How iRank Score Works
Full transparency on how we score and compare movies and TV series.
What is the iRank Score?
The iRank Score is a single number between 0 and 100 that represents the overall critical quality of a title. Rather than trusting a single review site, we aggregate four of the most reputable sources using a weighted average — giving more weight to platforms with larger, more diverse reviewer pools and longer track records of reliability.
| Source | Weight | Min. votes required | Scale |
|---|---|---|---|
| IMDb | ×3.0 | 1,000 | 0–10 |
| Rotten Tomatoes | ×2.0 | 20 reviews | 0–100 → 0–10 |
| Metacritic | ×1.5 | 7 reviews | 0–100 → 0–10 |
| TMDb | ×1.0 | 100 | 0–10 |
Formula
iRank = (IMDb×3.0 + RT×2.0 + Metacritic×1.5 + TMDb×1.0) ÷ 7.5
Only sources that meet the minimum vote threshold are included. If a source is unavailable, the remaining weights are re-normalised so the score stays on the 0–10 scale.
How We Measure Similarity
Similarity between two titles is calculated using cosine similarity on a weighted tag vector. Each title is represented as a vector of binary tags across eight content dimensions. The more aligned two vectors are, the higher the match score (0–100%).
Primary and secondary genres (action, drama, sci-fi…). The strongest signal for content similarity.
Emotional register — dark, comedic, inspirational, tense. Captures feel beyond genre labels.
Narrative rhythm: slow-burn vs. fast-paced. Affects viewer experience independently of genre.
Storytelling form: linear, non-linear, anthology, episodic, heist-structure, etc.
Decade of setting, not of release. A 2010 film set in the 1940s matches other period pieces.
Core ideas explored: redemption, identity, survival, political corruption, coming-of-age…
Target demographic — children, teens, adults, families. Prevents mismatched recommendations.
Narrative point-of-view: ensemble cast, single protagonist, unreliable narrator, etc.
The final match score shown on each page is the cosine similarity of the two weighted vectors, expressed as a percentage. A score of 90%+ means the titles share almost identical content DNA across all eight dimensions.
Data Sources
We pull data from the following sources. Ratings are refreshed monthly to keep scores current as audience and critic opinion evolves.
Primary metadata source: cast, crew, genres, release dates, posters, and community ratings. Data is accessed via the official TMDb API.
The world's largest movie database. We use IMDb's weighted average rating and vote count as the highest-weight component of the iRank Score.
Tomatometer score from professional critics. Known for binary fresh/rotten aggregation across hundreds of publications.
Metascore computed from weighted critic reviews. Smaller but highly curated pool of professional reviewers.
Used to retrieve IMDb and Rotten Tomatoes scores in a single API call when direct access is not available.