Netflix to replace star ratings with thumbs up and thumbs down in coming weeks after A/B testing showed thumbs got 200% more ratings
Get ready to say good-bye to star ratings on Netflix: The company is getting ready to replace stars with Pandora-like thumbs-ups and thumbs-downs in the coming weeks.
Context & Ripple Effects
The star rating was the last piece of Netflix's old, expressive feedback layer. Earlier in 2016 the company had already retooled its recommendation algorithms around global viewing habits rather than regional ones, so the signals feeding personalization were shifting toward behavior at scale.
The replacement of stars with thumbs follows an A/B test showing thumbs drew 200% more ratings — and the later record confirms this was the start of a deliberate simplification: Netflix went on to end written reviews entirely in 2018, then added a Double Thumbs Up option in 2022 to sharpen the same signal.
First-order effects
- Members lose five-star granularity immediately: every title page switches to a binary choice, and the ratings Netflix collects should jump sharply given the 200% lift measured in testing.
- Netflix's recommendation team gains a much denser explicit-signal dataset right on top of the global-habits algorithm rework, changing what its personalization models can train on.
Second-order effects
- With thumbs capturing far more member input, written reviews become redundant — a path the corpus shows Netflix completing by removing reviews altogether while keeping the thumb system.
- A ratings interface tuned purely for signal volume raises the bar for rival services whose catalogs compete on discovery quality; matching Netflix's personalization now requires comparable feedback data, not just content spend.
Third-order effects
- If the pattern holds, streaming interfaces converge on low-friction, algorithm-fed controls over expressive ones — culminating in the kind of wholesale homepage overhaul Netflix delivered with its first major TV redesign since 2013, where fewer titles and more video put the algorithm, not the user's judgment, front and center.
- Rating systems become infrastructure for recommendation engines rather than a channel for member opinion, making engagement metrics the deciding factor in what feedback UI survives.
The trend: Streaming platforms are trading expressive user feedback like stars and reviews for high-volume binary signals that feed recommendation algorithms, with each simplification compounding the last.