArcGIS	City of Virginia Beach Open Data terms	584	2026-10-30
bellevue	Bellevue, WA	ArcGIS	City of Bellevue Open Data terms	189	2026-11-01
edgewood_wa	Edgewood, WA	ArcGIS	City of Edgewood, WA open GIS data	3	2026-10-30
everett	Everett, WA	Socrata	City of Everett Open Data Terms	228	2026-10-30
pierce_county	Pierce County (unincorporated), WA	Socrata	Pierce County Open Data Terms	139	2026-10-28
puyallup	Puyallup, WA	ArcGIS	City of Puyallup open GIS data	29	2026-10-30
seattle	Seattle, WA	Socrata	Public Domain	887	2026-10-31
tacoma	Tacoma, WA	ArcGIS	City of Tacoma open data	139	2026-10-30

Some fields simply aren't published by some sources: for example Philadelphia, Maricopa County, Washington DC and Virginia Beach publish no valuation, and New Jersey publishes no street address (municipality and block/lot only, and its newest dates lag 2–4 weeks). Where a feed mixes company and private-person names in its contractor field, only business-looking names are returned. Missing values come back as null; nothing is invented. date_basis is approved for Cook County (approval date) and issued elsewhere.

Pricing

You pay only for results: one charge per permit returned (event permit-result). Price: $5 per 1,000 permits ($0.005 each) – so the prefilled 100-permit example costs about $0.50, and a 1,000-permit weekly lead list about $5. The current price is always shown on the Pricing tab. There is no separate charge for compute or proxies.

• Summary rows and the first "record only" monitoring run are free.
• Set Max results and/or a maximum cost per run in the run options; the Actor stops cleanly at that limit, so you never pay more than you chose.
• Apify's free plan includes monthly platform credits you can use to try the Actor.

Notes and limits

• Accuracy and freshness depend on each government feed. source_url links the original.
• Category, trade and property class are derived from the permit type and description text with rules, so they are very good but not perfect. property_class uses the source's occupancy/use field first, then permit type and description, then address unit, zoning, known residential installers and very large valuations; it is null when nothing gives a clear signal (20.3 % of records in the test week, down from 34.1 % in v1.2 on the same sources; most remaining blanks are NYC sidewalk sheds/scaffolds, San Antonio and Fort Worth trade permits and right-of-way work). The propertyClass filter drops those.
• Derived ZIPs (zip_source: zcta_derived) use Census ZIP Code Tabulation Areas, which approximate USPS ZIPs.
• contractor_normalized is a rules-based clean-up (case, punctuation, legal suffixes such as LLC/Inc. dropped, DBA, trailing license numbers, abbreviations such as Svc/Plbg/Elec, placeholders); some acronyms written in capitals may come out title-cased; it never merges different companies on fuzzy similarity, and personal-looking names are still dropped.
• Multi-row permits are deduplicated (one result per permit, date and filing).

FAQ

How fresh is the data? Each run queries the government feeds live, so you get whatever the city has published at that moment. Most cities publish daily; some lag a few days, and New Jersey lags 2–4 weeks. On 2026-11-03 every one of the 87 sources had a record from the last 30 days. Each run's RUN_STATS record shows the newest date per source and flags stale sources.

Which areas are covered? The 93 jurisdictions in the table above. Coverage is limited to places that publish permits through a public open-data API, so many cities (and most rural counties) are not available. Some fields aren't published everywhere, for example valuation or contractor name – see "Notes and limits".

Is this legal? The Actor only reads official open-data APIs that governments publish for public reuse. It does not log in, bypass CAPTCHAs or scrape private portals. Each source has its own license (linked in the table); see "Data sources and attribution" below.

Does it return personal data? No owner, applicant, phone, email or staff fields are returned, and they are stripped even from the optional raw record. Contractor fields are limited to business names and license numbers; names that look like private individuals are dropped. If you combine this data with other sources, you are responsible for complying with privacy and marketing laws (e.g. CAN-SPAM, TCPA, CCPA).

Why are some fields empty? If a city doesn't publish a value, it comes back as null – nothing is guessed or invented. Derived values (category, trade, property class, derived ZIP, geocoded coordinates) are labeled in property_class_basis, zip_source and geo_source.

