Goodreads Scraper

Extract Goodreads book metadata and full review text from the same AWS AppSync GraphQL endpoint the Goodreads website itself uses — ISBN-10 and ISBN-13, ratings, genres, editions, cover art, plus reviewer name, rating, date and complete review text — books and reviews in a single run.

• Books and reviews together. One run writes both datasets. No second Actor, no second bill, no re-scraping to get review text.
• No paid proxy by default. Talks to Goodreads' own AppSync GraphQL API rather than crawling HTML pages.
• No 10-page review ceiling. Because it uses the JSON API instead of ordinary HTML pagination, review depth is not capped at the 10 pages that HTML-based DIY scraping hits.
• About 5× cheaper per book than the verified all-in-one Goodreads competitor at $5.00 per 1,000, against $1.00 per 1,000 here.

Main Use Cases

• Book metadata dataset — build a structured catalogue with ISBN, publisher, format, page count, publication date, genres and cover URLs
• Book review dataset — collect complete review text with ratings, dates, likes, spoiler flags and reviewer profiles
• Book recommendation engine training data — join reviews to books to build rating and text-based recommendation models
• Genre reading list extraction — expand from a genre into thousands of books with ratings and reviews
• ISBN and edition data — use includeEditions to pull every other edition of a work, covering ISBN-10/ISBN-13 and format variants
• Author catalogue research — pull an author's works, contributor roles and Goodreads IDs

How It Works

1. Books are resolved from startUrls, bookIds or searchQueries — any mix of the three.
2. Each book is queried through Goodreads' AWS AppSync GraphQL endpoint and parsed into one flat Books row.
3. When includeReviews is on (the default), reviews for each book are fetched into a separate reviews dataset, filtered and sorted by reviewMinRating, reviewMaxRating and reviewSort, up to maxReviewsPerBook.
4. When includeEditions is on, sibling editions of the same work are enqueued and emitted as their own book rows, so ISBN and format variants are not missed.

Three dataset views are available: overview for a quick scan, details for the full record, and ratings for rating-only analysis.

Build Book and Review Datasets Without a Proxy

Goodreads' official public API has been documented as deprecated, which is why so many DIY approaches fall back to scraping HTML pages — and those approaches usually stall after about 10 pages of reviews and need a residential proxy to avoid blocks.

This Actor uses the same AWS AppSync GraphQL service that powers the Goodreads website, so you get structured, paginated JSON instead of scraped markup. That removes the review-depth ceiling and the proxy requirement.

Input

Field	Type	Required	Default	Notes
startUrls	array	no	https://www.goodreads.com/book/show/3	Direct book URLs. Always populated with at least the default.
bookIds	array	no	—	Numeric Goodreads book IDs.
searchQueries	array	no	—	Search terms, e.g. data science. Each becomes a discovery run.
maxItems	integer	no	100	Maximum books to collect. On the FREE tier this is clamped to 10.
includeReviews	boolean	no	true	Fetch reviews into the reviews dataset.
maxReviewsPerBook	integer	no	20	Maximum reviews per book.
reviewSort	string	no	DEFAULT	One of DEFAULT, NEWEST, OLDEST.
reviewMinRating	integer	no	—	Only keep reviews at or above this rating.
reviewMaxRating	integer	no	—	Only keep reviews at or below this rating.
includeEditions	boolean	no	false	Also emit other editions of the same work as separate book rows.
stripHtmlFromReviews	boolean	no	true	Remove HTML tags from reviewText. Raw markup stays in reviewTextHtml.
maxRequestsPerCrawl	integer	no	5000	Request budget for the run.
maxConcurrency	integer	no	5	Parallel requests.
proxy	object	no	—	Proxy configuration. Not needed on the default AppSync path.

Output

Default dataset — books (31 fields):

bookId, workId, title, titleComplete, author, authorId, authorUrl, isGoodreadsAuthor, authorRole, additionalContributors, isbn, isbn13, asin, format, pageCount, publisher, language, publishedAt, publishedAtIso, averageRating, ratingsCount, reviewsCount, ratingsDistribution, genres, genreUrls, coverUrl, coverUrlThumbnail, description, url, workUrl, sourceUrl

reviews dataset (27 fields, one row per review, each carrying its own book context so you can group or join without a second lookup):

bookId, workId, title, author, url, isbn, isbn13, averageRating, coverUrl, reviewId, reviewText, reviewTextHtml, reviewRating, reviewPreciseRating, reviewDate, reviewUpdatedAt, reviewLikes, reviewComments, reviewSpoilerStatus, reviewRecommends, reviewShelf, reviewTags, reviewerName, reviewerId, reviewerUrl, reviewerIsAuthor

Join on bookId.

