Redbubble scraper: print-on-demand design and price tracker

This Redbubble scraper searches Redbubble by keyword and returns one flat row per search result, a design printed on a product: the title, the artist, the product type and variant, the USD price, stock, the design's tags, the product link and the images. Sort by newest, trending, best selling or relevance, run it on a schedule and diff the dataset to get a print-on-demand trend, new-design and price watch: which new cat, retro or nurse designs appeared this week, which tags they use, and what they sell for.

No account, no API key, no login, no browser. One request returns up to 90 results. Nothing is fetched per result: every row comes from the search pages.

Input

Field	Notes
searchQueries	One or more search queries, one per line (cat, black cat sticker, retro sunset, 猫). Each query is read separately. A result found by two queries is returned once, under the first. A query with no match returns nothing and costs nothing.
maxItems	Results to read per query (default 100, up to 900). Redbubble sends up to 90 per page and serves only the first ten pages of any query, so about 900 is all one query can give, whatever the match count.
sortOrder	relevant (default), trending, recent (newest designs first) or top selling.
proxyConfiguration	Apify residential proxy by default. Redbubble refused datacenter addresses when this actor was built.

Example input:

{
  "searchQueries": ["cat", "black cat sticker"],
  "maxItems": 300,
  "sortOrder": "recent"
}

Output

Prices are numbers; currency is USD (every row measured, from exit addresses in Japan, Serbia and Brazil).

Field	Notes
query, sort, position	The query that first found the result, the sort order asked for, and the result's place in the pages read for that query (1 is first, counted over every entry of the pages walked, before duplicates are dropped). The site shuffles part of its order between requests, so this is the position in this run, not a stable rank.
product_id	Redbubble's inventory id, for example 110162644_bluey-gromit_TBCTK: the design id, a product type code and a blank id. Unique per row.
work_id	The design's id. It grows with upload time, so a larger number is a newer design; the same work_id repeats when one design is listed on several products.
title	The design title, whitespace collapsed.
artist_username, artist_name	The artist's shop name as it appears in links, and the display name. artist_name is null if the page sends none.
product_type	The product, as the site names it: Sticker, Classic T-Shirt, Poster, Magnet.
variant	The default variant the price is for, as the site labels it: Small (2.9 x 2.9 in) for a sticker, Black, Front, M for a shirt. Null if the page sends no option labels.
url	The product page link.
image_url, artwork_url	The product photo the card shows, and the bare artwork image. Null if the page lists none.
currency, price	The price of the default variant, as the page shows it.
discount_percent, discount_amount	The page's discount fields. Null when the site sends 0 (no markdown), never 0. See the note below.
in_stock	True when the page says in-stock, false for any other status (out-of-stock was seen). Null if the page sends no status.
tags	The design's tags as the artist set them, usually 10 to 15. Empty list when there are none (63 of 3,297 rows measured).
mature_content	The page's mature-content flag for the design.
query_total	How many results Redbubble said the query matches. It drifts by a few between requests (1,554,528 and 1,554,539 for cat minutes apart) and says nothing about depth: only ten pages can be read.

Example row (real values, 2026-10-29):

{
  "query": "cat",
  "sort": "recent",
  "position": 2,
  "product_id": "185127287_sticky-runners_EJUG5",
  "work_id": 185127287,
  "title": "Catostrophes the bumbling cat we love - pink",
  "artist_username": "Asterisk-a",
  "artist_name": "Asterisk-a",
  "product_type": "Sticker",
  "variant": "Small (2.9 x 2.9 in)",
  "url": "https://www.redbubble.com/i/sticker/Catostrophes-the-bumbling-cat-we-love-pink-by-Asterisk-a/185127287/7sgk",
  "image_url": "https://ih1.redbubble.net/image.6270028846.7287/st,small,507x507-pad,600x600,f8f8f8.jpg",
  "artwork_url": "https://ih1.redbubble.net/image.6270028846.7287/bg,f8f8f8-flat,750x,075,f-pad,750x1000,f8f8f8.jpg",
  "currency": "USD",
  "price": 2.78,
  "discount_percent": null,
  "discount_amount": null,
  "in_stock": true,
  "tags": ["cat", "kitty", "cat lover", "kitten", "catshirt", "cat mom", "cute animals", "pet", "cat owner", "pussycat", "funnytee", "cartoons", "pet humor", "cat dad", "cat christmas"],
  "mature_content": false,
  "query_total": 1554539
}

Honest nulls

A field the page did not send is null, never 0 or "". tags is an empty list when there are none.

• No sales, no favourites, no upload date. The search page carries none of them. work_id is the only age signal, and sortOrder: recent lists the newest designs first.
• Discounts are unverified. The page has discount fields on every price, but none of the 5,000+ rows measured on 2026-10-29 had a discount, so it is not known whether price is before or after one. Read discount_percent as "the site flagged a markdown" and compare price over time.
• The price is one variant. Stickers and posters show their default size; shirts show their default colour, print location and size (variant). It is not the price range of the product.

How it fetches

It reads the result list that redbubble.com's own search page embeds in its page data (__NEXT_DATA__). Measured 2026-10-29: HTTP 200 HTML (about 1 MB a page) from a plain connection and from Apify residential addresses; the pick step measured datacenter addresses 0 of 10 and residential 9 of 10. A refusal (403, 429, 5xx), a 200 that carries no page data, a page ahead of the one asked for, or a dropped connection is treated as the address being blocked, and the actor rotates to a fresh one and retries.

Things worth knowing, all measured:

• Ten pages, then nothing new. Every large query reports ten pages. Asking for page 11, 12 or 13 returns HTTP 200 with the page count still at 10 and a full page of results that is not a new page. The actor stops at the last page the site links to and ignores such an answer. A short list ends sooner (a query with 162 matches gave 90 and 80 results on its two pages; a third page was empty).
• Pages hold up to 90 results, although the site says 82. The first page of recent held 82, a last page 80.
• The order is not stable. The same search asked twice returned overlapping but different lists (72 of 90 results shared on relevant, 66 of 82 on recent), and a result can appear on two pages of one walk. The actor returns and charges a result once per run.
• Sorting is checked. A sort name the site does not know is silently answered as relevant; the actor reads which sort the page applied and fails the query instead of returning the wrong order.
• Loose matching. Multi-word queries match loosely: invented three-word phrases matched about 16,000 results. A single rare word narrows sharply (qwertyuiop matched 9).
• Country and currency follow the exit address, and are not settable by parameter (country= and currency= are ignored). Prices were USD from every exit measured.

If Redbubble changes its page data, the run fails with a message instead of returning empty rows.