Americanas Brazil Marketplace Scraper

Scrape americanas.com.br, one of Brazil's largest general marketplaces and department stores, and get one clean row per product: identity, seller, the full Brazilian price picture (current price, pre-discount price, discount, PIX cash price and instalment terms in BRL), stock, media, every colour / size / model variant, the description, the specification table, the warranty, and customer reviews.

Two ways to say what you want, both returning as many results as you ask for:

• Search by keyword, department, brand, price band and seller.
• URL by pasting product pages, department pages or search pages straight from the store.

---

What you get per product

Group	Fields
Identity	productId, skuId, name, brand, brandId, ean, referenceId, slug, url
Category	categories, breadcrumb, categoryPath, categoryName, department
Seller	sellerId, sellerName, sellerType (first_party or marketplace), sellerCount, otherSellers[], sellerReputation
Price	price, listPrice, originalPrice, discountAmount, discountPercent, onSale, currency (always BRL)
PIX cash price	pixPrice, pixDiscountAmount, pixDiscountPercent
Instalments	maxInterestFreeInstallments, interestFreeInstallmentValue, maxInstallments, installmentPlans[] (one row per payment method and instalment count, with the interest rate and total)
Stock	inStock, availableQuantity, priceValidUntil
Media	image, images[], videos[]
Variants	variants[] (per SKU: skuId, options such as colour / size / voltage, price, pixPrice, inStock, availableQuantity, sellerName, images), variantCount
Detail	description, shortDescription, specifications (flat name to value), specificationGroups[], warranty, itemCondition, releaseDate
Delivery (opt in)	shippingCost, deliveryEstimate, pickupAvailable, delivery.options[]
Reviews (opt in)	rating, reviewCount, recommendedPercent, ratingBreakdown (all five star buckets), reviews[]
Bookkeeping	searchMode, scrapedAt, and in incremental mode changeType, changedFields, firstSeenAt, lastSeenAt

The description, specifications, variants, discount, PIX price and instalments are free

They arrive with the listing, so this actor returns them on every row without any extra request and without any extra charge. Only two things cost an extra lookup per product, and both are off by default:

• Fetch delivery details at a Brazilian postal code (CEP): shipping cost, delivery time and pickup availability. The default CEP is central Sao Paulo (01310100); replace it with your destination when you need destination-specific delivery results.
• Fetch customer reviews: ratings are looked up for a whole result page in one go at no per-product cost, and only a product that actually has reviews costs a request of its own.

A product that receives at least one of those lookups is billed a single Detail enrichment event. A catalogue-only row is never billed one.

---

Brazil only, so there is no country input

Americanas is a Brazilian retailer serving Brazil, and this was verified live rather than assumed. Four independent signals, all observed on the live site:

1. The storefront declares itself Brazilian: lang="pt-BR" and locale: "pt-BR".
2. Every price on every route is in Brazilian reais, and there is no currency or market parameter that changes it.
3. The store's own delivery quote replies shipsTo: ["BRA"] for its products.
4. Asking for a delivery quote to a United States address returns zero delivery options and the store's own error, "this item cannot be delivered to the selected coordinates".

A country input would therefore be a knob with exactly one working value, so there is none. Use a Brazilian CEP for the delivery lookup.

---

Marketplace or the store's own stock

Americanas sells its own stock next to thousands of marketplace sellers, and the two behave very differently on price, delivery and returns. Every row says which it is:

• sellerType: "first_party" means Americanas itself is the seller.
• sellerType: "marketplace" means an independent seller on the Americanas marketplace, named in sellerName.
• otherSellers[] lists every other offer on the same SKU, with its own price and stock, so you can compare the winning offer against the rest.

The Sold by filter can narrow to either side. "Store itself only" is applied by the store as a native filter. "Marketplace sellers only" is applied to the results, because the store publishes no filter that excludes its own stock.

sellerReputation is always null. The store publishes no seller rating or reputation score anywhere in its catalogue, product or checkout data. The field is emitted as an explicit null so the gap is visible rather than looking like a dropped field.

