What it reads, and what it can't see

Sources

Most sources are read directly by CleanPull's data tools from Apify's platform, as a signed-out visitor. Some come through data providers we hold an agreement with, on CleanPull's accounts, so you never need a key of your own: Bright Data, a web-data provider, for eBay, Etsy, Walmart, Target, Best Buy, Home Depot, Lowe's, AliExpress, Alibaba and Trustpilot data, and ScrapeCreators, an ad data provider, for the Meta and TikTok ad libraries. Marketplace searches through the data provider usually take 1 to 4 minutes for Walmart, eBay, Home Depot and Lowe's, and up to about 12 minutes for Etsy, Target and Best Buy; Trustpilot about 2 to 3 minutes a business. Claude says so before starting. Their reads appear on your Apify bill as provider and library events at the tools' published prices.

Most sources are read directly, as a signed-out visitor. Some come through data providers we hold an agreement with, on CleanPull's accounts, so you never need a key of your own:

• Bright Data, a web-data provider: eBay, Etsy, Walmart, Target, Best Buy, Home Depot, Lowe's, AliExpress, Alibaba and Trustpilot.
• ScrapeCreators, an ad data provider: the Meta and TikTok ad libraries.

Marketplace searches through them take 1 to 4 minutes for Walmart, eBay, Home Depot and Lowe's, up to about 12 for Etsy, Target and Best Buy. Their reads appear on your Apify bill as provider and library events, at the tools' published prices. Tap a source for how it's read and what it can't see.

What each data tool can't do

Selling in Europe

MarketMage works for sellers in the UK and the EU. Claude takes your market from your first message (a .de or .co.uk address, “for France”, a price in euros or pounds) and keeps to it, without asking again.

• 22 Amazon marketplaces, including amazon.co.uk, .de, .fr, .it, .es, .nl, .se, .pl and .com.be: prices, sellers, ranks, charts and ratings, with Google Trends read for the same country. Each Amazon product read outside the US adds $0.003, which is in the estimate.
• Margins after Amazon's fees, ex VAT, on amazon.co.uk, .de, .fr, .it and .es. Claude takes VAT off the price, then Amazon's published referral fee and its local FBA fee (fulfilment from stock in that country), and says how sure the fees are beside them. Not modelled: storage, returns, inbound placement, cross-border and Pan-EU surcharges, dangerous goods and lithium battery fees, and the digital services fee; Claude names the ones that apply to your product. Freight, duty and ads come from your own figures. On other marketplaces the margin is labelled “before fees”.
• VAT taken off. On amazon.co.uk, .de, .fr, .it, .es, .nl, .pl, .se and .com.be, and any EU shop you name, VAT is taken off the shelf price at the country's standard rate before the margin is worked out. The answer says so, assumes you're VAT-registered unless you say otherwise, and asks your rate when your product could take a reduced one.
• Your currency, a dated rate. Prices are compared in their own currency, never converted from memory. When your supplier quotes in US dollars and you sell in euros or pounds, Claude uses your rate if you give one, otherwise the European Central Bank's reference rate, said with its date, and invites your landed cost in your own currency to refine it.
• Answers in your language. Write in Spanish, German, French, Italian or Portuguese and the answer comes back in it, with the verdict labels in fixed translations.
• Search terms in your market's language: “faltbare Trinkflasche” on amazon.de and for Trends in Germany, the English term for supplier sites.
• Ads in Europe: TikTok's ads library covers the EU, the UK and Switzerland, and Meta keeps stopped ads for ads that reached the EU.

These read the US sites, and Claude says so before you agree to the cost: eBay (ebay.com), Etsy, Walmart, Target, Best Buy, Home Depot and Lowe's; barcode store checks (Amazon US, Best Buy and Google Shopping US); TikTok Shop (the US shop); and AliExpress and Alibaba searches (US sites, in US dollars). They still give a US price reference or a market to compare with.

How sites are read

• Public data only. The data tools read what any visitor can see without signing in. Nothing signs in to a personal account.
• How sites are read. The data tools read public pages the way a browser does. Their own code never solves a captcha, never signs in, never gets past a paywall and never uses anyone's account. A page that still shows a captcha to a data tool is reported as not read, with the reason, and not charged.
  • Bright Data's Web Unlocker. Some reads go through the Web Unlocker, a service of Bright Data, a data provider we hold an agreement with. It fetches the page for the data tool and handles the site's bot checks, captchas included, on Bright Data's side and under Bright Data's terms. Marketplace Search Scanner uses it for Walmart, eBay, Home Depot and Lowe's search pages, and for Amazon searches refused on every other route; Supplier Price Finder for AliExpress and Alibaba searches, and for Made-in-China pages refused on every other route; Barcode Price Checker for some stores' pages; and Amazon Keyword Rank Tracker for searches Amazon refuses.
  • Residential addresses, from Apify's proxy service, are the first choice for some reads, not only a retry: Amazon Keyword Rank Tracker reads US searches with ads, and every marketplace outside the US, through them, and Amazon Product Scraper reads products outside the US through addresses in that country. Elsewhere they are a retry: Any Store Product Monitor uses them only for a store that doesn't answer its standard addresses, billed as an add-on only for that store.
  • Amazon's browser check. Where a page runs its own JavaScript check that a normal browser passes automatically, with nothing for a visitor to do, Amazon Keyword Rank Tracker runs that check in a short-lived browser and sends the cookie it gives with its next requests, until the site asks again: at most one check per keyword and up to 5 a run. A visual captcha stops it at once.
  • Otherwise a site that refuses is retried through a provider's network (Apify's proxy service, Bright Data, or, for Google's ads library, ScrapeCreators), or the item is reported as not read.
• Each site is paced. Store, review, promotion, listing and supplier sites get at most 2 requests at a time; Amazon Best Sellers Tracker reads 2 categories at a time and Amazon Product Scraper 3 products per marketplace; the ad libraries get one request at a time per platform; the data services' requests are capped and spaced.
• robots.txt. Supplier Price Finder reads only addresses each site's robots.txt allows. Any Store Product Monitor reads a store's robots.txt only to find its sitemaps and doesn't apply its Disallow rules: give it product pages you may read.
• Built for product and business data. Reviewer names are off by default, advertisers are businesses, creatives are linked, not stored. Some results still carry public names: eBay results carry the seller's username and Etsy results the shop name (a market watch keeps eBay usernames only as salted hashes between runs), and review text can mention personal details. All of it stays in your Apify account.

You are responsible for how you use the results: see the acceptable use policy. Where the data goes and who can see it: data handling.