Coldwell Banker Scraper [$2/1K💰] | Listings, Agents & Offices

Coldwell Banker Scraper

Pull US property listings, listing-agent contact details and full office/agent directories from coldwellbanker.com into three separate datasets — 75 listing fields, 26 agent fields, 27 office fields, including price, beds, sqft, lot size, open-house schedules, MLS attribution and every agent's email and phone.

Built for real-estate investors, proptech data teams, mortgage and insurance lead-gen, and agent recruiters.

Before you publish or resell anything: Coldwell Banker listing data is MLS-derived. That carries redistribution restrictions independent of this Actor. Read Legal & Compliance first.

What you get

A whole-city price snapshot for one request per 32 listings

This is the thing that makes large runs affordable. Alongside normal listing rows, the Actor emits listingSummary rows — one per city-search page, each holding a listings[] array of cheap search hits (price, beds, baths, sqft, type, MLS, status, lat/lng, first photo), plus total, the number of matching listings city-wide.

Those summary rows are not billed. A 1,000-listing market costs about 32 requests for the whole city, not 1,000.

Listing detail rich enough for valuation models

price, originalPrice, bathsDecimal, sqft, lotSizeSqft, lotSizeAcres, lotDimensions, yearBuilt, garageSpaces, parkingSpaces, stories, architecturalStyle, flooring, heating, basement, hasFireplace, isNewConstruction, daysOnMarket and listingContractDate — the fields a comps or pricing model actually needs, all in one row.

Open-house schedules as structured data

openHouses[] carries date, startTime, endTime, byAppointment and showOnInternet per event, plus a hasOpenHouse boolean. That is the field for agent-scheduling and open-house lead-gen, and you can filter for it server-side with the hasOpenHouses input instead of filtering 1,000 rows yourself.

Listing-agent contact details with honest coverage

listingAgentName, listingAgentPhone, listingAgentEmail, listingAgentMlsId, coListingAgentName, coListingAgentPhone, officeName, officePhone, officeMlsId. Agent phone is MLS-dependent — see Why is listingAgentPhone sometimes null? before you rely on it.

Agent and office rosters nobody derives from listings

A listing page's embedded payload never links to an agent or office profile, so agents and offices are not obtainable from a listing search. This Actor takes them directly: 26 agent fields (bio, specialties, languages, activeListingsCount, soldListingsCount, email, phone) and 27 office fields (totalAgents, areasServed[], agentsServed, agents[], totalStaff), via seed URLs or bulk sitemap discovery.

How it works

coldwellbanker.com is a Next.js application. Every route server-renders its entire dataset for that route into a single inline script tag:

<script id="__NEXT_DATA__" type="application/json">{ … pageProps … }</script>

So the Actor downloads plain HTML with CheerioCrawler — plain HTTP, no JavaScript execution — and JSON.parses that blob. Consequences:

• No browser. Roughly 10× cheaper than a Playwright run and about 10× faster, because there is no browser boot and no client-side render wait.
• No anti-bot to bypass. There is no Cloudflare, Turnstile, DataDome or fingerprint challenge to work around, because the data is already in the first HTTP response.

Purpose	URL	Data used
City listing search	/city/{state}/{city}	propertiesData.properties[] (32/page), .total, .paginationMetadata
Property detail	/{state}/{city}/{address}/lid-{id}	pageProps.propertyDetails + rawPropertyDetails
Agent profile	/{state}/{city}/agents/{name}/aid-{id}	pageProps.detail.agentDetails
Office directory	/{state}/{city}/offices/{slug}/oid-{id}	pageProps.detail
Listing sitemaps	/xml-sitemap/states/sitemapindex-listings.xml	<loc> — 2,267 per-state sitemaps
Agent sitemaps	/xml-sitemap/states/sitemapindex-agents.xml	<loc>
Office sitemaps	/xml-sitemap/states/sitemapindex-offices.xml	<loc>

Every one of those sitemaps is declared in robots.txt, which for User-agent: * contains no Disallow rules at all — only those six Sitemap: lines.

