Amazon Deals Scraper

Scrape Amazon's Today's Deals grid (/deals) across 23 Amazon marketplaces. Get every discounted product's ASIN, title, images, brand, department/category, variation options and pinned status. Filter by category, brand, price range, discount range and deal type (see FAQ for the current status of the minRating filter). No login required.

What this actor does

• 23 marketplaces — amazon.com, .co.uk, .de, .fr, .it, .es, .ca, .com.au, .co.jp, .in, .com.br, .com.mx, .nl, .se, .pl, .com.tr, .ae, .sa, .sg, .eg, .ie, .com.be, .co.za
• Filters: category/department, brand, deal price range, discount percent range, deal type (real collection bubbles Amazon shows on the page); a minRating input also exists but currently has no effect since Amazon's Deals grid never exposes per-deal ratings (see FAQ)
• Deal fields: title, images, deal/list price, savings amount/percent, deal badge text/messaging, countdown-badge expiry timestamp, claim progress, deal state/promotion type, lightning-deal flag, brand name/ID, department names/IDs, category, variation options, pinned flag
• Pagination: walks the deals grid page by page up to your limits
• Empty fields are omitted

Output per deal

• asin — Amazon Standard Identification Number
• title — product title
• sourceUrl — canonical product URL
• images[] — product image URLs (hi-res and low-res variants)
• dealPrice, listPrice — {value, currency, display} — included whenever Amazon's Deals grid carries live pricing for that specific deal (a real subset of deals — typically roughly half on any given fetch — genuinely have no price data attached, e.g. brand-level coupon promos; never fabricated, see FAQ)
• savingsAmount, savingsPercent — included whenever Amazon exposes them (derived from the deal's own price fields and/or its badge percentage, whichever is present)
• dealBadgeText — the deal grid's own badge label (e.g. "15% off"), when shown
• dealBadgeMessage — the badge's messaging line (e.g. "Limited time deal"), when shown
• dealEndsAt — ISO-8601 UTC expiry timestamp for a Lightning-Deal-style countdown badge (e.g. "Ends in"), when Amazon exposes one for that deal
• percentClaimed, percentClaimedMessage — live claim-progress for the deal (e.g. 34.0, "34% claimed"), when Amazon exposes it for that deal
• dealState — the deal's live state as Amazon reports it (e.g. AVAILABLE)
• dealPromotionType — Amazon's internal promotion classification for the deal (e.g. BEST_DEAL, LIGHTNING_DEAL)
• isLightningDeal — boolean, whether Amazon's own cart data flags this as a Lightning Deal
• brandId, brandName — Amazon's internal brand ID and its resolved display name
• departmentIds[], departmentNames[] — Amazon's internal department/category IDs and resolved names
• categorySymbol, productType — Amazon's internal category taxonomy codes
• brandLogoUrl — the brand's logo image, when shown
• variationDimensions — variant/size/color options and their ASINs (when present)
• isPinned — whether Amazon pinned this deal
• dealId — stable deal/promotion ID (when exposed by the widget)
• position — 1-based rank in the deals grid
• recordType: "deal", scrapedAt

When no deals are emitted

If a run ends with zero "deal" records, it also pushes exactly one recordType: "status" record instead of leaving the dataset empty with no explanation — so you can tell a real failure apart from a clean run that legitimately found nothing, without needing to read the run log:

• success: false — the Deals grid was never reached, with a reason of BLOCKED (couldn't get past bot-detection/redirect handling after retries — check classification for the specific cause), WIDGET_NOT_FOUND (page fetched cleanly but the deals widget wasn't present), or NO_ENTRIES (widget found but empty)
• success: true, reason: "NO_MATCHES" — the Deals grid was fetched and parsed successfully, but every entry was filtered out by your category/brand/priceMin-priceMax/discountMin-discountMax/dealType filters (or the marketplace's grid was genuinely empty that day)

Every status record also carries marketplace, message (human-readable), and scrapedAt.

Input

Field	Type	Default	Description
marketplace	string	amazon.com	Which Amazon domain's Deals page to scrape
dealType	string	(any)	Real collection bubble Amazon shows on the page — Lightning Deals / Coupons / Outlet / Customers' Most-Loved / Lowest Price in 365 Days / plus 30+ per-category bubbles (Electronics, Beauty, Fashion, Home, Toys & Games, etc.), all live-verified against amazon.com/deals
category	string	–	Free-text filter matched against department name / category taxonomy code
brand	string	–	Free-text brand substring filter
minRating	int	–	Minimum star rating (1-5) — currently a no-op; see FAQ
priceMin / priceMax	int	–	Deal price range (local currency, when price data is available — see FAQ)
discountMin / discountMax	int	–	Discount percent range (when available — see FAQ)
maxItems	int	30	Hard cap on emitted deals (1-2000)
maxPages	int	5	Hard cap on grid pages walked (1-50)
proxyConfiguration	object	AUTO	Apify proxy — AUTO by default, auto-escalates to residential on repeated blocks

Example: default US deals grid

{
  "marketplace": "amazon.com",
  "maxItems": 30
}

Example: home & kitchen deals from a specific brand

{
  "marketplace": "amazon.com",
  "category": "Home & Kitchen",
  "brand": "Bissell",
  "maxItems": 60
}

Example: UK Customers' Most-Loved deals with a 4-star minimum

{
  "marketplace": "amazon.co.uk",
  "dealType": "deals-collection-cml",
  "minRating": 4,
  "maxItems": 50
}

Use cases

• Deal aggregators — build a curated "today's deals" catalog feed
• Competitive intelligence — see which brands/categories Amazon is currently discounting
• Merchandising research — track which products get pinned/featured in the Deals grid
• Shopping assistants — surface deals matching a user's category/brand interest
• Catalog enrichment — pull variation options (size/color) and brand metadata for deal ASINs

FAQ

Does this require an Amazon account or cookies? No. The Deals page is publicly browsable; the actor scrapes it directly with no login.