Example Input

{
  "searchQueries": ["machine learning", "deep learning"],
  "maxItems": 50,
  "includeReviews": true,
  "maxReviewsPerBook": 30,
  "reviewSort": "NEWEST",
  "reviewMinRating": 3,
  "includeEditions": false
}

Example Output

{
  "bookId": "35093465",
  "workId": "35544962",
  "title": "Designing Machine Learning Systems",
  "titleComplete": "Designing Machine Learning Systems: A Practical Guide",
  "author": "Chip Huyen",
  "authorId": "136985838",
  "authorUrl": "https://www.goodreads.com/author/show/136985838",
  "isGoodreadsAuthor": true,
  "authorRole": "Author",
  "additionalContributors": ["Sam Chen (Foreword)"],
  "isbn": "1098106062",
  "isbn13": "9781098106064",
  "asin": "B09FHK7BKW",
  "format": "Paperback",
  "pageCount": 320,
  "publisher": "O'Reilly Media",
  "language": "en",
  "publishedAt": "2022-02-02",
  "publishedAtIso": "2022-02-02T00:00:00Z",
  "averageRating": 4.32,
  "ratingsCount": 2417,
  "reviewsCount": 512,
  "ratingsDistribution": { "5": 1389, "4": 612, "3": 214, "2": 108, "1": 94 },
  "genres": ["Computers", "Computer Science", "Artificial Intelligence", "Machine Learning"],
  "genreUrls": ["https://www.goodreads.com/shelf/show/computers", "https://www.goodreads.com/shelf/show/artificial-intelligence"],
  "coverUrl": "https://i.gr-assets.com/images/S/Designing-Machine-Learning-Systems.jpg",
  "coverUrlThumbnail": "https://i.gr-assets.com/images/S/Designing-Machine-Learning-Systems-s.jpg",
  "description": "An iterative, production-focused guide to building ML systems.",
  "url": "https://www.goodreads.com/book/show/35093465",
  "workUrl": "https://www.goodreads.com/work/editions/35544962",
  "sourceUrl": "https://www.goodreads.com/book/show/35093465"
}

And one row from the reviews dataset:

{
  "bookId": "35093465",
  "workId": "35544962",
  "title": "Designing Machine Learning Systems",
  "author": "Chip Huyen",
  "url": "https://www.goodreads.com/book/show/35093465",
  "isbn": "1098106062",
  "isbn13": "9781098106064",
  "averageRating": 4.32,
  "coverUrl": "https://i.gr-assets.com/images/S/Designing-Machine-Learning-Systems.jpg",
  "reviewId": "57129334801",
  "reviewText": "The chapter on data pipelines is the most practical treatment I have read. We adopted the feedback-loop structure almost verbatim.",
  "reviewTextHtml": "<p>The chapter on data pipelines is the most practical treatment I have read. We adopted the feedback-loop structure almost verbatim.</p>",
  "reviewRating": 5,
  "reviewPreciseRating": 4.5,
  "reviewDate": "2025-11-02",
  "reviewUpdatedAt": "2025-11-02",
  "reviewLikes": 87,
  "reviewComments": 4,
  "reviewSpoilerStatus": false,
  "reviewRecommends": true,
  "reviewShelf": "currently-reading",
  "reviewTags": ["machine learning", "engineering"],
  "reviewerName": "Priya R.",
  "reviewerId": "104392837",
  "reviewerUrl": "https://www.goodreads.com/author/show/104392837",
  "reviewerIsAuthor": false
}

Pricing

Pay-per-result. A book costs $1.00 per 1,000 on the FREE tier and each review costs $0.80 per 1,000. On the FREE tier maxItems is clamped to 10 books, so plan a paid tier for real dataset work.

Integrations & Automation

• Apify API — read both datasets from Python, Node or any HTTP client.
• Schedules — a daily run keeps a genre catalogue current; a weekly run is enough for review-volume tracking.
• Webhooks — push new reviews into a data warehouse as runs complete.
• MCP — callable by AI agents through the Apify MCP server.

Related Actors

• ahmed_jasarevic/link-preview-api — extract Open Graph and JSON-LD metadata from Goodreads URLs for enrichment
• ahmed_jasarevic/preply-scraper — comparable structured-catalogue extraction for the language tutoring market

FAQ

Is the official Goodreads API still available?

The official Goodreads public API has been documented as deprecated. This Actor uses the AWS AppSync GraphQL service behind the Goodreads website, which is a different endpoint and still returns structured data.

Why use this actor instead of scraping Goodreads HTML?

HTML scraping is fragile, usually needs a residential proxy, and runs into a review-depth ceiling of roughly 10 pages. The JSON path here returns structured, paginated results without that ceiling.

Can I get book reviews as well as metadata?

