Statistical Analysis of Steam Reviews

VaporLens now processes the original Steam review data - every review Steam returns, not just the subset that feeds the LLM analysis - and saves the statistical summary right next to the highlights, categories, and emotion breakdowns you already use. You'll now see a dedicated Statistical Analysis block on each game's detail page showing these numbers.

Included in this update:

• Sentiment × playtime matrix: Two-hour buckets separate refund-prone players, long-term detractors, and the advocates you want to amplify. Each quadrant shows counts, share of total reviews, and average hours.
• Velocity metrics: The block lists overall reviews per day, the most recent 30-day cadence, and a quarterly average based on older reviews so momentum shifts stand out quickly.
• Monthly sentiment trend: Expand the accordion to review the eight most recent months with positive/negative counts plus a simple bar to spot movement after patches or DLCs.

Behind the scenes VaporLens runs the analyzer on the full native Steam dataset each time it pulls fresh reviews, crunching playtime splits, sentiment trends, and velocity metrics before saving them with the rest of the snapshot. Working off the raw feed also means we can cover far more reviews than the LLM-ready chunks allowed.

Open any game's Stats tab to see the new block on snapshots processed from this release onward. Older games will pick it up the next time they are re-run. Feedback on additional slices or downstream uses is appreciated.