Best Web Scraping Services for Real Estate Data (2026): 15 Compared
Last updated November 2026
Most real estate web scraping does not fail loudly. One open-data project's Westchester County layer passed validation on 201,715 of 203,337 records (99.2%) while holding 203,337 of the county's 258,145 tax parcels: about four in five.
Breakage is louder. Canyon County, Idaho, moved its parcel service behind a token, and the weekly job reading it failed eight times running before it was fixed.
Listing data hides another. What looks like one national feed was 484 separate multiple listing services (MLSs) at the end of 2025, each with its own licence.
So choosing among real estate data scraping services means choosing who catches those gaps, and how you check. We judged 15, in two groups, on the sources their own pages name, MLS handling, coverage, refresh and QA, and one honest limit.
You leave with a shortlist, the model that fits your team, and the questions to ask before you sign.
- Shortlist: 15 services: 11 managed services that run collection for you, and 4 tools you run yourself.
- Channel first: MLS data is licensed per MLS, major portals' terms bar automated collection, and county offices such as Boulder's publish files.
- Coverage is unpublished: none of the 19 services here (15 plus four unnumbered), Forage AI included, publishes a per-county record count against an independent parcel count, or per-county recall or value accuracy.
- Pilot before you sign: test three counties you know and score coverage against a count you hold, not job success.
Comparing providers for property data? We build custom real estate data to your own fields and refresh cadence. Talk to us about coverage.
How we judged web scraping services for real estate
Source and channel decide more than anti-bot strength.
| Criterion | What we looked for |
|---|---|
| Real-estate sources on its own pages | Which of six types it names: listing portals, rentals, commercial, short-term rental, county records, permits |
| MLS handling | What its own pages say about MLS data |
| Coverage proof beyond volume | A count, recall or schema-alert statement |
| Refresh and change detection | Stated cadence and changes detected |
| Delivery | Formats and routes |
| Pricing model | Published floor or quote; not a rank |
Examples, not a national norm:
| Source type | Channel | Example (dated) |
|---|---|---|
| MLS listings | Licensed | Per MLS (IDX or VOW licence); 484 MLSs, end of 2025 |
| Listing portals | Web pages only, terms bar automated collection | Zillow (terms 28 Oct 2025), Realtor.com, Redfin |
| Rental sites | Web pages only, terms bar automated collection | Apartments.com (terms 6 Oct 2026) |
| Commercial listings (CRE) | Web pages only, terms bar automated collection | LoopNet (terms 7 May 2025) |
| Short-term rental (STR) | Web pages only, terms bar automated collection | Airbnb (terms 5 Feb 2026) |
| County assessor | Download / feed | Boulder CO (daily CSV files); Riverside CA (sells files) |
| County recorder | Download / feed, or Search only | Riverside CA (paid daily index); Fayette PA (free search) |
| Building permits | Download / feed | Chicago; Montgomery County MD (daily open-data feeds) |
A portal scrape is complete only for what the portal shows: delayed-marketing exempt listings and seller opt-outs never reach it. Our reasoning: ask which channel a service uses per source.
MLS data is licensed: access is granted per MLS, under that MLS's licence, and the MLS issues the credentials. There were 484 MLSs at the end of 2025.
Reading the list:
- Our service: Forage AI is our service and sits first by house convention, not by evidence; on the counted rule it would not be first.
- No paid placement: no vendor paid to be listed.
- Counted order: within each group: source types named on own pages, then MLS handling, contradictions or unsupported ratings, coverage statements, sites named.
"Not stated by vendor" means the vendor's own pages do not say it; pages read 5 November 2026.
