Redfin Scraper

Scrape Redfin home search results for properties for sale, recently sold homes, and rentals from a search URL. Export Redfin property data to CSV, JSON, or Excel with prices, addresses, bedrooms, bathrooms, floor area, location, status, and source links. The actor provides a structured alternative to manually copying records from a map search.

Start with the exact search page you would use in a browser. The actor reads the filters already encoded in that URL, collects matching listing records, follows additional result pages, and returns one flat record per property. It can divide a crowded map into smaller areas when the search reaches its nine-page result limit.

What data can you extract from Redfin?

The output schema is stable across sale, sold, and rental searches. Fields that a particular listing category does not provide are null. Monetary fields are the numeric amounts shown in the search result; this actor does not convert currencies, calculate taxes, or infer missing prices. Areas use square feet where the source supplies an area. A rental range uses the minimum in the main field and the maximum in the matching Max field.

Field	Description	Example
propertyId	Site property identifier	121243941
listingId	Sale listing identifier	223846331
rentalId	Rental record identifier	null for a sale
mlsId	Public listing reference	1088464
url	Direct property page link	https://www.redfin.com/FL/Mims/4109-Sweet-Bay-Dr-32754/home/121243941
offerType	Search result category	sale
mlsStatus	Listing status text	Active
price	Asking price, sold price, or minimum rent	340000
priceMax	Upper rent amount if a range is shown	null
beds	Bedrooms, or lower bedroom count for a rental range	3
bedsMax	Upper bedroom count for rentals	null
baths	Bathrooms, or lower bathroom count for rentals	2
bathsMax	Upper bathroom count for rentals	null
sqFt	Interior square footage or lower rental area	1726
sqFtMax	Upper rental area in square feet	null
pricePerSqFt	Search result price per square foot	197
lotSize	Lot size in square feet	36155
hoa	Association amount if shown	null
streetLine	Street address	4109 Sweet Bay Dr
unitNumber	Apartment or unit label	Unit C206
city	City name	Mims
state	State abbreviation	FL
zip	Postal code	32754
countryCode	Country code	US
location	Neighborhood or location label	Spruce Hills 1st Addn To
latitude	Property latitude	28.668593
longitude	Property longitude	-80.8638018
propertyType	Site property type code	6
listingType	Site listing type code	1
searchStatus	Site search status code	1
yearBuilt	Construction year	1957
soldDate	Date of a recorded sale, when present	1989-02-01
daysOnMarket	Search result days on market	2
timeOnRedfin	Listing time value supplied by the source	141341844
photoCount	Number of listing photos	32
listingAgentName	Public agent name if shown in the result	null
isHot	Hot listing flag	false
isNewConstruction	New construction flag	false
hasVirtualTour	Virtual tour flag	true
hasVideoTour	Video tour flag	false
has3DTour	3D tour flag	false
propertyName	Rental building or community name	null for a sale
numAvailableUnits	Available rental unit count	null for a sale
listingRemarks	Public description displayed with the result	NO HOA + . 83 FENCED ACRE! ...
lastUpdated	Rental update timestamp	null for a sale
resultIndex	Output order within the supplied search URL	1
foundFromSearchUrl	Input URL that produced the property	https://www.redfin.com/zipcode/32754
zoomQuadrantSequence	Quadrants visited after a map was split	[]

A recorded soldDate can be a prior transaction on an active listing. Use offerType and mlsStatus to identify the result's current category. The numeric type and status codes are preserved from the search data so you can compare records consistently; the actor does not assign labels to undocumented numeric values.

How to scrape Redfin

1. Open a Redfin search page that displays homes on a map. Select the area, sale or rental category, and any filters you need.
2. Copy the full search URL into searchUrls. Keep its filter and viewport portion; that part defines the search you intend to export.
3. Set maxResults to your intended output size. Keep zoomIn enabled when the area may contain thousands of listings.
4. Run the actor. When it finishes, open the default dataset and download the records in JSON, CSV, or Excel format.
5. For recurring work, reuse the same input in a schedule and compare stable propertyId values between runs.

Each search URL is processed in the order supplied. Listings that appear in more than one input search are deduplicated by property identifier within the run. The first search to return a property determines its foundFromSearchUrl. A result limit applies to the combined dataset, so place the most important search first when the limit is small.

Input

Name	Type	Default	Description
searchUrls	Array of URL objects	Required	One or more map search URLs, including sale, sold, or rental search pages.
zoomIn	Boolean	true	Split dense map searches into smaller areas after the page cap is reached.
maxResults	Integer	100	Stop after this many unique property records across all URLs. Use 0 for no result cap.
proxyConfiguration	Object	Enabled	Connection setting for the search requests. A US residential connection is used.