Why don't I see dealPrice, listPrice, savingsPercent, dealBadgeText or percentClaimed on every record? Amazon's own Deals grid data genuinely omits price/badge/claim-progress information for a real subset of deals on any given fetch (commonly brand-level coupon promos with no single "deal price" to show). The actor never fabricates these fields — it populates them whenever Amazon's page includes them for that specific deal, and omits them entirely otherwise. All the catalog/structural fields (title, images, brand, category, variations) are always populated when Amazon shows them.

Why do a few records occasionally show an implausibly large dealPrice/listPrice? On rare fetches, Amazon's own Deals grid data itself carries an inflated numeric value for a handful of deals — independently confirmed live by inspecting the raw page data byte-for-byte, not an actor parsing/arithmetic error (the actor's savingsPercent math on these records is still internally correct — e.g. a genuinely-computed "30% off" — only the absolute base price is implausible). This appears tied to sessions Amazon treats as higher-risk/automated (the same class of session-dependent behavior that can also cause Amazon to render a different regional currency than expected); the actor does not attempt to "correct" these values since there is no reliable way to distinguish a genuinely expensive real deal from an inflated one without fabricating a number. If you filter downstream by price, treat an outlier far outside a category's normal range with appropriate skepticism.

Why is dealType filtering only "best effort" for some values? Amazon's Deals page groups deals into filter "bubbles". Category-style bubbles (e.g. Customers' Most-Loved) are backed by an explicit department list, so the actor can verify membership per deal. Promotion-style bubbles (Lightning Deals, Coupons, Outlet) aren't tagged per-item in the grid data the actor can see, so those filters pass every deal through rather than risk wrongly excluding a real match. Additionally, not every dealType value is guaranteed to render as a live bubble on every one of the 23 supported marketplaces on a given day (e.g. a smaller marketplace's Deals page may simply not have a "Jewelry" bubble today) — when this happens the actor logs a warning and passes every deal through unfiltered rather than silently returning zero results; check the run log if a dealType filter appears to have had no effect.

Why does sourceUrl return a 503 if I curl it directly? Amazon's edge (CloudFront) aggressively blocks plain, cookie-less, non-browser requests — a bare curl from a clean shell with no session/cookies/JS will get a 503 "Service Unavailable" even though the same URL loads fine for a real shopper's browser. This actor itself reaches the Deals grid successfully using a real browser session and proxy rotation; the URLs in the output are the correct, canonical Amazon product URLs and will render normally when opened in a browser or fetched with an equivalent session.

Does minRating actually filter anything? Not currently, in practice. Amazon's Deals grid data (the same page payload documented above) never includes a per-deal star rating — only structural/catalog fields do. Since the actor never excludes a deal on the absence of rating evidence (per its no-fabrication rule), minRating currently passes every deal through unfiltered rather than risk wrongly dropping real matches. The input is kept because a future/differently-trusted session could expose rating data the way it can expose pricing; if you need rating-based filtering today, filter the output dataset by stars after the run (it will simply be absent on every current record).

Do priceMin/priceMax/discountMin/discountMax exclude deals with missing price/savings data, or let them through? They let them through. Every filter in this actor follows the same rule: a deal is only ever excluded when it has real, present evidence that contradicts the filter — never on the absence of the field the filter checks (see "Why don't I see dealPrice..." above; the same real subset of deals genuinely has no price/savings data on any given fetch). So discountMin: 10 will exclude a deal it can see has e.g. a 5% discount, but will still include a deal with no savingsPercent at all — its real discount is simply unknown, not verified to be below your threshold. If your workflow needs a hard guarantee ("only deals I can confirm meet the discount floor"), filter the output dataset post-run by checking savingsPercent is present and satisfies your threshold yourself, rather than relying on the actor's own filter to have done that exclusion for you.

How many deals can I get per run? Amazon's Deals grid returns up to 30 items per page (occasionally fewer, depending on catalog/session state). The actor attempts to walk further pages by reproducing the grid's own "load more" request live in-browser; if Amazon doesn't expose a further page in a given session, the actor stops cleanly rather than guessing. Each deal's position is always a gapless, 1-based rank across the full run regardless of how many items any individual page actually returned.

What if my category/brand filter returns 0 results? These are free-text substring filters matched against the department name / category code / brand name Amazon exposes on each deal. If nothing matches, widen or clear the filter.

Is this affiliated with Amazon? No, this is an independent third-party actor that scrapes Amazon's public Deals page.

How fresh is the data? As fresh as the live page at run time — Amazon refreshes its Deals grid continuously throughout the day.

More Amazon Actors

Part of a full Amazon scraper suite — also check out:

• Amazon Category Scraper
• Amazon Product Scraper
• Amazon Best Sellers & Rankings Scraper
• Amazon Reviews Scraper
• Amazon Reviews Scraper Pro
• Amazon Offers Scraper (All Offers Display)
• Amazon Seller & Shop Scraper
• Amazon Brand Store Scraper
• Amazon Creator Shop Scraper
• Amazon Wishlist & Registry Scraper
• Amazon Keyword Suggestions Scraper