Real estate web scraping services at a glance
The whole list on one screen, flags in the cells.
| Service | Model | Sources named (of 6) | MLS stance (own pages) | Prices published? | Best for |
|---|---|---|---|---|---|
| Forage AI | Managed | 4 (types): listings, rents, county, permits | Not stated by vendor | Quote only | County and permit programmes |
| Import.io | Hybrid | 5 (types): all but STR | Not stated by vendor | Self-serve from $9/mo | Listing history plus permits |
| DataOx | Managed | 5: all but permits | Not stated by vendor | Quote only | Most named sites |
| Actowiz Solutions | Managed | 4: portals, rentals, CRE, STR | States MLS is licensed; excluded | Pricing pages disagree | MLS out of scope in writing |
| ScrapeHero | Managed | 4: portals, rentals, county, permits | Names MLS sites as a target | Minimums published | Published per-site minimums |
| Grepsr | Managed | 3: portals, rentals, STR | Names MLS sites as a target | From $100 | Feeds priced by record |
| Xwiz Analytics | Managed | 3: portals, rentals, CRE | Names MLS sites as a target | Quote only | Named CRE sites |
| Zyte | Hybrid | 2 (types): listings, rents | Not stated by vendor | API from $0.13 per 1,000; feeds quoted | Schema checks each run |
| Outsource BigData | Managed | 1: portals | Not stated by vendor | From $50 per schedule | Low-cost scheduled pulls |
| dataforest | Managed | 1 (type): listings | Not stated by vendor | Not stated by vendor | A data-engineering partner |
| X-Byte | Managed | 1 (type): listings | Not stated by vendor | Quote; free pilot | A free pilot first |
| PromptCloud (unnumbered) | Managed | 0 named | Not stated by vendor | Quote only | Not ranked: no source named |
| Browse AI | No-code | 4: portals, rentals, CRE, STR | Not stated by vendor | From $19/mo | Monitoring pages without code |
| Zenrows | API | 3: portals, rentals, county | Not stated by vendor | From $19/mo | Engineers on assessor pages |
| Oxylabs | API | 2: portals, STR | Not stated by vendor | From $30/mo | Pages by URL or ID |
| Apify | Platform | 2 (types); community actors | Not stated by vendor | From $19/mo | Low-cost tests |
| Bright Data · Octoparse · ScraperAPI (unnumbered) | Platform / API · No-code · API | 1 · 1 · 1 | Not stated by vendor | Pay as you go · from $69/mo · from $49/mo | Portal scrapers · one-off pulls · portal guides |
Read 5 November 2026; "(types)" means types named, not sites.
The best real estate data scraping services, by service model
Forage AI sits first by house convention; the counted rule orders the rest. Pages read 5 November 2026. For other sectors, see web scraping service companies across industries.
Managed web scraping services for real estate
Managed web scraping services run collection for you, to your schema. You trade field-change speed for not running pipelines.
1. Forage AI
Managed real estate data extraction, built to your schema.
| Attribute | Detail |
|---|---|
| Best for | County, permit and listing programmes |
| Type | Managed; no self-serve tool |
| Listing portals | Type only ("listings") |
| Rental sites | Type only ("rents") |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Type only (deeds, mortgages, assessor rolls) |
| Building permits | Type only ("building permits") |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Sample can be reviewed against agreed coverage criteria |
| Refresh | Your cadence, by segment and source |
| Delivery | Delivered to your fields and systems |
| Pricing model | Quote-based, scoped to sources, fields and cadence |
| User reviews | G2 4.8 (9), read 5 November 2026 |
| Watch-out | Managed means setup time and a request per field change; does not license MLS data |
Forage AI builds and runs managed pipelines for real estate teams: it finds the sources, extracts and structures the records, checks quality and delivers into your own fields and systems. Its page names listings and rents, county deed and mortgage records, assessor rolls, building permits and parcel attributes, normalised into one schema and refreshed on your cadence.
It fits a county-heavy recurring programme without a data-engineering bench. A sample can be reviewed against agreed criteria such as coverage, field completion and duplicate rate; see our managed real estate data extraction. The limit: setup time, a request per field change, and Forage AI does not license MLS data.
