Netflix is rolling out a Double Thumbs Up ratings option, which will appear next to the thumbs up and down buttons, to let it refine content recommendations
For nearly five years, Netflix has had simple thumbs-up and thumbs-down icons to express viewing preferences and help its algorithms provide better recommendations.
Context & Ripple Effects
Netflix is adding more signal to a ratings system it adopted after A/B testing favored thumbs over star ratings. The change extends a long-running effort to use viewer behavior to shape discovery, including its earlier shift toward globally informed recommendations.
The new option also sits alongside Play Something, Netflix’s opt-in response to viewers’ decision fatigue. Together, the features make recommendation quality central to how Netflix reduces the friction of choosing what to watch.
First-order effects
- Netflix gains a distinct positive-preference input, allowing its recommendation system to separate titles subscribers merely liked from titles they especially valued.
- Subscribers receive a more precise way to train their own recommendations without replacing the existing thumbs-up and thumbs-down controls.
Second-order effects
- Netflix’s content-presentation systems can use the stronger preference signal to prioritize particular titles for individual viewers, reinforcing the role of personalization in discovery.
- Play Something has more useful preference data to draw on, potentially making its randomized viewing mode better aligned with a subscriber’s strongest interests.
Third-order effects
- Streaming discovery is moving from broad catalog browsing toward interfaces that continuously collect explicit and behavioral signals to determine what each subscriber sees.
- As platforms rely more heavily on recommendation inputs, the competitive value of a library increasingly depends on how effectively a service matches titles to individual viewers rather than on catalog size alone.
The trend: Netflix is deepening personalized streaming discovery by turning more moments of viewer feedback into recommendation data.