How much does a typical run cost? $5 per 1,000 permits returned. Examples from the Oct 26 – Nov 1, 2026 week: all of Chicago (583 permits) ≈ $2.90; all of Los Angeles (3,418) ≈ $17; LA residential roofing only (341) ≈ $1.70. Filters are applied before you are charged, so narrow filters cost less. Set Max results or a maximum cost per run to cap spend.

Does it include homeowner names, phone numbers or emails? No. You get the permit, address, work description, valuation and the contractor's business name (where the city publishes it). Owner and applicant fields are removed.

What counts as "new construction"? is_new_construction is true when the permit's category is new_construction, which the rules assign from the permit type, work class and description (for example "New Construction of a 2-Story Single Family Res."); additions and remodels have their own categories. In the sample week, 560 of 9,226 permits in five big cities were new construction.

Can I get only new permits every day? Yes: schedule the Actor with incremental: true. Each run returns only permits it hasn't returned before, so a daily schedule is a daily lead list.

How do I get the data into Google Sheets, a CRM or my own code? Download as CSV/Excel/JSON from the run, use Apify's Google Sheets, Zapier or Make integrations, or call the Actor from the Apify API / Python and JavaScript clients and read the dataset.

How is this different from scraping a city permit portal? It doesn't touch Accela/EnerGov web portals or log in anywhere; it reads the official open-data APIs (Socrata, ArcGIS, CKAN, CARTO, OpenDataSoft), so runs are fast (a week of 5 big cities takes well under a minute) and don't break when a portal redesigns.

Can you add my city? Open an issue on the Issues tab with the city name and, if you know it, the link to its open-data permit dataset. Only sources with a public open-data API can be added.

Data sources and attribution

Data is provided by the cities, counties and states listed in the table above, under each publisher's open-data license or terms of use (linked per source). Every record includes data_source (portal and dataset id) and source_url (the original record or dataset) for attribution. Some sources require attribution when you republish the data, for example CC BY (Tempe, Denver, Washington DC, Pittsburgh/WPRDC) and ODbL (Marin County, Somerville, which also requires share-alike for databases built from it). If you republish, credit the publisher, e.g. "Source: City of Denver Open Data, CC BY 3.0", and follow the linked terms. This Actor is not affiliated with or endorsed by any government agency.

Disclaimer

Permit data is provided "as is" from public sources and may be incomplete, delayed or contain errors made by the publisher. Derived fields (category, trade, property class, normalized contractor name, derived ZIP, geocoded coordinates) are produced by rules and can be wrong. A permit is evidence of a project, not an exclusive lead or a promise of work. Verify important details with the issuing agency before relying on them.

Support

Questions, a broken source or a city you'd like added? Open an issue on the Issues tab of this Actor.

Changelog

• 1.6 (Nov 2026) – Two input modes. Human mode (default): simpler form (place, permit type/trade, dates, max results) with advanced options in collapsed sections. AI/Agent mode ("mode": "agent"): coverage record (COVERAGE) with per-jurisdiction status codes completed / no_match / failed and skipped reasons. Existing inputs, output fields and pricing are unchanged.
• 1.5 (Nov 2026) – 6 new sources (93 total): Atlanta GA, Harris County TX (unincorporated), St. Charles IL, Jackson County OR, Edgewood WA, LA County fire-rebuild permits. Week counts for the new rows are from a 2026-11-04 test of the same week.
• 1.4.6 – README: one-week sample numbers, CSV preview, coverage by state, more FAQ. No code changes.
• 1.4.5 – Added an Actor output schema (Output tab links to permits, contractor leads, per-source summary and run stats).
• 1.3 (Nov 2026) – 87 sources (116 datasets) in 26 states + DC; status-change monitoring (changes, new_and_changes); better residential/commercial classification; consistent normalized contractor names; offline ZIP derivation from coordinates; null-like text values ("NA", "None") cleaned to null; simpler input form with a quick prefilled example.
• 1.2 (Nov 2026) – Expanded coverage and enrichment (more cities, normalized contractor names, location filters).
• 1.0 (Nov 2026) – First version: 22 cities in one normalized schema.