2. Import.io
| Attribute | Detail |
|---|---|
| Best for | Listing history plus permits |
| Type | Hybrid: self-serve plus managed feed |
| Listing portals | Type only ("property portals") |
| Rental sites | Type only (rental listing sites) |
| Commercial | Type only (commercial listings) |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Type only (assessor and recorder sites) |
| Building permits | Type only (permit portals) |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Scoped per jurisdiction; changes versioned |
| Refresh | Listings and status daily; permits weekly |
| Delivery | API or managed feed |
| Pricing model | Self-serve $9 to $699 a month by queries; managed quoted |
| User reviews | G2 3.5 (4), read 5 November 2026 |
| Watch-out | History starts at its own first capture |
Import.io names the widest spread of source types here, county and permit portals included, and is the only page we read that scopes coverage by jurisdiction. That is a statement, not a count: ask for the county list.
Each price and status change is versioned from first capture, so days on market are observed, not inferred.
3. DataOx
| Attribute | Detail |
|---|---|
| Best for | Named portals, rentals, CRE, STR |
| Type | Managed |
| Listing portals | Named: Zillow, Realtor.com, Redfin, Trulia |
| Rental sites | Named: Apartments.com, Rent.com, Craigslist |
| Commercial | Named: LoopNet |
| Short-term rental | Named: Airbnb, Vrbo |
| County assessor / recorder | Type only ("property tax records") |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Automated plus manual checks; no coverage measure |
| Refresh | Hourly or daily by plan |
| Delivery | CSV or JSON, to API or warehouse |
| Pricing model | Custom estimate by platform count; no figure |
| User reviews | G2 5.0 (1), read 5 November 2026 |
| Watch-out | Most named sites, but no coverage or QA measure |
DataOx names more specific sites than any other managed page here, from four listing portals to LoopNet and Airbnb. Its data-sources list adds property tax records, making five types. The page was modified on 15 October 2026.
It publishes no measure of what it delivers beyond automated checks with manual review, and one G2 review says nothing either way.
4. Actowiz Solutions
| Attribute | Detail |
|---|---|
| Best for | Feeds with MLS out of scope |
| Type | Managed, plus API plans |
| Listing portals | Named: Zillow, Redfin, Realtor.com, Trulia |
| Rental sites | Named: Apartments.com |
| Commercial | Type only (where portals publish it) |
| Short-term rental | Type only (where portals publish it) |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | States MLS is licensed and excluded |
| Coverage and QA beyond volume | Schema checks, sampled review, completeness scoring |
| Refresh | Daily; hourly for hot markets |
| Delivery | JSON, CSV, Parquet; webhooks, S3 |
| Pricing model | Both appear on its pages: crawler plans from $99/month and plans from $500/mo; managed quoted |
| User reviews | G2 profile, no rating shown, read 5 November 2026 |
| Watch-out | Two different starting prices; get the plan in writing |
Actowiz Solutions is the only managed page we read that describes MLS systems as licensed, credentialed environments and lists MLS content among what it will not collect. That is the written scope boundary question 6 below asks for.
It is also candid that its price history begins when its own runs first saw a listing. Ask every vendor the same question.
5. ScrapeHero
| Attribute | Detail |
|---|---|
| Best for | Published per-site minimums |
| Type | Managed, plus self-serve scrapers |
| Listing portals | Named: Zillow, Realtor.com, Redfin, Trulia, Homes.com |
| Rental sites | Type only (rental listings) |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Type only (county valuation sites) |
| Building permits | Type only ("building permits") |
| MLS (as stated) | Names MLS sites as a target |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | One-time, weekly, monthly or any, by plan |
| Delivery | JSON, CSV, Excel, XML |
| Pricing model | "$550 minimum per site per refresh"; Business "Starts at $199" per site per month; Enterprise minimums $1,500 and $8,000 per month |
| User reviews | G2 4.7 (63), read 5 November 2026 |
| Watch-out | 2024 listicle names MLS sites as targets (see below) |
ScrapeHero publishes the clearest minimums of any managed service here. Its real-estate page, modified July 2026, runs from listings to building permits and names county valuation websites by type only; test that in the pilot.