Search filters all apply server-side: minPrice, maxPrice, minBedrooms, maxBedrooms, minBathrooms, minSqft, maxSqft, propertyType (SFR, CONDO, COOP, TOWNHOUSE, MFR, MFD, LAND, FARM), status (ACTIVE, PENDING, COMING_SOON, …), hasOpenHouses, isRental, isNewConstruction, sortBy, keyword, page.

Why sitemap discovery is budgeted per kind

Bulk enumeration is where scrapers usually cannibalise themselves. Here each child sitemap queues at most maxListings/maxAgents/maxOffices URLs minus what has already been collected, so listing discovery can never drain the request budget and starve agent or office collection. Search results seeded from searchLocations always process first, and listing detail pages are dequeued with forefront, so bulk discovery never delays your primary output.

Stale search entries don't eat your quota

City search indexes occasionally list properties whose detail page 404s (the page renders listingNotFoundContent and carries no payload). The Actor detects this, logs Listing not found (stale search entry), and backfills the slot from the rest of the same search page — so a dead entry never consumes maxListings. Expect roughly 1 stale hit per 15–30 listings on fast-moving markets.

Input

Field	Type	Default	Notes
searchLocations	string[]	["ny/new-york"]	City slugs in the form "<state>/<city>", e.g. ny/new-york, ca/san-francisco.
startUrls	string[]	[]	Direct listing, agent or office URLs. Route is auto-detected.
includeDetails	boolean	true	Expand each search hit to a detail request for the full 75-field row.
minPrice / maxPrice	integer	—	Price range, applied server-side.
minBedrooms / maxBedrooms	integer	—	Bedroom range.
minBathrooms	integer	—	Minimum bathrooms.
minSqft / maxSqft	integer	—	Living-area range.
propertyType	string[]	—	Any of SFR, CONDO, COOP, TOWNHOUSE, MFR, MFD, LAND, FARM.
status	string[]	—	Any of ACTIVE, PENDING, COMING_SOON, …
hasOpenHouses	boolean	—	Only listings with a published open-home schedule.
isRental / isNewConstruction	boolean	—	Boolean filters.
sortBy	string	—	newest, price_asc, price_desc, living_area_desc.
keyword	string	—	Free-text keyword pass-through.
maxListings	integer	50	Hard cap on listing records.
maxPagesPerLocation	integer	2	32 listings per page.
sitemapDiscovery	object	{ enabled: false, states: [], kinds: ["listings"], maxUrls: 100 }	Bulk-enumerate from XML sitemaps. kinds: listings, agents, offices.
maxAgents	integer	20	Cap on agent records.
maxOffices	integer	20	Cap on office records.
maxRequestsPerCrawl	integer	300	Runaway guard.
proxy	object	{ useApifyProxy: false }	Standard Apify proxy picker.

Agents and offices need their own source

searchLocations alone will always yield agents: 0, offices: 0, because a listing page's payload never links to an agent or office profile. Use one of:

• put agent/office URLs (or a whole directory URL) in startUrls — the router detects the route;
• or set sitemapDiscovery: { enabled: true, kinds: ["agents","offices"], states: ["ny"] }.

If you set maxAgents/maxOffices without a source, the Actor logs an explicit warning rather than silently returning nothing.

Output

Three datasets, so each record type validates, exports and bills independently.

Dataset	Alias	Contents
Listings	default	listing, listingSummary
Agents	agents	agent
Offices	offices	office

Run totals are not a dataset record — they go to the key-value store as RUN_SUMMARY.json (listings, agents, offices, billed, sitemapUrlsQueued, pagesFetched, isPaying, finishedAt, aborted), so a summary row is never billed as data.

Listings (default) — 75 fields

type, listingId, mlsNumber, mlsSource, mlsName, url, price, originalPrice, currency, priceText, isPriceHidden, isSoldListing, address, city, state, stateName, zip, county, fullAddress, latitude, longitude, neighborhood, beds, baths, bathsDecimal, bathsFull, sqft, lotSizeSqft, lotSizeAcres, lotDimensions, yearBuilt, garageSpaces, parkingSpaces, stories, architecturalStyle, flooring, heating, basement, hasFireplace, propertyType, isNewConstruction, isRental, description, photos, photoCount, openHouses, hasOpenHouse, listingAgentName, listingAgentPhone, listingAgentEmail, listingAgentMlsId, coListingAgentName, coListingAgentPhone, officeName, officePhone, officeMlsId, status, displayStatus, isComingSoon, isVirtualTourAvailable, listingContractDate, firstAppearedDate, daysOnMarket, lastUpdated, mlsDisclaimer, idxFeedAttributionCompany, source, detailFetched, scrapedAt, location, searchUrl, placeMasterId, page, total, listings

Views: Overview, Details, Search pages.