---

Reviews

Reviews are a real, fully available feature on this store, verified live on reviewed products. Turn on Fetch customer reviews to get:

• rating, reviewCount, recommendedPercent;
• ratingBreakdown, always all five star buckets, so a bucket with no reviews reads 0 rather than going missing;
• reviews[] with author, date, rating, title, text, verifiedPurchase, recommended, helpfulCount, unhelpfulCount and images[].

A product with no reviews returns reviews: [], reviewsAvailable: false and a reviewCount of 0 rather than invented values, and costs no extra request. Use Max reviews per product to cap the rows; the rating, count and breakdown stay complete regardless of that cap.

---

Which filters the store applies, and which this actor applies

Applied by the store, so they narrow the result count at the source and cost nothing extra:

Filter	How
Department	Resolved against the store's own published department list, then filtered at the source
Brand	Resolved against the store's own published brand list, then filtered at the source
Minimum / maximum price	The store's own price band
Sold by: store itself	The store's own seller filter
Sort by	All nine orderings are the store's own

Two details worth knowing about those, both measured on the live store rather than assumed:

• Picking several departments is a union, and it is walked one department at a time. The store's catalogue route cannot combine two departments in a single request: asked for both, it silently answers with only one of them. So this actor walks each department you picked as its own pass and merges the results, sharing Max products across them. Brands do combine natively, so several brands stay a single request.
• Price sorting is the store's own, and the store ranks a product by its cheapest variant. A product sold in many variants can therefore appear earlier in a price sorted run than its headline price suggests, because price is the price of the variant the row describes. The full list is in variants[].

Applied to the results, because the store publishes no filter for them:

Filter	Why
Sold by: marketplace only	The store offers no way to exclude its own stock
Minimum interest free instalments	Not a published filter
Only products that can be ordered	The store returns sold out listings alongside available ones on this route
Only reduced price products	Not a published filter

If a brand or department you asked for does not exist on the store, it is reported by name and skipped. If none of them exist, the run stops with that message instead of quietly handing back the unfiltered catalogue. If the store cannot be reached to check them at all, the run stops and says so: an unanswered check is a connection problem, and it is never reported as "that department does not exist".

A note on keywords

Some very common Portuguese keywords ("notebook", "cafeteira", "mouse") are configured on the storefront to jump to a curated landing page instead of running a search. This actor reads the catalogue directly and is not affected: a keyword search returns catalogue results for the words you typed, every time.

---

Modes

Search mode

Give one or more keywords, or leave keywords empty and pick a department to browse it whole. Each keyword is searched separately, and Max products is shared across all of them.

URL mode

Paste any mix of:

• a product page, ending in /p;
• a department page, for example /computadores-e-informatica/notebook;
• a search page, for example /busca/cafeteira.

A link that already names a page (?page=2) starts there rather than restarting at page 1. Pasted product links are always fetched first, so a large department can never fill your item cap and drop the products you actually named. The filters above still narrow what a department or search link returns.

---

Input parameters

Every input the actor accepts, in the order the input form shows them. The three MCP connector inputs are documented in their own section further down.

Start here: pick your mode

Parameter	Type	Default	Description
mode	string	search	search searches the catalogue by keyword and filters. url scrapes pasted links: product pages, department pages and search result pages are all accepted.
queries	array	(empty)	Search mode only. One or more keywords, for example notebook dell, cafeteira or tenis nike. Each keyword is searched separately and maxItems is shared across all of them. Leave empty to browse by department, brand or price alone.
urls	array	(empty)	URL mode only. Mix freely: product pages ending in /p, department pages such as /computadores-e-informatica/notebook, and search pages such as /busca/notebook-dell. A link that already names a page (?page=2) starts there.

