Gumtree Scraper - Extract UK Classifieds Listings

What We Extract from Gumtree

Each Gumtree section fills its own attribute block, which is why a Gumtree scraper is built per category, not per site. Every advert shares a core:

• Title, description, price in GBP and posted date - the date is relative, as Posted 5 hours ago, and is resolved against collection time.
• Location - town and county as displayed, such as Fraserburgh, Aberdeenshire, plus the internal location id behind the area tree.
• Advert id - a ten-digit number ending the URL and printed as Ad ID. The only stable key: the title slug moves when the advert is edited.
• Seller type - Private, Trade or Agency. A real field and a search filter, not a guess from the wording.
• Photo count, image URLs, video flag, and whether the seller ships.
• Promotion state - Featured, Urgent, Standout, Bump Up and Premium flags, separating paid placement from organic ranking.

Cars carry the deepest block: make, model, registration year, mileage in miles, transmission, fuel type, body type, doors, seats, colour, engine size, CO2, Euro emissions band, insurance group, mpg and Previous Keepers, beside a history check flagging stolen, scrapped, written-off and road-worthy status. Trade adverts add a VAT marker and finance figure. MOT is not a structured field - where it shows, the seller typed it in.

Property splits by intent: To Rent, To Share, Commercial, Parking and Garage, For Sale, To Swap, Property Wanted and Holiday Rentals. Its attributes are property type, bedrooms, bathrooms, date available and a price frequency flag of pm or pw - the same rent figure means a different amount under each. Deposit, furnishing and bills sit in the description, not in attributes, and are parsed from the text. Jobs carry salary text, hours, closing date, immediate-start and work-from-home flags and a company page link; pets carry age and a ready-to-leave date.

What Else Can Be Scraped with Gumtree Scraper

Beyond the advert body, what matters is how the site is addressed and how far it lets a crawler go:

• Browse structure - an address stacks section, subcategory and attribute onto a location path, so one make, one model, one town resolve to a single readable URL. A distance control widens the set to a radius around a postcode.
• Pagination and its ceiling - results come thirty to a page and stop at fifty pages. That is 1,500 adverts per query whatever the headline count says: a London motors browse reported roughly 22,000 matches and still capped there. Coverage comes from splitting a query by town, subcategory and price band.
• Seller shopfronts - each advert links to the rest of that seller's stock, with a total advert count and how many were posted in the past year. A feedback summary exists but is empty on most accounts - a bonus field, not a dataset.
• Sitemaps - a robots file, per-section category indexes and an advert sitemap rebuilt every hour. One hourly file in September 2026 listed roughly 2,800 fresh adverts - the cheapest way to catch new stock without re-walking browse pages.

What is missing matters as much: Gumtree shows no per-advert view counter and no seller trust score, so neither appears in what we deliver.