A small ZIP search is a useful first run because it gives you a manageable dataset and lets you inspect field coverage before expanding to a wider map. This example requests 50 homes from a current public ZIP search:

{
  "searchUrls": [
    { "url": "https://www.redfin.com/zipcode/32754" }
  ],
  "maxResults": 50,
  "zoomIn": true
}

A map URL with an explicit viewport is also supported. The viewport format is part of the site URL, so paste it as copied instead of entering coordinates separately. A recently sold search uses a URL with a sold filter, such as https://www.redfin.com/zipcode/32754/filter/include=sold-3mo. A rental search uses a rental URL, such as https://www.redfin.com/zipcode/32754/rentals. Search pages must be public and reachable without an account.

Output example

The following values come from the first item of a successful 50-result ZIP search. This is a shortened selection of fields from the actual output record; the dataset also includes the other fields in the table above.

{
  "propertyId": 121243941,
  "listingId": 223846331,
  "mlsId": "1088464",
  "url": "https://www.redfin.com/FL/Mims/4109-Sweet-Bay-Dr-32754/home/121243941",
  "offerType": "sale",
  "mlsStatus": "Active",
  "price": 340000,
  "beds": 3,
  "baths": 2,
  "sqFt": 1726,
  "pricePerSqFt": 197,
  "lotSize": 36155,
  "streetLine": "4109 Sweet Bay Dr",
  "city": "Mims",
  "state": "FL",
  "zip": "32754",
  "latitude": 28.668593,
  "longitude": -80.8638018,
  "yearBuilt": 1957,
  "daysOnMarket": 2,
  "photoCount": 32,
  "resultIndex": 1,
  "foundFromSearchUrl": "https://www.redfin.com/zipcode/32754",
  "zoomQuadrantSequence": []
}

The same 50-result run returned a propertyId, property URL, offer type, price, street, city, state, ZIP, bedroom count, and bathroom count on every row. Floor area appeared on 96% of rows. That run is a sample from one area at one point in time; other areas and property categories can have different coverage.

Use cases

• Market inventory tracking: Save a daily set of active properties in a defined ZIP code or viewport. Match records by propertyId to identify additions and removals.
• Price monitoring: Compare price for the same property across runs to find changes. Retain the original dataset for each date because the search result reflects what was visible at collection time.
• Neighborhood research: Group listings by city, zip, or location, then compare bedroom counts, floor areas, and asking prices. The coordinates allow a map view without extracting addresses again.
• Rental supply checks: Use rental search URLs to collect available buildings and units. The minimum and maximum fields keep rent and size ranges explicit.
• Recently sold research: Use a sold filter URL to capture public transaction search results, then distinguish current sold listings with offerType and mlsStatus.
• Listing quality checks: Monitor the presence of photos, tour flags, descriptions, and key property dimensions across a set of listings.

How much does it cost to scrape Redfin?

The configured result price is $1.00 per 1,000 records, or $0.001 for each pushed result. At that rate, 50 results account for $0.05 in result charges, 1,000 results for $1.00, and 5,000 results for $5.00. Your run dashboard also shows resource and connection usage. Check the price displayed when starting a run because account charges and Store prices can change.

Only emitted dataset records count toward the result limit. A URL that yields no listings does not produce result rows. A large map can require more source requests than a small ZIP search, particularly when it must be divided into quadrants. The maxResults setting gives a straightforward cap on the number of paid result events. For an initial evaluation, start with 50 or 100 results and expand after you check the output fields.

Tips for best results

Use a search URL copied after applying the filters you want. The actor follows the endpoint parameters embedded in the page, including the category and map viewport. A short URL can be useful for a whole ZIP or city; a viewport URL is more precise for a custom geographic box. If you need only one type of property, select that type before copying the URL.

Leave zoomIn enabled for broad, dense maps. A search response can contain up to 350 homes per page, and the site exposes at most nine pages for a map. When the ninth page is still full, the actor can split the geographic bounds into four quadrants, search each one, and deduplicate properties that fall on an edge. Splitting continues only where the cap is still reached, with a safety depth to prevent unbounded requests. zoomQuadrantSequence records the path used for a listing.

For a quick sample, set maxResults below 350. The actor can then use the listing data already embedded in the first page and stop without fetching additional result pages. For a complete export of a large area, set the limit to 0 and allow more time. A narrow area or detailed filters will usually need fewer requests than a major metropolitan map. Combine several focused URLs when you need a known set of neighborhoods.