Filters (apply to a keyword search AND to a pasted department or search link)

Parameter	Type	Default	Description
categories	array	(empty)	Keep only products in these store departments. Resolved against the store's own published department list and applied at the source. Selecting several returns products from any of them, walked one department at a time because the store cannot combine two in one request.
brands	array	(empty)	Keep only products from these brands, typed as the store writes them (Lenovo, Samsung). Accents and capitals do not matter. Resolved against the store's own brand list and applied at the source. A brand the store does not publish is reported by name and skipped; if none of your brands exist, the run stops rather than returning unfiltered results.
sellerType	string	any	any, first_party (store's own stock, applied by the store) or marketplace (independent sellers, applied to the results because the store publishes no filter that excludes its own stock).
minPrice	integer	(none)	Keep only products priced at or above this amount in BRL. Applied by the store.
maxPrice	integer	(none)	Keep only products priced at or below this amount in BRL. Applied by the store.
minInterestFreeInstallments	integer	0	Keep only products payable in at least this many interest free instalments (parcelas sem juros), 0 to 24. Applied to the results. 0 means no requirement.
inStockOnly	boolean	false	Keep only products the store will currently accept an order for. Applied to the results, because the store returns sold out listings alongside available ones.
onSaleOnly	boolean	false	Keep only products whose current price is below their advertised original price. Applied to the results. The PIX cash discount does not count here; it is returned separately in pixPrice.
sortBy	string	relevance	One of relevance, price_asc, price_desc, best_selling, best_rated, newest, biggest_discount, name_asc, name_desc. All nine are the store's own orderings, applied at the source.

Output, limits and extra lookups

Parameter	Type	Default	Description
fetchDetails	boolean	false	Read the delivery estimate, shipping cost and pickup availability for each product at the postal code below. One extra request per product, billed as one Detail enrichment event per product returned.
deliveryPostalCode	string	01310100	Destination for delivery enrichment. The default is central Sao Paulo; replace it with your own 8 digit Brazilian postal code (CEP) for destination-specific shipping cost and delivery time.
fetchReviews	boolean	false	Add each product's rating, review count, per star breakdown and individual reviews. Ratings are looked up for a whole result page in one request; only a product that actually has reviews costs a request of its own, and only those products are billed a Detail enrichment event.
maxReviewsPerProduct	integer	20	Cap on individual reviews returned per product. The rating, count and per star breakdown stay complete regardless. 0 returns all reviews a product has.
maxItems	integer	20	Maximum products returned across the whole run. 0 means unlimited.
maxPages	integer	0	Maximum result pages read per keyword, department or link. 0 means no limit, which is the default.

Resume and recurring updates

Parameter	Type	Default	Description
resumeFromRunId	string	(none)	Continue ONE interrupted run. Paste that run's id or dataset id and this run skips, and does not bill for, everything already collected there.
incrementalMode	boolean	false	Turn on for daily or recurring monitoring of the same search. The first run returns everything as NEW; later runs return only NEW, UPDATED and REAPPEARED.
stateKey	string	(none)	Incremental mode only. Name a monitoring campaign to keep its state stable, or to deliberately share state across runs. Left empty, the key is derived from the mode, keywords, links, filters and lookup settings, so two different setups never mix state.
emitUnchanged	boolean	false	Incremental mode only. Also return products that have not changed, marked UNCHANGED. This returns, and bills, extra rows.
emitExpired	boolean	false	Incremental mode only. Also return products that were present before and are no longer found, marked EXPIRED. Only produced after a complete scan, so never when maxItems capped the run or resumeFromRunId was used. This returns, and bills, extra rows.

Connection

Parameter	Type	Default	Description
proxy	object	Apify Proxy, standard pool	The standard pool is the default and is sufficient for this store. A residential pool is available but is not needed here.

---

Recurring runs: resume and incremental mode

These are two different features and they solve two different problems.

Resume from a previous run continues ONE interrupted run. Paste that run's id or dataset id and this run skips everything it already collected.