Its 2024 listicle lists MLS sites among scraping targets. MLS listing data is available only under a licence from each MLS, and one MLS's 2026 rules require licensed IDX and VOW displays to impede scraping.
6. Grepsr
| Attribute | Detail |
|---|---|
| Best for | Scheduled feeds priced by record |
| Type | Managed, plus self-serve platform |
| Listing portals | Named: Zillow, Redfin, Realtor.com, Trulia |
| Rental sites | Named: Apartments.com |
| Commercial | Not stated by vendor |
| Short-term rental | Named: Airbnb, Vrbo |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Names MLS sites as a target |
| Coverage and QA beyond volume | Automated plus human checks; no coverage measure |
| Refresh | Daily (Zillow page); your schedule |
| Delivery | CSV, JSON, Parquet; API, S3, SFTP |
| Pricing model | Starter from $100, by data records; higher tiers quoted |
| User reviews | G2 4.5 (32), read 5 November 2026 |
| Watch-out | December 2025 guide names MLS platforms as targets (see below) |
Grepsr's real estate page, modified September 2026, names portal, rental and short-term rental sites; its Zillow page states a daily refresh. Starter bills by data records, one of the clearer units here. Permit pages are use cases, not counted.
Its December 2025 guide lists local MLS platforms among scraping targets. MLS listing data is available only under a licence from each MLS, and one MLS's 2026 rules require licensed IDX and VOW displays to impede scraping.
7. Xwiz Analytics
| Attribute | Detail |
|---|---|
| Best for | Named portal, rental and CRE feeds |
| Type | Managed |
| Listing portals | Named: Zillow, Realtor.com, Redfin, Trulia |
| Rental sites | Named: Apartments.com, Zumper |
| Commercial | Named: LoopNet, Crexi |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Names MLS sites as a target |
| Coverage and QA beyond volume | Validation only; no coverage measure |
| Refresh | Hourly, daily or weekly by plan |
| Delivery | CSV, Excel, JSON, API |
| Pricing model | Quoted by portals, markets, frequency, fields; free sample |
| User reviews | No third-party profile read |
| Watch-out | Real-estate page names MLS sites as targets (see below) |
Xwiz Analytics names sites in three source types, including LoopNet and Crexi for commercial listings, and offers a free sample dataset before a quote. It states validation but no coverage measure.
Its real-estate page (read 5 November 2026) lists MLS sites among scraping targets. MLS listing data is available only under a licence from each MLS, and one MLS's 2026 rules require licensed IDX and VOW displays to impede scraping.
8. Zyte
| Attribute | Detail |
|---|---|
| Best for | Scoped listing feeds, schema-checked |
| Type | Hybrid: managed feeds plus API |
| Listing portals | Type only (listing sources) |
| Rental sites | Type only (rent comps) |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Schema-validated runs; layout changes raise alerts |
| Refresh | Price and status daily; attributes usually weekly |
| Delivery | JSON Lines or CSV; S3, GCS, SFTP, API |
| Pricing model | Feeds quoted; API from $0.13 per 1,000 pay as you go, $0.06 on a $500-a-month commitment |
| User reviews | G2 4.4 (122), read 5 November 2026 |
| Watch-out | Names no specific portal; ask which sites are in scope |
Zyte sits in the managed group because its real-estate page leads with scoped feeds. Each run is validated against your agreed schema, so a layout change raises a coverage alert, not a silent gap.
It sets refresh per portal to how often each site changes, the nearest any page gets to a source's own cadence.
