ReelRecs, a free movie and TV recommendation app, has released findings from an internal evaluation showing that its top personal picks recovered 26.9% of a viewer's known favorites, compared with just 1.0% when titles were ranked by the app's crowd ratings. That means personal picks found about 27 times as many hidden favorites, a result that challenges the common practice of relying on aggregate star ratings to choose what to watch next.
The evaluation withheld titles that users had already rated as favorites and then tested whether those titles appeared in the app's top 20 recommendations. The stark difference between personal and crowd-based rankings suggests that a highly rated film can still leave an individual viewer cold, and that a one-size-fits-all approach misses the mark for many people.
ReelRecs addresses this by using a person's own likes and dislikes to generate a personal Rec Score, a visible number that indicates how well a title fits that individual's taste. Each recommendation also includes an explanation of why the title was suggested, helping viewers understand the reasoning behind the pick. In a recorded first session, six saved ratings produced a shortlist that included The Godfather, with an explanation connecting the suggestion to the user's ratings of GoodFellas and Breaking Bad. The walkthrough, available through the app, gives new viewers a concrete example of how to begin building a profile.
The app also incorporates Licensed PeerLogix viewing history, a dataset of more than a billion viewing observations, to learn connections across streaming services. While this broad data helps the system understand relationships between titles, it is the viewer's own ratings that make the recommendations personal. In a separate evaluation, ReelRecs placed a film a person loved ahead of one they disliked 94% of the time, compared with 50% for a placebo comparison, further demonstrating the impact of incorporating individual preferences.
These findings matter for viewers who feel overwhelmed by the sheer volume of content available across streaming platforms. Relying on crowd ratings can lead to disappointment when a critically acclaimed film does not resonate personally. By contrast, a system that learns from a user's own ratings can surface hidden gems that align with their unique tastes, making the search for something to watch more efficient and satisfying. The results also highlight a potential shift in how recommendation algorithms are designed, moving away from aggregate popularity toward individualized modeling.
For those interested in trying the approach, ReelRecs is free on iPhone and Android. New users can create an account, rate six movies and shows they know, including both favorites and disappointments, and then explore picks shaped by those opinions. The company provides a guide to see the recorded first session for a step-by-step look at the process.