Incremental mode is for a schedule. Run the same search daily and get only what changed. The first run returns everything as NEW; later runs return NEW, UPDATED and REAPPEARED, with changedFields naming exactly what moved. UNCHANGED rows are suppressed (and not billed) unless you ask for them. EXPIRED rows are only produced once a run has fully scanned the tracked search, so a run capped by Max products, a resumed run, or a run that collected nothing can never wrongly tombstone live products.

State is kept per search, so two different keyword or filter setups never mix. Use State key to name a campaign, or to deliberately share state between runs.

Four fields are deliberately left out of change detection and are never named in changedFields:

• scrapedAt and priceValidUntil are clocks. The store stamps priceValidUntil about a year ahead of the request itself, so it moves on every fetch.
• the nested delivery object carries absolute estimated dates that roll forward daily. The two values derived from it, shippingCost and deliveryEstimate, ARE tracked, so a genuine shipping change is still detected.
• reviews, installmentPlans and specificationGroups are bulky. A change in any of them is still detected through reviewCount, rating, ratingBreakdown, maxInterestFreeInstallments, interestFreeInstallmentValue, maxInstallments and specifications, all of which are tracked.

---

Cost and limits

• Max products defaults to 20 so a first run cannot surprise you. Set it to 0 for no limit.
• Max result pages defaults to 0, meaning unlimited: a run stops at your item cap, at the end of the results, or when there is nothing new left to return.
• The store answers over the standard connection pool, which every Apify plan includes. That is the default and it is the cheapest option. A residential pool is available if you want one, but this store does not need it.
• A run that could not read any result page fails loudly with the reason. It never reports "succeeded, 0 items" when the real problem was a connection.

---

Example row (abridged, values illustrative)

{
  "productId": "8365907",
  "skuId": "8844834",
  "name": "Notebook Lenovo IdeaPad Slim 3 Intel Core 3 100U 8GB RAM SSD 256GB 15.3\" Full HD Windows 11",
  "brand": "Lenovo",
  "url": "https://www.americanas.com.br/notebook-lenovo-ideapad-slim-3-.../p",
  "department": "Computadores e Informática",
  "categoryPath": "Computadores e Informática > Notebook",
  "sellerId": "1",
  "sellerName": "AMERICANAS SA",
  "sellerType": "first_party",
  "sellerReputation": null,
  "currency": "BRL",
  "price": 3998.95,
  "originalPrice": 3998.95,
  "discountPercent": null,
  "onSale": false,
  "pixPrice": 3799.0,
  "pixDiscountPercent": 5.0,
  "maxInterestFreeInstallments": 8,
  "maxInstallments": 12,
  "inStock": true,
  "warranty": "12",
  "itemCondition": "Novo",
  "variantCount": 1,
  "rating": 5.0,
  "reviewCount": 2,
  "ratingBreakdown": { "1": 0, "2": 0, "3": 0, "4": 0, "5": 2 },
  "shippingCost": 12.9,
  "deliveryEstimate": "0bd",
  "pickupAvailable": true,
  "searchMode": "search",
  "scrapedAt": "2026-08-27T13:00:00Z"
}

---

Send results into your apps (MCP connectors)

Optionally pipe results into Notion, Linear, Airtable or Apify through a Model Context Protocol (MCP) connector. Authorize one under Apify, Settings, API and Integrations, then pick it in the input. Notion gets a page per item; the other connectors get a best effort write or digest. This is a delivery step after the scrape: the export never changes the dataset, and a connector that cannot be reached never breaks the run.

Parameter	Type	Default	Description
mcpConnectors	array	(empty)	Which authorized MCP connectors to send results to. Authorize one under Apify, Settings, API and Integrations, then select it here. Leave empty to skip.
notionParentPageUrl	string	(empty)	Notion connector only. URL or id of the Notion page under which item pages are created. Ignored by the other connectors.
maxNotifyListings	integer	50	Cap on items written to each connector per run (1 to 1000). Does not affect the dataset.