Streaming homepages are flooded with content. Unseen filters that noise using the strongest signals of quality: critic and audience ratings, awards, and major film canon lists.
Just movies — no TV, no filler. Each title gets an Unseen Score— a single 0–100 number blended from multiple sources, weighted by reliability — so you can quickly find what's worth your time.
How the Unseen Score works
The composite score blends multiple ratings sources into a single 0–100 number to help you find movies worth watching.
Target weights
Audience ratings from millions of users. We apply a Bayesian average so movies with fewer votes are pulled toward the global mean — a 9.0 with 2,000 votes counts much less than a 9.0 with 500,000. Movies need roughly 100,000+ votes before their raw rating is fully trusted.
The community rating from Letterboxd's film-focused audience, on a 0.5–5 scale (we normalize to 0–100). It's a more cinephile crowd than IMDb, so pairing the two captures both the broad audience and the film-buff one. We Bayesian-smooth it the same way — a glowing 4.8 from only a few hundred ratings gets pulled toward the site average until enough people have weighed in.
A weighted average of professional critic scores on a 0–100 scale. Unlike Rotten Tomatoes, Metacritic measures how good critics think a film is, not just whether they gave it a thumbs up.
The Tomatometer is the percentage of critics who gave a positive review — but it doesn't distinguish a masterpiece from pretty good. A film can score 95% with critics all saying it's decent. The least reliable input, hence the lower weight.
Why IMDb is required
A movie must have an IMDb rating with at least 1,000 votes to receive an Unseen Score. IMDb is the only source where we can verify a meaningful sample size. Without it, a film might rank highly based on a handful of critic reviews alone, which isn't a reliable signal. Movies without an IMDb match still appear in results with their individual badges, but won't have a composite score.
When other scores are missing
Not every movie has a Letterboxd, Metacritic, or RT score. When one is unavailable, the remaining weights are proportionally scaled up. For example, if only IMDb and Letterboxd are available, IMDb gets ~62% and Letterboxd ~38%.
The math
Bayesian-adjusted IMDb rating — smooths the raw rating toward the global mean based on how many votes the film has:
adjusted = (v / (v + m)) × R + (m / (v + m)) × C where R = raw IMDb rating (0–10) v = IMDb vote count m = 100,000 (trust threshold) C = 6.8 (global mean rating)
Unseen composite — weighted mean of the sources we have, with weights re-normalized when Letterboxd, Metacritic, or RT is missing:
composite = round(
Σ (score_i × weight_i / total_weight)
for i in available sources
)
target weights
IMDb (adjusted × 10) 40
Letterboxd (adjusted × 20) 25
Metacritic 20
Rotten Tomatoes 15Worked example: raw IMDb 9.2 with 150,000 votes → adjusted (150k / 250k) × 9.2 + (100k / 250k) × 6.8 ≈ 8.24. With no MC/RT, composite = round(8.24 × 10) = 82.
Where the data comes from
- IMDb — public ratings + vote counts via the official datasets
- Rotten Tomatoes + Metacritic — critic scores via OMDb
- TMDB — posters, runtimes, languages, related films, trailers
- Watchmode — streaming availability per platform
- Awards + canon— Academy Best Picture, AFI Top 100, BFI Sight & Sound, Library of Congress National Film Registry
Scoring methodology v2 — last updated July 2026.