9. Outsource BigData
| Attribute | Detail |
|---|---|
| Best for | Low-cost scheduled portal pulls |
| Type | Managed |
| Listing portals | Named: Zillow, Realtor.com, Trulia |
| Rental sites | Not stated by vendor |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | Daily, weekly or monthly |
| Delivery | Excel, XML, JSON |
| Pricing model | From $100; bulk from $250; $50 per schedule |
| User reviews | G2 5.0 (28), AIMLEAP profile, read 5 November 2026 |
| Watch-out | One source type; no QA statement |
Outsource BigData publishes the lowest managed entry prices we found. It also lists brokerage websites such as Keller Williams and RE/MAX, which we note but do not count.
With one source type and no QA statement, it suits a defined portal pull more than a gap-sensitive programme.
10. dataforest
| Attribute | Detail |
|---|---|
| Best for | A data-engineering partner |
| Type | Managed (data engineering) |
| Listing portals | Type only (real estate platforms) |
| Rental sites | Not stated by vendor |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | Hourly to daily |
| Delivery | Not stated by vendor |
| Pricing model | Not stated by vendor |
| User reviews | G2 profile, no rating shown, read 5 November 2026 |
| Watch-out | Names no site; ask for the list in writing |
dataforest positions itself as a data-engineering partner, updating real estate platform data hourly to daily. Its page names no site, delivery format or price.
So the first call is a scoping call: ask for the site list, format and a sample in writing.
11. X-Byte
| Attribute | Detail |
|---|---|
| Best for | A free pilot first |
| Type | Managed |
| Listing portals | Type only ("multiple property listing websites") |
| Rental sites | Not stated by vendor |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | Daily, weekly, monthly or custom |
| Delivery | Not stated by vendor |
| Pricing model | Free pilot run, then quote |
| User reviews | G2 profile, no rating shown, read 5 November 2026 |
| Watch-out | Names no site; use the free pilot to see coverage |
X-Byte collects from what its page calls multiple property listing websites. A vacation-rental heading is not a stated source, so it counts one type.
Use its free pilot run to see which websites it covers and what format arrives.
PromptCloud is a managed data service; its real-estate page names no specific source, so it is not numbered here.
Scraping platforms and APIs you run yourself
If you have engineers and want to run collection yourself: web scraping as a service, self-serve. You own parsing, change detection and QA.
12. Browse AI
| Attribute | Detail |
|---|---|
| Best for | Monitoring listing pages without code |
| Type | No-code platform |
| Listing portals | Named: Redfin, Rightmove (brokerage Compass noted) |
| Rental sites | Named: Apartments.com |
| Commercial | Named: LoopNet |
| Short-term rental | Named: Airbnb |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Field-level change alerts; no coverage measure |
| Refresh | You set it, every minute to monthly |
| Delivery | Sheets, CSV, JSON, API, webhooks |
| Pricing model | Free; Personal $19/mo annual ($48 monthly); Professional $69; Premium from $500 |
| User reviews | G2 4.8 (61), read 5 November 2026 |
| Watch-out | You train each robot; checks and terms stay yours |
Browse AI leads this group on counted sources, the only self-serve page here naming both a commercial and a short-term rental site. Change monitoring is its stated strength; coverage is not.
13. Zenrows
| Attribute | Detail |
|---|---|
| Best for | Engineers on portal and assessor pages |
| Type | Scraping API |
| Listing portals | Named: Zillow, Redfin, Realtor.com |
| Rental sites | Named: Apartments.com, Zumper |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Type only (county assessor sites) |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Suggests count versus portal, empty fields per run |
| Refresh | You schedule it; daily diffs suggested |
| Delivery | HTML, Markdown or JSON; webhook |
| Pricing model | 5,000 free credits; $19 to $2,999 a month; Enterprise quoted |
| User reviews | G2 5.0 (18), read 5 November 2026 |
| Watch-out | Its count check is against the portal, not an independent count |
Zenrows is the only self-serve page here naming county assessor sites. Its suggested per-run count check is a sound habit, though the portal is not an independent count.
