Steam Tag Analyzer

Website · Report an issue

Developed while working on Pocket Cat Desktop, this tool grew out of the experimental project to analyze market and Steam.

Choose accurate Steam tags and understand the games competing for your audience.

Steam Tag Analyzer helps indie developers describe their game, find relevant comparison titles, and research a niche using observed Steam data. It recommends tags that fit the game and have useful commercial associations within similar games, rather than simply promoting the most popular tags.

What you can do

• Analyze a game pitch. Get primary and secondary tags, misleading tags to avoid, and similar successful games that explain the recommendations.
• Audit an existing store page. Inspect which tags describe the game, correspond to declared store features, or deserve further investigation.
• Choose paid or free games. All game analytics offer All games, Paid games and Free games, with the selection saved on your device and in shared links.
• Build a comparison cohort. Filter by release years, tag positions, included/excluded tags, observed price, features, and publisher-name states. Save filters on your device, share links, and export game lists as CSV.
• Compare outcomes. Explore review distributions, reception, concentration, and validated H3/H6/H12 review histories. Review thresholds are configurable, with 250 reviews as the default.
• Share tag research. Every tag has a permanent page such as /tags/492-indie/, with a description and basic statistics readable without JavaScript. Charts and the configurable review threshold load in the browser.
• Research positioning. Inspect tag combinations, market trends, pricing, declared features, release timing, upcoming release windows, and publisher catalogs relevant to your niche.

Area	Pages
Tag recommendations	Analyzer, Tag Audit
Comparable games	Niche Benchmarks, Compare Games, Review Lifecycle
Market research	Tag Explorer, Tag Combos, Trends, Trending Now, Validated Momentum, Market Map
Store and launch research	Features, Pricing, Release Timing, Upcoming Releases
Catalog context	Publisher Portfolios, Tag Changelog, Data Coverage

Reading the results

Reviews are public observations, not sales, revenue, or profitability. Reaching a review threshold is an outcome to compare, not a universal definition of success. Unknown observations remain distinct from observed zeros.

Paid/free groups use the latest observed final price: positive or zero. All games also includes unknown prices. Discounts can affect membership; zero observed price is not verified free-to-play status. Historical charts use the current price group. Data Coverage and Tag Changelog describe the entire collection and taxonomy.

Current tags, prices, features, and publisher credits can differ from their launch-time values. Comparisons show associations and expose sample sizes and coverage limits. Equal-age comparisons require validated histories and release anchors; missing months are not filled with zeros. The Data Coverage page explains the snapshot, exclusions, and metric definitions.

This project is not affiliated with Valve or Steam.

How it runs

The application runs in the browser. It reads the browser SQLite database, and Transformers.js performs embedding inference in a Web Worker. The recommendation engine ranks relevant tags against comparable games.

Large runtime assets are downloaded separately:

• browser database;
• embedding models, rankers, and game vectors;
• immutable analytics snapshots and their manifest.

The analyzer and audit can download substantial data on first use. Analytics pages load smaller JSON snapshots and selected cohorts. Data and models are needed at runtime.

Database snapshots

Download monthly database snapshots from Releases. Each data-YYYY-MM release contains the full working SQLite database, including review histories, detail observations, tag snapshots and collection state.

• steam-tags-working.db.gz — compressed SQLite database.
• metadata.json — source URL, CDN modification time and ETag, packaging time, table counts, SQLite user_version, file sizes and SHA-256 checksums.
• SHA256SUMS — checksums for the archive and metadata.

Download all three files into the same directory, then verify and unpack:

sha256sum --check SHA256SUMS
gunzip --keep steam-tags-working.db.gz

The monthly database workflow runs on the first day of each month at 12:17 UTC, downloading the latest available full database from the CDN. It checks SQLite integrity and the required history tables before publishing. Collection dates vary by record; the metadata records the exact downloaded snapshot. The site's browser database remains a separate, smaller export.

Support & community

Join on Discord — chat about the project, follow progress, ask questions, share ideas, or help test.

Support development — you can support me on Patreon.

License

Client source code is available under the MIT License. Steam data, trademarks, model artifacts, and third-party dependencies retain their respective terms and licenses.

Static tag pages

Paid/free aggregate views require a data snapshot containing the corresponding population exports. Older snapshots report unavailable filtered data instead of showing all-game results. Static tag pages without JavaScript show the all-game snapshot.

npm run build validates the current CDN analytics manifest and summary checksum, generates one HTML page per tag, builds the browser bundles, and checks the resulting site. No database, model files, API keys, or backend are needed at build time. npm run dev also supports /tags/<id>-<name>/ routes.

The numeric ID identifies a tag; the readable name suffix is decorative. Existing /tags/#492 links continue to work. On GitHub Pages, obsolete or mistyped suffixes use the custom 404 page to redirect by ID with JavaScript. Canonical links point directly to generated HTML; unknown IDs stay on the 404 page.

Each page includes its snapshot date, canonical URL, sharing metadata, and a sitemap entry. Static figures use 250 reviews by default; interactive charts retain the selected threshold and use the same immutable snapshot as the HTML. Publication dates are not observation dates for every game.

The Pages workflow rebuilds on pushes, manual dispatch, and daily at 13:17 UTC (GitHub may delay scheduled runs). If the source is unavailable or fails validation, deployment stops and the published site stays intact. For an offline build, set TAG_PAGES_SOURCE to a local directory containing manifest.json and its immutable releases/<snapshot>/tags-summary.json.