Agents (agents) — 26 fields

type, agentId, name, firstName, lastName, email, phone, photo, bio, specialties, languages, officeName, officeId, officeUrl, officePhone, address, city, state, stateName, zip, latitude, longitude, activeListingsCount, soldListingsCount, url, source

Views: Overview, Profiles.

Offices (offices) — 27 fields

type, officeId, name, companyName, brandCode, phone, email, website, address, addressLine2, city, state, zip, country, latitude, longitude, photo, logo, areasServed, totalAgents, totalStaff, agentsServed, agents, activeListingsCount, soldListingsCount, url, source

Views: Overview, Directories, Agent roster.

Field notes that change how you read the data:

• type is your record discriminator: listing, listingSummary, agent or office. Filter on it before loading to a warehouse.
• sqft falls back through structure.livingArea → buildingAreaTotal → the SSR display string. Land, farm and vacant listings legitimately have none, so null is normal.
• openHouses is [] unless the MLS published a schedule. That is an empty schedule, not a missing field.
• officeName/officePhone on a listing come from the MLS attribution block, so they name the listing brokerage — often an independent franchise, not Coldwell Banker. Use type: "office" rows for true CB office directories.
• Agent source matters: a directory page (/city/{state}/{city}/agents, 12 agents per request) gives name, email, phone, photo, office and URL; a profile page adds bio, specialties, languages, activeListingsCount and soldListingsCount. Directory rows leave the profile-only fields null. Seed profile URLs when you need the extras.

Example input

City search with filters:

{
  "searchLocations": ["ny/new-york"],
  "propertyType": ["SFR"],
  "minBedrooms": 3,
  "maxPrice": 2500000,
  "status": ["ACTIVE"],
  "maxPagesPerLocation": 3,
  "maxListings": 100
}

Cheap whole-city price snapshot — no detail requests, summary rows only:

{
  "searchLocations": ["ca/san-francisco", "ny/brooklyn"],
  "includeDetails": false,
  "maxPagesPerLocation": 10,
  "maxListings": 2000
}

Agent roster for recruiting:

{
  "searchLocations": [],
  "sitemapDiscovery": { "enabled": true, "states": ["ny"], "kinds": ["agents"], "maxUrls": 300 },
  "maxAgents": 300,
  "maxRequestsPerCrawl": 400
}

Specific listing, agent and office pages:

{
  "searchLocations": [],
  "startUrls": [
    "https://www.coldwellbanker.com/ny/springfield-gardens/21626-131st-ave/lid-P00800000HI8DebheIrz6ty0mDIFY6RivDTTN4Mm",
    "https://www.coldwellbanker.com/ny/brooklyn/agents/anthony-marra/aid-P00200000FDdqmuwdqxLZyHkaMD67yyZH1zRejeU"
  ]
}