14. Oxylabs
| Attribute | Detail |
|---|---|
| Best for | Pages by URL or listing ID |
| Type | Scraping API |
| Listing portals | Named: Zillow, Redfin |
| Rental sites | Not stated by vendor |
| Commercial | Not stated by vendor |
| Short-term rental | Named: Airbnb |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | You schedule it |
| Delivery | HTML or parsed results; API or bucket |
| Pricing model | $30, $100 or $500 a month; Enterprise from $2,500 |
| User reviews | G2 4.5 (421), read 5 November 2026 |
| Watch-out | Parsing, change detection and QA are yours |
Oxylabs fits engineers who know which pages they need: Zillow and Redfin by URL, Airbnb by listing ID. Checking what comes back is your work.
15. Apify
| Attribute | Detail |
|---|---|
| Best for | Low-cost tests with community actors |
| Type | Platform with community marketplace |
| Listing portals | Type only (properties, housing market) |
| Rental sites | Type only (apartments, rentals) |
| Commercial | Not stated by vendor |
| Short-term rental | Not stated by vendor |
| County assessor / recorder | Not stated by vendor |
| Building permits | Not stated by vendor |
| MLS (as stated) | Not stated by vendor |
| Coverage and QA beyond volume | Not stated by vendor |
| Refresh | Platform scheduling |
| Delivery | Community actors; quality varies by publisher |
| Pricing model | Free; $19, $199 or $999 a month, plus usage |
| User reviews | G2 4.7 (699), for the platform, read 5 November 2026 |
| Watch-out | Actors are third-party builds; check publisher and last update |
Apify's real-estate category lists community-built actors for listings and rentals. Their claims are their publishers', not Apify's: check who built each and when.
Also: Bright Data, a scraping platform with ready-made scrapers for Zillow and Realtor.com; Octoparse for a one-off no-code pull with portal templates; ScraperAPI, a scraping API with portal guides.
Not on this list, and why
- Property data providers and datasets: they sell data they already hold, a different category from services that collect the sources you name.
- Services offering credentialed or MLS-member access: not listed; shortlisting them would imply MLS data can be collected that way.
- Single-portal tools and browser extensions: built for one site, not a programme.
Real estate web scraping services compared
All 15 side by side: read across from your source list; cells come from each service's own pages.
| Service | Portals | Rental | Commercial | STR | County | Permits |
|---|---|---|---|---|---|---|
| Managed services | ||||||
| Forage AI | Type only | Type only | Not stated | Not stated | Type only | Type only |
| Import.io | Type only | Type only | Type only | Not stated | Type only | Type only |
| DataOx | Named | Named | Named | Named | Type only | Not stated |
| Actowiz Solutions | Named | Named | Type only | Type only | Not stated | Not stated |
| ScrapeHero | Named | Type only | Not stated | Not stated | Type only | Type only |
| Grepsr | Named | Named | Not stated | Named | Not stated | Not stated |
| Xwiz Analytics | Named | Named | Named | Not stated | Not stated | Not stated |
| Zyte | Type only | Type only | Not stated | Not stated | Not stated | Not stated |
| Outsource BigData | Named | Not stated | Not stated | Not stated | Not stated | Not stated |
| dataforest | Type only | Not stated | Not stated | Not stated | Not stated | Not stated |
| X-Byte | Type only | Not stated | Not stated | Not stated | Not stated | Not stated |
| Platforms and APIs you run yourself | ||||||
| Browse AI | Named | Named | Named | Named | Not stated | Not stated |
| Zenrows | Named | Named | Not stated | Not stated | Type only | Not stated |
| Oxylabs | Named | Not stated | Not stated | Named | Not stated | Not stated |
| Apify | Type only | Type only | Not stated | Not stated | Not stated | Not stated |
"Not stated" means Not stated by vendor; brokerage and foreclosure sources are noted in entries, not counted.
Our read, November 2026: county records are named on the pages of 5 of the 19 services in this article, and building permits on 3. Push hardest there, because "named" on a page is not "delivered in production".
