Hotels

Because hotels at different price points and in different cities live in different review-inflation regimes. A budget hotel that reads well for its category and a luxury property that reads merely average for its can sit at the same raw number. Global averages don't let you tell these apart.

Yes, provided enough of its reviewers carry a rating history we can normalize against. The catalog filter is 50 or more reviews per property; obscurity is not a disqualifier, thin reviewer overlap is. A hotel that clears the filter is ranked on the mean reviewer z itself, reported as measured. Thin samples are charged for separately, in the AI-adjusted percentile, rather than guessed at.

Two mechanisms, neither of them a prior over categories. First, every reviewer is normalized against their own rating history, so a reviewer who hands out top marks to everything adds little by handing out one more. Second, the AI-adjusted percentile discounts the mean by n / (n + 53), so a bulk positive push on a new property is pulled toward the corpus average until genuine coverage accumulates; bulk negative campaigns are diluted the same way. Outlier review clusters are flagged for editorial review.

No. We hold no price feed from Booking.com or anyone else, and Expedia and Hotels.com are not currently integrated. The "Book" button builds a Booking.com search from the property name, city and your chosen dates and hands you off; the rate you see is Booking.com's, on Booking.com. That link carries no affiliate tracking and earns Rankquant nothing.