The idea
Ratings are published as a current value and consumed as one. Almost nothing keeps the series, and the series is where the interesting parts are.
What a history makes visible
- A step change
- A rating that moves sharply on a particular day usually has a cause on that day — a new version, a supplier change, a listing taken over by a different seller.
- An unusual burst
- Review counts normally grow at a rate that tracks sales. A week that does not resemble any other week is worth a second look.
- A slow drift
- Quality changes rarely announce themselves. A rating declining steadily over months is a signal no single review contains.
- A disagreement across marketplaces
- The same product rated differently in two places is either a different audience or a different product.
What it would not try to be
- Not a fake-review detector. Calling a specific review fake is a claim about a person, and it is not one a rating curve can support.
- Not a review aggregator. The individual reviews are not the object of interest; the movement is.
- Not a scraping service sold to others.