How should you choose a real estate web scraping service?
Pick by the source types you need and the team you have, then test before you sign.
- County and permit sources at scale, no data-engineering bench: a managed service; ask for per-county counts. Forage AI is one managed option for county, permit and listing pipelines built to your schema.
- Portal and rental pages, with an engineering team: a platform or API; you own parsing, change detection and QA.
- A one-off pull or page monitoring: a no-code tool.
- Listing data at MLS fidelity: MLS data is licensed: access is granted per MLS, under that MLS's licence, and the MLS issues the credentials.
A managed service owns the breakage and builds to your schema, but a field change becomes a request and a wait. With a platform or API, your engineers own the breakage, the schema and each source's terms; see managed versus automated web scraping services.
Success rates are a property of the target, not the vendor: in Proxyway's 2026 market research the same proxy networks succeeded on a median of 99.3% of requests to a small test page and 74.4% on Amazon, Google and Instagram. In practice, a job can succeed on every request and still miss a fifth of a county.
Questions to ask before you sign
Run a small pilot first:
- Pick three counties you know: one easy, one hard, one whose office recently changed systems.
- Add your one or two key portals and one permit source.
- Ask for a paid sample to your schema, on the cadence you would buy.
- Score it on the questions below, not on row volume.
Then ask each vendor, in writing:
- How many records per county do you deliver against an independent parcel count (the county roll or a statewide parcel file), and per recording office for recorder data? The Westchester layer passed validation while missing about a fifth of its parcels.
In our read of the 19 services here (November 2026), Forage AI included, none publishes a per-county record count against an independent parcel count, or per-county recall or value accuracy, for a real-estate source.
- On a hand-checked sample, how many source records are missing, and how many delivered fields are wrong?
- Does refresh follow each source's cadence (some assessors publish daily, others weekly), with price, status and delisting diffs? Our reasoning, from one county's notice: an assessor's owner name can lag the recorder by weeks.
- Where does price and status history start, and is it observed or inferred?
- What alerts you when a source returns partial or empty results, how fast, and with what fix commitment?
- Which channel do you use per source (scrape, download or API), and what will you not collect, in writing? Expect an MLS answer: in our read, 3 of the 12 managed services here, PromptCloud included, name MLS sites among scraping targets, and 1 states MLS is licensed and excluded.
- Which of my sources' terms bar automated collection, and how does the contract allocate that? Your counsel decides.
- What does a field or county change cost in time and money?
- What delivery format, schema versioning and per-run manifest of row counts per source do you provide?
- What is the pricing unit (site, record, query, refresh), and what triggers a re-quote?
Is real estate web scraping legal?
This article is for informational purposes only and does not constitute legal advice. Consult a qualified attorney for legal guidance specific to your situation.
No court has settled whether collecting public data is lawful, so question 7 starts with what the portals' contracts say.
| Portal | Terms version | What the terms say | AI / ML clause |
|---|---|---|---|
| Zillow | Updated 28 Oct 2025 | users agree not to "conduct automated queries (including screen and database scraping" | No |
| Realtor.com | Undated; read 5 Nov 2026 | no one may "collect, scrape, text mine or data mine" content "by any automated means" | Yes |
| Redfin | Updated 29 Sep 2025 | may not "automatedly crawl or query the Services" without written permission | No |
| Apartments.com | Updated 6 Oct 2026 | no "robot, spider, scraper" to "access, monitor, copy, extract, collect" | Yes |
| LoopNet | Revised 7 May 2025 | no "data mining, gathering or extraction tool, or any robot" | No |
| Airbnb | Updated 5 Feb 2026 | "Do not use bots, crawlers, scrapers, or other automated means" | No |
Robots.txt does not settle it either. Realtor.com's terms of service say a user's rights are not expanded "in any way by our use or configuration of exclusionary protocols", robots.txt included.
The AI clauses matter twice if the data will train or ground a model: Realtor.com bars use "for any machine learning or artificial intelligence purposes", and Apartments.com bars collection to train AI systems.
