Amazon search results by keyword in 20 marketplaces, plus Best Sellers and New Releases charts. One row per product: ASIN, title, price, rating, ratings count, image, sponsored flag, position on the page, chart rank. Free filters: skip sponsored (organic rank), min rating, department.
maxItems is now shared fairly: every keyword and category first gets an equal share (maxItems / targets), then rows left unused go to targets that still have products, so 5 keywords with maxItems 20 return 4 products each instead of 20 from the first; a target the cap never reaches is still never fetched. Rows read but not yet delivered wait for the next turn instead of being read again, and a later page that fails no longer loses the rows already read. Near the run timeout the run starts no new keyword or page and ends green with SUMMARY written instead of being killed. Without maxItems, results are unchanged. No prices, output fields, defaults or input fields changed.
1.7 — 2026-11-16
Never billed past maxTotalChargeUsd: each charge asks the platform only for the products that still fit under the spending limit, under one lock shared by every concurrent search, and exactly those are delivered; the run then ends — spending limit reached. Apify SDK 4.2 answers a request for more events than still fit by billing one event past the limit (a $0.30 limit was billed $0.302 on google-hotels-prices on 2026-11-16), and a limit below the price of one product was billed one product. Runs without a spending limit, or whose limit is not reached, are unchanged: no prices, input fields, defaults or output fields changed.
1.6 — 2026-11-16
maxItems (sent by AI agents) now caps the primary rows delivered and charged; add-ons follow the delivered rows. It is one total across all keywords and categories, and each page asks for its rows before it is fetched, so a run stops fetching the moment the cap is reached; keywords it never reached get a free limit status row. With pages left at its default, maxItems also lifts the page limit (up to 20 pages per keyword or category) so the run can deliver up to maxItems products; an explicit pages or maxResultsPerQuery is the stricter cap and wins. Without maxItems nothing changes. No prices, output fields, defaults or input fields changed.
2026-11-13 — Restore previous listing keywords
Restore title, description and SEO text to the November 5 baseline at the owner’s request. Preserve current runtime fixes, input/output contracts, billing, prices and defaults.
2026-11-07 — Agent contract clarification
Correct page data-index versus organic/global rank, cap-before-filter semantics, and run-wide ASIN deduplication that makes the old multi-keyword tracking recipe incomplete. Runtime, defaults, prefills, output fields and prices are unchanged.
Changelog
1.5 — 2026-11-13
Fetching: Amazon now puts a short browser-check page in front of every search results page, on every IP. The actor answers that check in the same session and fetches the page again, so searches return products again. Bestseller charts were not affected. Same inputs, rows and prices.
1.4 — 2026-10-19
Fetching: a passing proxy error on a residential session now just rotates the IP; only a 407 (no residential access) drops the run back to the cheap pool. Before, one such error disabled the residential fallback for the rest of the run. Same inputs, rows and prices. Also up to 8 tries per search page (was 5).
2026-10-16 — Keyword Planner and locale documentation
Validated search terminology against Google Ads Keyword Planner and Store shelves. Added localized usage guidance and bounded examples; retained existing English phrases. Clarified locale limitations where measured. No runtime, output fields, defaults, or pricing changed.
1.3 — listing and input form reworded, 2026-10-11
No behaviour, output or price changed; the same inputs return the same rows.
Title and description now say what one row is: one row per product with ASIN, title, price, rating, ratings count, image, sponsored flag, position on the page or chart rank; the free filters; and that the charts here render 30 of 50 entries per page (the Amazon Best Sellers Scraper fills the rest).
The input form is reordered: the keyword search inputs first (queries, domain, department, pages, maxResultsPerQuery, skipSponsored, minRating), then the optional charts (categories, chart), then proxy and pacing. Nothing renamed or removed.
Every input now says what to pass, with an example, its default and what it costs: pages gives the rows-per-page arithmetic; maxResultsPerQuery says it applies after the filters; skipSponsored says the default keeps ads in the shopper's slot; chart says Movers & Shakers is not available; proxyConfiguration gives the exact RESIDENTIAL input.
The pricing text gives the measured rows per page, says price is null where the page shows none, lists what is never charged, that there is no start fee, and that a run stops at the maximum cost you set.
README: a section mapping common requests to the exact input, with the output and empty-result contract.
1.3 — 2026-09-07
Fix: every run ended with amazon page failed after 5 tries (last: 503). Amazon started refusing the browser identity the actor sent (a Chrome/139 user agent) with a 503 on every IP, datacentre and residential alike, so a run burned its five attempts and delivered nothing. The actor now identifies as a current Chrome (152), which gets search and chart pages again from every IP tested, and a product-style page that arrives with its below-the-fold section empty is treated as a soft block that rotates the IP instead of a page with no results. No change to inputs, output rows or pricing.
2026-08-30
Reliability: a run that cannot act on its input — a URL from the wrong site, a name nothing resolves to, an empty target list, a number outside the documented range — now ends cleanly with the reason on the run card instead of as a failed run. Nothing is charged in either case; a blocked or throttled source still fails loudly.
Store listing: moved into the categories buyers actually browse (the old Business category has no Store page).
1.2 — 2026-08-29
Automatic residential fallback. Amazon refuses search pages (/s?k=) from whole datacentre ranges for minutes at a time, which occasionally ended a run with amazon page failed after 5 tries (last: 503) even though the same input had just worked. A run now starts on the cheap proxies you chose and escalates only when it is pushed back: after the same page is refused twice in a row the sessions re-open on a residential proxy and stay there for the rest of the run, so nothing repeats attempts that were never going to work. Nothing to configure, no change to pricing, and a run that is never refused never leaves the cheap pool — a run costs the least that still succeeds. Reported as http.escalated in the run's SUMMARY.
1.1 — 2026-08-29
Search pages answered 503 for the previous browser signature from 2026-08-29; the fetcher now presents a current Chrome build (verified: full pages on 4 of 4 residential IPs where the old one got 503 on 3 of 4).
chart: "movers" removed. Amazon renders Movers & Shakers client-side (the page HTML carries no products), so it always returned zero rows; the actor now says so instead of running an empty job. Diff two daily Best Sellers runs for the same signal.
Chart rows carry their image again — chart tiles moved to the images-na.ssl-images-amazon.com host, which the parser did not recognise.
A chart page renders 30 of its 50 entries in HTML. For every rank of the Top 100 use the Amazon Best Sellers Scraper , which reads Amazon's own load-more call.
1.0 — 2026-08-26
First release.
Amazon search results for any keyword, in any of 20 marketplaces: ASIN, title, price, rating, rating count, image, and the exact slot on the page.
Category charts: Best Sellers, New Releases and Movers & Shakers, each row carrying its chart rank.
Sponsored placements are flagged, not hidden — and skipSponsored drops them before billing, so keyword-rank tracking measures organic position honestly.
Multi-page walking with automatic stop when a page adds nothing new; an ASIN found by two keywords is delivered and billed once.
Filters that run before billing: skipSponsored, minRating and maxResultsPerQuery. Filtered rows are never charged.
Billing: pay per delivered product row. Status rows, filtered rows, empty searches and failed jobs are free.
Reliability: Amazon's captcha page is detected and the session rotates to a fresh IP rather than being parsed as an empty result page.