Yes, and that is the default — includeReviews is true. Reviews go into a separate reviews dataset, one row per review, each row carrying its own book context (bookId, title, author, isbn) so you can group without a join if you prefer.

How do I get ISBN and format data for every edition?

Set includeEditions: true. Sibling editions of the same work are then emitted as their own book rows, giving you ISBN-10, ISBN-13, ASIN, format, publisher and page count per edition.

Does it strip HTML from review text?

Yes by default, via stripHtmlFromReviews: true, which cleans reviewText while the original markup is preserved in reviewTextHtml. Set it to false if you need the raw text field to keep tags too.

What are alternatives to this actor?

The main alternatives are HTML-based Goodreads scrapers, and an all-in-one tool that also covers authors, lists and series. The all-in-one option is roughly 5× more expensive per result and defaults to a residential proxy; several competitors split books and reviews into two separate Actors, which doubles your runs. Goodreads itself, StoryGraph and LibraryThing are alternative data sources rather than equivalent scrapers.

For AI Agents & LLM Apps

Purpose: returns Goodreads book rows in the default dataset and review rows in a named reviews dataset, each review row carrying its own book context.

Minimal working input — smallest input that produces a valid run:

{ "searchQueries": ["machine learning"], "maxItems": 5, "includeReviews": false }

Books plus reviews variant:

{
  "searchQueries": ["science fiction"],
  "maxItems": 200,
  "includeReviews": true,
  "maxReviewsPerBook": 50,
  "reviewSort": "NEWEST",
  "reviewMinRating": 4
}

Editions variant:

{ "bookIds": ["35093465"], "includeEditions": true, "maxItems": 50 }

Output fields — books: bookId, workId, title, titleComplete, author, authorId, authorUrl, isGoodreadsAuthor, authorRole, additionalContributors, isbn, isbn13, asin, format, pageCount, publisher, language, publishedAt, publishedAtIso, averageRating, ratingsCount, reviewsCount, ratingsDistribution, genres, genreUrls, coverUrl, coverUrlThumbnail, description, url, workUrl, sourceUrl. Reviews (named dataset reviews): bookId, workId, title, author, url, isbn, isbn13, averageRating, coverUrl, reviewId, reviewText, reviewTextHtml, reviewRating, reviewPreciseRating, reviewDate, reviewUpdatedAt, reviewLikes, reviewComments, reviewSpoilerStatus, reviewRecommends, reviewShelf, reviewTags, reviewerName, reviewerId, reviewerUrl, reviewerIsAuthor.

Behaviors an agent should know

• startUrls is never empty — it ships with a default book URL. If you want only your own targets, pass bookIds or searchQueries and explicitly set startUrls to [], or the default book will also be collected.
• Reviews land in a separate named dataset (reviews), not in the book row. Join on bookId.
• includeReviews defaults to true, so reviews are billed unless you turn it off. Set includeReviews: false to avoid surprise charges.
• maxItems counts books, not reviews. maxReviewsPerBook controls review depth per book.
• On the FREE pricing tier maxItems is clamped to 10 books.
• includeEditions: true adds sibling editions as extra rows, so row counts will exceed the number of works you asked for. Deduplicate on workId if you want one row per work.
• stripHtmlFromReviews only affects reviewText; the raw markup is always available in reviewTextHtml.
• reviewMinRating / reviewMaxRating filter reviews only, never books.
• Three dataset views exist: overview, details, ratings.
• No proxy is required on the default path.

Billing: pay-per-result — $1.00 per 1,000 books on the FREE tier, plus $0.80 per 1,000 reviews.

Legal & Compliance

This Actor is an independent tool. It is not affiliated with, endorsed by, or sponsored by Goodreads or Amazon.

It reads only publicly accessible book pages, public review pages and public search results through Goodreads' own GraphQL service. It does not bypass authentication, does not use CAPTCHA solving, and does not attempt to defeat rate limits.

You are responsible for your own compliance with Goodreads' Terms of Service and with applicable data-protection law. Review output contains personal data — reviewer names, reviewer profile IDs and URLs. Under GDPR and comparable regimes you need a lawful basis for processing it, and you must handle it accordingly.

Please do not use extracted personal data for unsolicited commercial outreach in violation of applicable law, including CAN-SPAM, GDPR or TCPA.

SEO Keywords

goodreads scraper, goodreads books dataset, goodreads reviews, goodreads api, goodreads api deprecated, goodreads review text, book review dataset, book metadata dataset, isbn and edition data, goodreads genre lists, reading list data, book recommendation dataset, goodreads series data, author data, shelf and list scraping, book price comparison, reading statistics database, goodreads book search, analyse reader sentiment, goodreads alternatives, build a book recommendation engine, 10 review page limit workaround