Use propertyId as the main property key. listingId can change when a property is relisted, and rental records can have a separate rentalId. Keep foundFromSearchUrl with the row to preserve the search context. If the same property appears in overlapping searches during one run, it is pushed once. Across separate runs, deduplicate or compare using the stable key in your own data store.

Limits

The actor reads public search results, not full property detail pages. The output includes fields available in the search response; it does not promise every attribute visible after opening an individual home page. Some records omit bedroom and bathroom counts, floor area, association amount, agent name, photo count, or remarks. Land and other nonstandard property records can have especially sparse interior details. Rental records have a different structure and often provide ranges rather than a single bed count, bath count, area, or rent. Null indicates that the search response did not supply a value.

A result page can be capped at 350 homes, and a map search can stop after nine pages. The quadrant option addresses this for searches with usable geographic bounds, but a highly dense area can still reach the maximum split depth. In that case the run logs a cap warning and the dataset may be incomplete. Divide the area manually or apply tighter filters if complete coverage is essential. Map edges can cause the same property to appear in adjacent quadrants; the run removes duplicates by identifier.

Search pages and their endpoint formats can change. The actor requires the page's embedded response cache; it reports an error when the expected listing endpoint is missing rather than returning an empty success. Blocks and rate limits can also occur. The connection uses US residential access because ordinary data center requests were rejected during verification. Running a smaller search first is the best way to confirm that a new filter URL is supported.

A sold search can include surrounding properties in the site's own response. The actor excludes records without a listing status from that response. Historic soldDate values may also be present on active listings, so do not treat the presence of a date alone as proof that the current listing is sold. The source can revise listing text, prices, availability, and status between runs; each dataset is a snapshot of the search at collection time.

Automated workflows

A workflow can pass one or more saved map URLs as input, wait for the run to finish, and read the default dataset. For example: “Collect up to 100 active homes from this ZIP search and return the price, beds, baths, floor area, and URL for each property.” Another example: “Run this recently sold search every week, retain the dataset for each week, and compare prices and property IDs.”

For repeatable comparisons, keep the search URL and the run date with the downloaded dataset. A new run using the same URL can reflect changes in inventory and filters maintained by the site. If an area is unusually broad, make the workflow provide several focused URLs and set a practical maxResults cap. The actor's flat fields are suitable for tabular storage and simple filtering without unpacking nested source objects.

FAQ

Is it legal to collect these listings?

The actor handles publicly visible search data. Check the site's terms and the rules that apply to your use, including privacy and data protection rules. Use the results for legitimate research and avoid personal data misuse. This actor does not access account-only pages or collect private contact details.

Do I need to set up a proxy?

The default connection is configured for a US residential route because the site rejected ordinary data center requests in verification. Most users can keep the default setting. If your account has a custom connection arrangement, the input exposes proxyConfiguration for that setting. Connection availability and charges are shown by the platform.

How fast is a run?

A verified 50-result ZIP search completed in about 12 seconds after its build was ready. Time varies with area size, page count, rate limits, and whether the map must split. A large unlimited search can take substantially longer than a 50-result sample. The run log reports progress and any failed search URLs.

Can I schedule and monitor it?

Yes. Save the input and set a recurring schedule in the platform. Check each run's status, item count, and logs. For market history, retain each dated dataset rather than replacing old data with the newest values.

Can I export to a spreadsheet?

Yes. The default dataset can be downloaded as CSV or Excel, as well as JSON. A spreadsheet works well for sorting by price, filtering by ZIP or status, and making simple comparisons. Keep identifier columns as text if your spreadsheet software changes their formatting.

Why is a field empty?

The source search response may not include it for that listing or property category. Rental listings, for example, can have propertyName and numAvailableUnits while sale listings have listingId and yearBuilt. Some fields also depend on the data supplied by a particular listing. Null is preferable to an invented value.

Will this return every home in a very large area?

It attempts to page through results and split a saturated map, but no finite map strategy can guarantee completeness after the split depth is reached or if the site changes access rules. Check the run log for a cap warning. Use smaller map boxes or more specific filters when you need a complete inventory.

Integrations

The dataset can be read through the platform API, downloaded on completion, or passed to another step through a webhook. Common patterns include scheduled exports, spreadsheet updates, database imports, and alerts when the inventory or price of a tracked property changes. The output is flat and uses stable property identifiers, so downstream steps can compare runs directly.

Support

Open an issue on the Issues tab; we reply within 24h and add fields on request. Include the search URL, run link, and a short description of the missing or unexpected data so the behavior can be reproduced.