Open ranking data
Download the data behind Rankquant's rankings of hotels, films and TV, books, wine, cruises and consumer products. Each dataset page identifies the review source, scoring method, export date and coverage limits. The JSON files are free to access without an API key and carry a Creative Commons Attribution 4.0 license.
Machine-readable datasets
Every scored vertical publishes its ranked table as JSON — the same numbers the browsable catalogs render, with the scoring parameters that produced them. Free, CC-BY, no key required. Where a catalog is too large to serve in one file the download is the top slice and the remainder ships as per-category parts; each entry below states which it is, and the row count it actually contains. Every rating column is one Rankquant computed itself from individual reviewer-level ratings — no source's own star average or ratings count appears in any of these files.
Rankquant normalized wine ratings
69,234 wines from a Vivino review corpus.
JSON·25.5 MB··CC BY 4.0Rankquant normalized film and TV ratings
52,376 film or TV titles from a IMDb review corpus.
JSON·22.1 MB··CC BY 4.0Rankquant normalized consumer-product ratings
1,542,626 consumer products from a Amazon customer reviews review corpus. 1,403,756 carry a normalized score; the rest ship unscored — no Rankquant percentile, and nothing borrowed from the retailer to stand in for one. The single-file download is the top 150,000 by global percentile; the full catalog ships as 14 per-category files alongside it.
JSON·65.1 MB (top slice)··CC BY 4.0Rankquant normalized hotel ratings
71,244 hotels from a TripAdvisor / Booking.com / Agoda / Expedia / Trip.com / Ostrovok review corpus.
JSON·41.1 MB··CC BY 4.0Rankquant normalized book ratings
5,285 books from a Goodreads / Amazon review corpus.
JSON·1.8 MB··CC BY 4.0Rankquant normalized cruise-ship ratings
298 cruise ships from a Cruise Critic review corpus.
JSON·1.1 MB··CC BY 4.0
How to read and use the data
A Rankquant percentile is a position within a comparison population, not a source's star average or the percentage of customers who were satisfied. The global and in-cohort percentiles rank the mean reviewer-normalized score. The AI-adjusted percentile ranks that mean after the sample-size adjustment n/(n+53).
Review-count fields differ by category. In the hotel dataset, n1 is the total number of collected reviews and n2 is the number contributing to the normalized score. Do not substitute a platform's public review total for the sample used in a score. The dataset pages describe their columns and the methodology sets out the admission rules and formulas.
Check the export date before comparing releases. A page edit, image update or new price observation does not mean the review scores have been recomputed. The downloads do not provide a complete history of every score, raw review text or every review on the source platforms. Missing values represent unavailable measurements; they are not zeros.
A comparison you can reproduce
The Miami Beach hotel comparison uses the hotel dataset, an explicit location filter and a minimum of 50 normalized reviews. It publishes the ordering rule, the qualifying review counts and all three percentiles, so you can reproduce the shortlist or change its evidence threshold yourself.
How to cite a dataset
Cite Rankquant, the dataset title, its export date, the dataset landing-page URL and your access date. State any filters you applied. Keep a copy of the downloaded file when you need reproducibility because later exports can contain different records or scores. The license applies to Rankquant's published data; it does not grant rights to hotel photographs, platform logos or underlying third-party review text.
Use the data
Rankquant's published ranking datasets are licensed under CC-BY 4.0 — republish, cite, or remix with attribution to Rankquant + a link to the original dataset URL. For data-access questions or collaboration, email [email protected].