One MLS's 2026 rules also oblige IDX and VOW display operators to take commercially reasonable steps to impede scraping.
The most cited case settles less than people think: the Ninth Circuit's 2022 hiQ v. LinkedIn rulings came on a preliminary-injunction record, and the case ended in December 2022 with a $500,000 consent judgment and an injunction against hiQ.
Vendor pages often call public listing data generally legal to collect, while these portals' terms bar automated collection. What you collect is yours to settle with counsel; start with our overview of US web scraping laws.
Which real estate scraping service should you shortlist?
With the terms read, prioritise by programme:
- County-heavy recurring programme: a managed service whose pages name county and permit sources. Forage AI and Import.io name both; ScrapeHero names county websites and permits; DataOx names county tax records.
- Engineering team on portal and rental pages: Zenrows, the only self-serve page here naming county assessor sites, or Oxylabs for portal pages by URL.
- One-off pull or page monitoring without code: Browse AI, or Octoparse.
A November 2026 shortlist of real estate data scraping services has a shelf life. Re-read vendor pages, prices and G2 each quarter, and portal terms by version date. Then run the pilot. Hold every shortlisted service to a per-county count, and compare what comes back.
Why Forage AI for real estate web scraping
Forage AI is a fully managed service, never self-serve. It builds and runs pipelines for listings and rents, county deed and mortgage records, assessor rolls, building permits and parcel attributes, normalised to your schema, refreshed on your cadence and delivered to your fields and systems, and you own the data. Forage AI prices each programme from a quote based on the sources, fields and refresh cadence you specify.
Choose Forage AI if you run a recurring programme across county records, permits and listings and want the pipeline, QA and delivery run for you.
Choose a self-serve scraping API if you have engineers who will own parsing, change detection and QA, or you need a one-off pull.
FAQs about real estate web scraping services
Does Zillow allow data scraping?
Zillow's terms, updated 28 October 2025, ask users not to conduct automated queries, including screen and database scraping. Whether they bind a given collection is unresolved: your call, with counsel. To build a scraper yourself, see our guide to Zillow data scraping.
Can a scraping service collect MLS data?
MLS data is licensed: access is granted per MLS, under that MLS's licence, and the MLS issues the credentials. The RESO Web API page says access is gained through local MLSs, under their licensing policies (read 5 November 2026). There were 484 MLSs at the end of 2025.
How much do real estate scraping services cost?
As of November 2026, self-serve plans start at $9 to $69 a month. Published managed entry points run from $50 per schedule (Outsource BigData) to $550 per site per refresh (ScrapeHero), and most managed services quote. Units differ; see our real estate API pricing breakdown.
What is the difference between a real estate data provider and a scraping service?
A provider sells data it already holds, in its own schema and on its own cycle. A scraping service collects the sources you name, to your schema, on your cadence. For packaged datasets, see our list of real estate data providers.
Can a scraping service collect county assessor or recorder records?
Five of the 19 services here name them. Offices also publish files: Boulder County, Colorado, refreshes its assessor CSV files daily, and Riverside County, California, sells files. For property data scraping at scale, ask which channel the service uses per county. Recorder data is organised by recording office.
How often should listing data be refreshed?
Match each source's own change rate, not a vendor's default schedule: "real-time" is a vendor's word, not a measured freshness. See our comparison of daily versus change-triggered refresh.
Related articles
- PropStream alternatives: routes to property and owner data
- Data extraction enterprise evaluation checklist: vetting any extraction vendor
- How real estate investors use web data: use cases for listing and county data
- What is managed web data extraction?: how managed extraction works
Sai is a data infrastructure enthusiast who has spent the past two to three years following the AI space closely, from the infrastructure layer to the fast-growing world of data for AI. He is genuinely curious about how modern data pipelines get built and where the data industry is heading, and he writes insightful pieces on the core topics that shape this niche.