artbrokerage-scraper

Scrapes artist profiles and artwork listings from ArtBrokerage.com, a US-based secondary-market art marketplace — via the site's own public sitemaps, no login required.

Contents

• Key features
• Output
• Input
• Input examples
• Incremental (delta) mode
• How to scrape ArtBrokerage.com
• FAQ

🔑 Key features

• Two entity types, one Actor. Set entityType to artworks or artists.
• Rich per-listing detail most marketplace scrapers don't carry: condition, signature details, framing, provenance ("Purchased from" / "Provenance history"), a free-text backstory, and the certificate-of-authenticity source — genuinely server-rendered on every listing page.
• A real buyer-demand signal. Every artist record includes collector_wanted_count — genuine named-collector buy requests for that artist, a field no sibling art-marketplace Actor in this collection has (kept separate from broker_wanted_count, ArtBrokerage's own generic standing interest, present for nearly every artist and not itself a strong demand signal — see FAQ).
• A genuine for-sale/sold flag on every artwork.
• Delta mode built in. Every run classifies each item as new, changed, unchanged, or delisted against a persisted baseline.
• maxItems defaults to 50 — a fast preview, and what keeps an unconfigured run within Apify's automated 5-minute QA check. ArtBrokerage's real catalog is substantial (~40,000 listings, ~5,800 artists) — clear maxItems (null) for a full run.

📋 Output

One dataset item per artwork or artist, depending on entityType — see .actor/dataset_schema.json for the full field list, or the Output tab's Artworks / Artists views for a readable table.

Example artwork record:

{
  "source": "artbrokerage",
  "entity_type": "artworks",
  "external_id": "120261",
  "url": "https://www.artbrokerage.com/Larry-Rivers/Madama-Butterfly-1978-120261",
  "title": "Madama Butterfly 1978",
  "artist_name": "Larry Rivers",
  "artist_url": "https://www.artbrokerage.com/Larry-Rivers",
  "category": "Limited Edition Print",
  "medium": "Mixed Media Work in Eight Colors in Lithography And Three Colors",
  "size_raw": "22.75x30.5 in | 58x77 cm",
  "height_in": 22.75,
  "width_in": 30.5,
  "edition": "From the edition of 250",
  "sold": false,
  "price": 2700.0,
  "currency": "USD",
  "price_note": "Mixed Media - Blue Chip",
  "year": "1978",
  "condition": "Excellent",
  "purchased_from": "Gallery",
  "provenance": "Circle Gallery, NY",
  "certificate_of_authenticity": "Art Brokerage",
  "art_styles": ["Contemporary"],
  "art_collections": ["Well Priced", "Blue Chip - Post War - MASTERS"],
  "change_type": "new"
}

Example artist record:

{
  "source": "artbrokerage",
  "entity_type": "artists",
  "external_id": "Pablo-Picasso",
  "url": "https://www.artbrokerage.com/Pablo-Picasso",
  "name": "Pablo Picasso",
  "location": "Spain",
  "bio": "Art Brokerage: Park West Artist: Pablo Picasso Blue Chip Spanish...",
  "for_sale_count": 151,
  "recent_sold_count": 16,
  "broker_wanted_count": 4,
  "collector_wanted_count": 7,
  "change_type": "new"
}

🔧 Input

Field	Type	Default	Description
entityType	string	artworks	artworks or artists.
maxItems	integer	50	Stop after pushing this many items. Set to null for a full crawl.
mode	string	auto	auto / full / incremental — see Incremental mode.
concurrency	integer	10	How many listing/artist pages to fetch in parallel. Automatically capped to 8 for entityType=artists regardless of this value — see FAQ.
impersonate	string	chrome	curl_cffi TLS-impersonation target.
proxyConfiguration	object	off	Apify Proxy config — not needed; no anti-bot friction found.

📥 Input examples

Default preview (50 artworks):

{ "entityType": "artworks" }

Full artist directory:

{ "entityType": "artists", "maxItems": null }

Full listing catalog:

{ "entityType": "artworks", "maxItems": null }

🔁 Incremental (delta) mode

auto mode does a full scan the first time it runs for a given entityType, then only pushes new/changed items on later runs. Only an uncapped run (no maxItems limit reached) can detect delistings, sold-status changes, or price changes, or update the baseline.

🌐 How to scrape ArtBrokerage.com

ArtBrokerage has no JSON API of any kind — this Actor discovers its scope entirely via the site's own public sitemaps (sitemap-artists.xml, sitemap-listings.xml) and reads each artwork/artist page's own server-rendered HTML, the same content your browser sees.

❓ FAQ

Why is price null on some for-sale (not sold) listings? Some (typically higher-value) listings show a price tier instead of an exact figure — the site displays literal "$$$$$$$" characters with the real range only in a tooltip. This Actor reads that range into price_min/price_max instead, leaving price null since no single figure exists.

What's the difference between broker_wanted_count and collector_wanted_count? broker_wanted_count is ArtBrokerage's own standing "we'll help you sell this artist's work" interest — present at a low count for nearly every artist checked, including ones with zero for-sale listings, so it isn't a meaningful demand signal by itself. collector_wanted_count is genuine named-collector buy requests, confirmed to be entirely absent (not zero) from the page until at least one real request exists — this is the field worth watching for real demand.

Does this include realized/historical sale prices? No. A sold listing on this site shows only a plain "SOLD" tag — no price. An artist's "Recently Sold" section is a small, capped preview (also with no price), not a queryable total or archive. recent_sold_count reflects only what that preview shows, not a true total sold count.

Why is price sometimes accompanied by odd text like "23 Watchers" or "Inquire"? That's price_note — ArtBrokerage's own marketing copy next to the price on some listings. It's kept verbatim (with decorative emoji stripped) since it occasionally carries real signal (a watcher count, a negotiability hint), but it's free text, not a structured field.

Some size_raw values look implausible (e.g. "1312x1034 in"). A real, confirmed site-side data-entry bug on ArtBrokerage's own pages (a missing decimal point) — not a scraping error. The paired cm value is mathematically derived from the same (buggy) inch value, so it inherits the same error rather than independently confirming it. Passed through as-is.

Does this need a proxy? No — no anti-bot friction was found anywhere on this site (plain nginx, no Cloudflare or equivalent challenge). The origin server is simply slow (several seconds per page even under concurrency) — a full uncapped crawl takes hours, which is expected, not a sign of blocking.

Why is entityType=artists capped at concurrency 8 no matter what I set? A real large-scale test found artist pages are fast/light enough that even moderate concurrency drives a high sustained request rate, and above roughly 8 concurrent requests that rate reliably triggers a real HTTP 403 block for the rest of that run (confirmed: concurrency 20 failed within 3 minutes; concurrency 8 ran clean across 4,500+ consecutive real artist pages). This Actor enforces the safe value automatically rather than exposing a setting that can silently lose half a run's results.