WatchScore™ is built from real audience comments and reviews - not critic reviews, and not a personalization algorithm guessing what you specifically will like.
For each title, we pull real audience comments and reviews from across the web - YouTube review videos, TMDB reviews, Trakt comments, Bluesky and X posts, Tumblr, and (for anime titles) AniList - filtering out spam, promotional content, and reactions to trailers rather than the finished film or show.
An AI model reads every comment and tags what it actually says about specific aspects of the title - pacing, acting, the ending, humor, and more - rather than just counting positive or negative words. A comment praising the acting but criticizing the pacing gets credit for both, accurately.
Similar opinions from different commenters are grouped together, so genuine agreement across many people counts more than one person's opinion restated. A point made by twenty independent commenters carries real weight; a single strongly-worded comment doesn't dominate the score just because it's vivid.
Different things matter for different kinds of titles - pacing and worldbuilding carry more weight for a slow-burn drama than for a comedy, where humor matters most. Each title's WatchScore is a genre-aware weighted combination of everything audiences said, not a single flat average.
It's not a critic score (like Rotten Tomatoes or Metacritic), and we show those separately when available for comparison. It's not personalized - every visitor sees the same WatchScore for a given title, based on the same real audience data. And for very new or low-visibility titles without enough real audience comments yet, we show a clear pending state rather than a score we can't stand behind.
WatchScore also doesn't require an account and doesn't track individual users - every visitor sees the same score and recommendations for a given title.