Carousell Scraper - Listings, Prices & Seller Leads

Scrape Carousell listings, prices, conditions, photos, meet-up locations and seller leads across Singapore, Malaysia, the Philippines and Hong Kong - fast, over plain HTTP, with no browser and no login.

Carousell is Southeast Asia's largest consumer marketplace. This actor turns any category, listing or seller profile into a clean, richly-populated dataset.

---

Why this scraper

Most Carousell scrapers drive a headless browser per page. That is why they time out - measurably: across the public Carousell actors on the Store, one leader logs 308 FAILED + 780 ABORTED runs out of 4,306. This actor takes a different route.

Carousell server-renders its entire application state into the page. We parse that state directly, so:

• No browser. ~1-2 s per page instead of 15-30 s. No Chromium, no memory spikes.
• No timeouts by construction. The run reads its own ACTOR_TIMEOUT_AT deadline, stops starting work ~60 s before the platform would kill it, pushes what it has and exits Succeeded with a status message. A run can never end TIMED-OUT.
• Never throws. A transient rate-limit ends the run Succeeded with 0 items and an explanatory message - it does not fail, and pay-per-event means you are not charged for it.
• Richer records. The state blob carries fields the rendered page never shows: exact meet-up GPS coordinates, postal codes, seller verification flags, follower counts and the positive/neutral/negative review split.

We parse JSON, not CSS classes. Carousell's class names are obfuscated build hashes that rotate on every deploy - class-based scrapers rot within weeks.

---

What you get

Listing records - title, numeric + formatted price, condition, full description, likes, buyer-protection badge, images, category, seller handle, and an always-present listedAt timestamp.

Enriched listing details (fetchDetails) - 50+ fields including brand, availability, offers/comments counts, full category path, meet-up address with postal code, latitude/longitude, deal methods, and the complete seller profile.

Seller leads (includeSellerLeads) - one de-duplicated record per seller: verification flags (email / mobile / ID / Facebook), followers, feedback score, positive-rate, tenure in years, response rate, and a 0-100 leadScore ranking commercial reachability.

Deep pagination, solved

Carousell returns only 48 results per URL - there is no ?page= parameter and no infinite-scroll endpoint (verified: 12 scrolls, 0 requests fired). This actor partitions the inventory by price band and sort order, subdividing any saturated band recursively, then de-duplicates by listing ID.

Measured: 481 unique listings from a single starting point in 12 requests.

Category browsing (and why not keyword search)

Carousell's robots.txt disallows /search/, and the site enforces it - the search route returns an unresolved shell with no results to non-approved clients. Category pages are SEO-critical, explicitly crawlable, and always resolve. So this actor browses categories - the lane Carousell sanctions - and applies your keyword filter itself, while crawling, so filtering never starves your item limit.

If you point it at a category that is only a navigation hub (or a slug that does not exist in that region), it automatically falls back to that region's working subcategories rather than returning an empty run.

---

Input

Field	Description
region	sg, my, ph, hk. The proxy is auto-pinned to the matching country.
categoryUrls	Category pages to scrape - full URL or bare slug. Slugs are region-specific (see below).
keywords	Optional. Keep only listings whose title/description matches. Filtering happens during the crawl.
startUrls	Any Carousell URL - category, listing (/p/...) or seller profile (/u/...).
sellerUsernames	Pull public listings from specific sellers.
maxItems	Record cap. 0 = no limit (still exits safely before timeout).
fetchDetails	Open each listing for the full 50+ field record.
includeSellerLeads	Emit de-duplicated seller lead records.
deepPagination	Price-band + sort partitioning. On by default.
sortBy, minPrice, maxPrice	Standard filters.

Verified category slugs (ids differ per region - a Singapore slug will not work on Philippines):

Region	Working slugs
Singapore	computers-tech-213, womens-fashion-4, men-s-fashion-3, luxury-20, video-gaming-264, furniture-home-living-13, toys-collectibles-12
Malaysia	women-s-fashion-4, hobbies-toys-6245, cars-32, property-102
Philippines	luxury-20, womens-fashion-4, toys-board-games-12
Hong Kong	luxury-20, womens-fashion-4, furniture-home-living-13, toys-collectibles-12

Run it with no input at all and it browses a sample category so you always see real data.

{
  "region": "sg",
  "categoryUrls": ["luxury-20"],
  "maxItems": 100,
  "fetchDetails": true,
  "includeSellerLeads": true
}

Output (abridged)

{
  "type": "listingDetail",
  "listingId": "1451172552",
  "title": "HP Pavilion x360 Convertible Laptop",
  "price": 340,
  "priceFormatted": "S$340",
  "currency": "SGD",
  "condition": "Lightly used",
  "brand": "HP",
  "categoryPath": ["Computers Tech", "Laptops Notebooks"],
  "locationAddress": "439B Bukit Batok West Avenue 8, 652439",
  "latitude": 1.357249,
  "longitude": 103.739253,
  "dealMethods": ["meetup"],
  "sellerUsername": "collectors88",
  "sellerFeedbackScore": 5,
  "sellerFollowers": 14,
  "sellerIsVerified": true
}

Use cases

• Price intelligence - track second-hand pricing for any category across four markets.
• Resale arbitrage - spot underpriced listings the moment they appear, filtered by condition and price band.
• Seller lead generation - build a ranked list of high-volume, verified Carousell sellers with leadScore.
• Market research - measure supply, condition mix and price distribution by category.
• Geographic analysis - map listing density using real meet-up coordinates.

Use with AI assistants (MCP)

This actor works as a tool for AI assistants through the Apify MCP server. Point your assistant at it and ask for Carousell data in natural language.

Agent-ready: autonomous payments (x402 & Skyfire)

This actor is agent-ready - AI agents can discover it, run it, and pay for it autonomously, with no Apify account and no human in the loop. It uses pay-per-event pricing and limited permissions, so it qualifies for Apify's agentic-payment standards:

• x402 - an open, HTTP-native payment protocol. Agents pay per run in USDC on the Base network directly through the Apify MCP server - no account, no API key.
• Skyfire - agent-to-service payments for fully autonomous AI-agent workflows.

Building an AI agent, MCP tool, or autonomous data pipeline? This scraper is ready to plug in and pay as it goes.

Pricing

Pay-per-event - you pay only for records you receive:

Event	Price
Listing scraped	$0.003
Listing detail enriched	$0.006
Seller lead	$0.008

Notes

Only public Carousell data is collected - the same information any visitor sees without logging in. No accounts, no private messages, no personal data behind authentication. Taiwan (carousell.tw) is not supported: that site server-renders no listings, so it is deliberately excluded rather than shipped as a region that returns nothing. Please use the output in line with Carousell's terms and applicable law.