Example output

{
  "type": "listing",
  "listingId": "P00800000HI8DebheIrz6ty0mDIFY6RivDTTN4Mm",
  "mlsNumber": "1234567890",
  "mlsSource": "onekey",
  "mlsName": "OneKey MLS",
  "url": "https://www.coldwellbanker.com/ny/springfield-gardens/21626-131st-ave/lid-P00800000HI8DebheIrz6ty0mDIFY6RivDTTN4Mm",
  "price": 685000,
  "originalPrice": 725000,
  "currency": "USD",
  "priceText": "$685,000",
  "isPriceHidden": false,
  "isSoldListing": false,
  "address": "21626 131st Ave",
  "city": "Springfield Gardens",
  "state": "NY",
  "stateName": "New York",
  "zip": "11012",
  "county": "Queens",
  "fullAddress": "21626 131st Ave, Springfield Gardens, NY 11012",
  "latitude": 40.65412,
  "longitude": -73.74431,
  "neighborhood": "Springfield Gardens",
  "beds": 4,
  "baths": 3,
  "bathsDecimal": 2.5,
  "bathsFull": "2.5",
  "sqft": 2340,
  "lotSizeSqft": 4800,
  "lotSizeAcres": 0.1101,
  "lotDimensions": "60x80",
  "yearBuilt": 1954,
  "garageSpaces": 1,
  "parkingSpaces": 1,
  "stories": 2,
  "architecturalStyle": "Cape Cod",
  "flooring": "Hardwood",
  "heating": "Forced Air",
  "basement": "Full",
  "hasFireplace": true,
  "propertyType": "SFR",
  "isNewConstruction": false,
  "isRental": false,
  "description": "Well-maintained Cape Cod on a quiet block, finished basement, updated roof 2021.",
  "photos": ["https://imagecache.coldwellbanker.com/1234567890/photo-01.jpg"],
  "photoCount": 28,
  "openHouses": [
    { "date": "2026-10-11", "startTime": "12:00 PM", "endTime": "02:00 PM", "byAppointment": false, "showOnInternet": true }
  ],
  "hasOpenHouse": true,
  "listingAgentName": "Anthony Marra",
  "listingAgentPhone": null,
  "listingAgentEmail": "anthony.marra@cbrealty.com",
  "listingAgentMlsId": "P00200000FDdqmuwdqxLZyHkaMD67yyZH1zRejeU",
  "coListingAgentName": null,
  "coListingAgentPhone": null,
  "officeName": "Marra Realty Group",
  "officePhone": "(718) 555-0132",
  "officeMlsId": "OFF-7781",
  "status": "ACTIVE",
  "displayStatus": "Active",
  "isComingSoon": false,
  "isVirtualTourAvailable": true,
  "listingContractDate": "2026-09-18",
  "firstAppearedDate": "2026-09-19",
  "daysOnMarket": 21,
  "lastUpdated": "2026-10-04T06:12:00.000Z",
  "mlsDisclaimer": "The data is deemed reliable but is not guaranteed.",
  "idxFeedAttributionCompany": "OneKey MLS",
  "source": "search",
  "detailFetched": true,
  "scrapedAt": "2026-10-04T06:12:41.220Z"
}

Integrations & automation

Call it over the Apify API (POST /v2/acts/1cOxtsUkKezlOoke4/runs), or from Python, JavaScript, the CLI or MCP. All three datasets export as JSON, CSV, Excel, XML or RSS.

For market monitoring, includeDetails: false plus a weekly schedule over a list of searchLocations gives you city-level price and inventory movement cheaply — the listingSummary rows are unbilled, so this is the cheap path. For agent lead-gen, run sitemap discovery per state monthly and upsert on agentId; agent rosters change slowly.

Typical pipeline: run on schedule → split datasets on type → upsert listings on listingId and agents on agentId → alert when a tracked mlsNumber goes ACTIVE → PENDING, when daysOnMarket passes a threshold, or when a tracked neighbourhood's median price/sqft moves more than ~5%. PostGIS works directly: latitude/longitude are on listings, agents and offices.

What a run costs

Configuration	Events	Cost
City snapshot, includeDetails: false, 3 pages	summary rows only — unbilled	$0
50 listing detail rows	50 × listing	$0.10
300 agent rows	300 × agent	$1.80
200 office rows	200 × office	$1.00

Free accounts get a 10-listing preview; paying accounts get the full maxListings. listingSummary rows are never charged — one row is a whole page of results, not a single listing.

Related Actors

Actor	Why
century21-scraper	The same problem on the other big US brokerage brand — the closest structural comparison in this portfolio.
cargurus-scraper	US vehicle-marketplace listings with seller and vehicle detail data.
cars-com-scraper	Same US-marketplace shape, different vertical — useful for schema and price-comparison patterns.
housinganywhere-scraper	Long-term rental listings in the UK and Europe, for geographic expansion.
wallapop-scraper	Marketplace listings in Spain, with the same listing-detail + seller structure.

FAQ

Why not use the official Coldwell Banker API?

