Gumtree UK Scraper — All Categories, Any Keyword

Unofficial. This Actor is not affiliated with, authorised by, endorsed by, or in any way officially connected to Gumtree. "Gumtree" is a trademark of its respective owner and is used here only to describe what this tool reads. It reads the same public pages a browser does.

Gumtree Scraper searches any category on Gumtree UK - cars, phones, property, furniture, jobs - and returns priced listings with the structured vehicle attributes Gumtree publishes, plus battery health, storage and damage wording parsed out of the seller's own free text.

What does it do?

Gumtree server-renders its search results, so this Actor reads the tiles directly rather than driving a browser. It sends the price, category and location filters Gumtree itself supports, then applies seller-type, battery-health, storage, damage and accessory filters after fetching, which the site does not offer at all.

Why use it?

• Every category, one Actor. Vehicle ads return year, mileage, engine size, fuel type and Private/Trade; phone ads return battery health and storage. Nothing is category-locked.
• Built to stay up. Gumtree blocks an unproxied client after about five pages; we route through Apify's residential proxy and measured it completing 25 of 25 sequential result pages with no block.
• Battery health from free text. UK sellers write it nine different ways — BH85%, "battery health is 78%", "88% battery", "Battery 83 per cent", even "%90". All of them are parsed into one integer field.
• Cases and chargers filtered out. A large minority of results for any phone search are accessories whose titles contain the search term. "Case for iPhone 12" is not an iPhone.
• Paid placements are labelled. Exactly five promoted ads occupy the first five slots of every page and ignore the sort order - measured on pages 1, 2, 3 and 5. isFeatured marks them, and excludeFeatured drops them (off by default, so they are returned).
• Multi-variant shop ads are flagged, not mangled. A phone shop advertising "64gb £70 256gb £80" is one tile with two units. Reporting it as a 64 GB phone at the tile price would be wrong, so isMultiVariant is set and storageOptionsGb lists both.
• Honest sellers are not punished. "no cracks", "no scratches whatsoever" and similar negations are understood, so advertising the absence of damage does not get an ad dropped.

How to use it

1. Put one or more search terms in searchQueries. There is no default.
2. Optionally narrow with category, location, a price range, sellerType, minBatteryHealth or minStorageGb.
3. Run it. Results land in the dataset.

Input

Field	Type	Meaning
searchQueries	array	One search per term. Required, no default.
category	string	Optional Gumtree category slug. Validated before the first request, so an unknown slug fails immediately.
location	string	Location slug, e.g. uk, london. Defaults to the whole UK.
minPrice / maxPrice	integer	Optional. Omitted from the request entirely when empty — never defaulted.
sellerType	string	any | private | trade. Applied after fetching. Only motoring categories publish it, so setting it elsewhere returns zero rows.
minBatteryHealth	integer	Rejects ads that never state a figure, not just ads below the floor.
minStorageGb	integer	TB is converted, so 1TB counts as 1024. Rejects ads that never state a capacity.
excludeDamaged	boolean	Drops ads admitting damage. Default on.
excludeAccessories	boolean	Drops cases, chargers, screen protectors and spare parts. Default on.
excludeFeatured	boolean	Drops paid promoted placements. Off by default.
requireKeywordMatch	boolean	Drop items whose title is missing any word of the query. Default on.
sortBy	string	relevance | date | price_lowest_first | price_highest_first | distance.
maxItems	integer	Hard cap on rows pushed. Default 1000.
maxPages	integer	Pages per term, clamped to the site's ceiling of 50.
proxyConfiguration	object	Residential by default. Gumtree blocks an unproxied client after about five pages.

{
    "searchQueries": ["iphone 13"],
    "category": "mobile-phones",
    "minBatteryHealth": 85,
    "minStorageGb": 128,
    "maxPrice": 400,
    "maxItems": 200
}

Output

{
    "id": "1801736377",
    "title": "Apple iPhone 13 128GB Unlocked BH92%",
    "description": "Comes with box and charging cable. No scratches.",
    "url": "https://www.gumtree.com/p/apple/apple-iphone-13-128gb/1801736377",
    "imageUrl": "https://img.gumtree.com/…/86",
    "price": 300,
    "currency": "GBP",
    "location": "Enfield, London",
    "sellerType": null,
    "isFeatured": false,
    "photoCount": 4,
    "year": null,
    "mileage": null,
    "engineCc": null,
    "fuelType": null,
    "batteryHealth": 92,
    "storageGb": 128,
    "storageOptionsGb": [128],
    "isMultiVariant": false,
    "matchedKeyword": "iphone 13"
}

For motoring searches year, mileage, engineCc, fuelType and sellerType are filled from Gumtree's own structured attributes and batteryHealth / storageGb are null.

Notes and limits

• 50 pages per search, about 1,500 ads. Gumtree's own ceiling, visible in its page title (Page 5/50), regardless of how many ads it says match. Narrow with category, location or a price range to get past it.
• A proxy is not optional. Gumtree challenges an unproxied client after about five pages. Measured over 25 sequential pages: no proxy 5, datacenter 22, residential 25.
• Seller identity is never collected. Gumtree's search tiles publish only Private or Trade, never a name, and this Actor does not fetch listing pages to find one.
• Ad descriptions are reproduced exactly as the seller wrote them and are not filtered. UK sellers frequently put a phone number or a WhatsApp handle in that free text. Handle the description field accordingly.
• Battery health and storage are seller claims, parsed from prose, not verified facts. sellerType, year, mileage, engineCc and fuelType come from Gumtree's own structured fields and are more reliable.
• A run that returns zero rows fails rather than succeeding empty. If every listing is filtered out - an over-strict minBatteryHealth, or sellerType on a non-motoring category - the Actor raises instead of writing an empty dataset, so a silent misconfiguration cannot look like "no stock today".
• A keyword can end early without failing the run. If Gumtree keeps challenging after three proxy rotations, that keyword stops and the run continues with what it has. The log records how many pages were challenged; check it before treating a small result set as the whole market.
• searchOptionsExactMatch is not offered. Gumtree's own UI emits it, but it changes nothing: measured on iphone 13 mini, results were identical with and without it - same 26 tiles, same first listing, same total. No input is exposed for a control that does nothing.
• Reads the same public page a browser does. See the disclaimer at the top.