Because there isn't a usable one for this data. Coldwell Banker is built on an internal real-estate platform (api.anywhere.re) whose public runtime config leaks an API key and a /properties route, but that key is scoped and rotated — calling it directly returns 401 {"statusCode":401,"message":"Invalid ApiKey for given resource"}. What that internal endpoint would give you is the same data already present in the server-rendered page. So the Actor reads the server-rendered JSON instead: no key, no auth, more robust.

Agents and offices come back as zero. Why?

Because a listing page's payload never links to an agent or office profile, so a listing search cannot produce them. Supply agent or office URLs in startUrls, or enable sitemapDiscovery with kinds: ["agents","offices"]. Setting maxAgents/maxOffices without a source logs an explicit warning rather than silently returning nothing.

Why is listingAgentPhone sometimes null?

It is MLS-dependent, and the Actor will not invent one. Midwestern feeds such as Midwest Real Estate Data (IL, WI) populate it; New York's OneKey MLS feed does not ship the field at all. Agent email and office phone are populated in both. If you need guaranteed agent phone numbers, scrape agent profile pages — those always carry a business phone number.

Why do some listings have no sqft?

Because sqft is resolved through a fallback chain ending at the rendered display string, and land, farm and vacant listings genuinely have no living area. null is the correct value there, not a parse failure.

Why is officeName on a listing not "Coldwell Banker"?

Because it comes from the MLS attribution block, so it names the brokerage that actually holds the listing — frequently an independent franchise. For true CB office directories, read the offices dataset instead.

Why am I getting fewer listings than maxListings?

Three causes, all normal: New York's OneKey index limits how many listings a city search returns; maxPagesPerLocation may be too low; and stale search entries whose detail page 404s are detected and skipped (with the slot backfilled where possible). Expect roughly 1 stale hit per 15–30 listings on fast-moving markets — the log line is Listing not found (stale search entry).

Do I need a browser or a proxy?

No. The data is in the first HTML response, so CheerioCrawler with plain HTTP is enough. There is no Cloudflare, DataDome, Akamai or fingerprint challenge on this target. Proxy defaults to off; enable it only if you plan high-volume runs.

SEO Keywords

coldwell banker scraper, coldwellbanker.com, us real estate listing data, real estate leads, listing agent contact data, property listing api, open house schedule data, mls listing data, house for sale data, realtor directory data, real estate agent directory, real estate office directory, proptech data pipeline, comps data, us property market research, real estate price monitoring, listing photos and details, property valuation data, agent lead generation, mortgage lead generation, real estate market analysis, bulk property scraping

For AI Agents & LLM Apps

Returns Coldwell Banker data as three flat datasets — default (listings + page summaries), agents, offices — discriminated by a type field.

Minimal input — smallest input that produces a valid run:

{ "maxListings": 25 }

Cheapest useful run — whole-city snapshot, no detail requests, summary rows are unbilled:

{ "searchLocations": ["ny/new-york"], "includeDetails": false, "maxPagesPerLocation": 5 }

Filtered listing search:

{ "searchLocations": ["ny/new-york"], "propertyType": ["SFR"], "minBedrooms": 3, "status": ["ACTIVE"], "maxListings": 100 }

Agent roster (requires sitemap discovery or agent URLs — listings alone yield zero agents):

{ "searchLocations": [], "sitemapDiscovery": { "enabled": true, "states": ["ny"], "kinds": ["agents"] }, "maxAgents": 200 }

Output fields — listings (75): type, listingId, mlsNumber, mlsSource, mlsName, url, price, originalPrice, currency, priceText, isPriceHidden, isSoldListing, address, city, state, stateName, zip, county, fullAddress, latitude, longitude, neighborhood, beds, baths, bathsDecimal, bathsFull, sqft, lotSizeSqft, lotSizeAcres, lotDimensions, yearBuilt, garageSpaces, parkingSpaces, stories, architecturalStyle, flooring, heating, basement, hasFireplace, propertyType, isNewConstruction, isRental, description, photos, photoCount, openHouses, hasOpenHouse, listingAgentName, listingAgentPhone, listingAgentEmail, listingAgentMlsId, coListingAgentName, coListingAgentPhone, officeName, officePhone, officeMlsId, status, displayStatus, isComingSoon, isVirtualTourAvailable, listingContractDate, firstAppearedDate, daysOnMarket, lastUpdated, mlsDisclaimer, idxFeedAttributionCompany, source, detailFetched, scrapedAt, location, searchUrl, placeMasterId, page, total, listings

Output fields — agents (26): type, agentId, name, firstName, lastName, email, phone, photo, bio, specialties, languages, officeName, officeId, officeUrl, officePhone, address, city, state, stateName, zip, latitude, longitude, activeListingsCount, soldListingsCount, url, source

Output fields — offices (27): type, officeId, name, companyName, brandCode, phone, email, website, address, addressLine2, city, state, zip, country, latitude, longitude, photo, logo, areasServed, totalAgents, totalStaff, agentsServed, agents, activeListingsCount, soldListingsCount, url, source

Behaviours an agent should know

• Three datasets, one type field. Always filter on type (listing, listingSummary, agent, office) before processing — the default dataset mixes full listings with per-page summary rows.
• listingSummary rows are not billed. Each holds a listings[] array plus total (city-wide match count). This is the cheapest way to get market-level price data: one request per 32 listings.
• Agents and offices require their own source. searchLocations alone always yields agents: 0, offices: 0. Use startUrls with agent/office URLs, or sitemapDiscovery.kinds: ["agents","offices"]. Otherwise the run logs an explicit warning.
• includeDetails defaults to true. Setting it false skips detail pages entirely — much cheaper, but you get only search-level fields.
• listingAgentPhone is MLS-dependent and may be null (OneKey MLS does not ship it). listingAgentEmail and officePhone are reliable. Agent profile rows always have a phone.
• Stale search entries (detail page 404s) are detected and skipped, and the slot is backfilled from the same page where possible — so counts stay close to maxListings without exceeding it.
• Billing is per-event: $0.002 per listing, $0.006 per agent, $0.005 per office. listingSummary rows cost nothing. Agents are the most expensive record type by far — check the tiered pricing table for paid-plan rates before scraping a full state roster.
• Free accounts are capped at 10 listings; maxAgents/maxOffices are not plan-capped.

Legal & Compliance

This Actor is an independent third-party tool. It is not affiliated with, endorsed by, or sponsored by Coldwell Banker, by any Coldwell Banker franchisee, or by any MLS.

What it actually accesses: publicly visible Coldwell Banker listing, agent and office pages over plain HTTP, plus the XML sitemaps that robots.txt explicitly declares. No login, no authentication bypass, no captcha solving, no headless browser. robots.txt for User-agent: * contains no Disallow rules — only Sitemap: declarations.

The real constraint is MLS data licensing, not robots.txt. Coldwell Banker listing data is sourced from regional Multiple Listing Services (this Actor surfaces mlsSource, mlsName, mlsDisclaimer and idxFeedAttributionCompany on every row precisely so you know whose data you are holding). MLS data is licensed to brokerages for member use. Redistribution, resale, republication or use in a competing data product is generally restricted by those agreements and by NAR rules, even where the web page is publicly viewable. If your use case is anything other than internal analysis of markets where you operate, get written clearance first — this Actor's README is not that clearance.

Personal data. The agents dataset contains email and phone for real estate professionals, which are business contact details published for buyer enquiries. B2B outreach to agents is regulated in the US by CAN-SPAM (email) and TCPA (automated or prerecorded calls to mobile numbers). Note also that named agent consent is not implied by publication: claimedStatus is not exposed on agent rows, so verify before you treat a record as an opt-in contact. GDPR applies to agents with EU ties, and Canadian PIPEDA to agents in Canada.

Platform terms. Coldwell Banker's own Terms of Service and franchise agreements also bind, and individual franchises may set additional restrictions.

You are solely responsible for compliance with those agreements, Coldwell Banker's Terms of Service, robots.txt, and applicable law. Keep volumes modest: maxRequestsPerCrawl (default 300), maxConcurrency 3 and maxRequestRetries 3 bound a run without effort. Do not use this Actor to build or resell a competing property-listing data product without a licence.