Your Streaming App Carousels Could Get Better — Here's Why

When you open Netflix, Spotify, or YouTube, you usually see rows—carousels—of suggested shows, albums, or videos. These rows decide a big part of what you watch next, so streaming services work hard to get them right. But the standard mathematical tool used to judge which recommendations are better was originally designed for lists of web search results, not for rows on a screen. New research fixes that by redesigning the score so it matches how people actually browse.

The old method, called N2DCG (a formula that estimates how useful a list of recommendations is), made two flawed assumptions. First, it imagined an "ideal" carousel that doesn't respect the real limits of how many items a row can hold. Second, it decided how much to discount items further down based on web searching habits. Real eye-tracking data shows people browse carousels' rows and columns differently. The researchers propose a corrected version and tested it against both eye-tracking data and simulations of real viewing patterns. It performed much better.

So why should you care? Because this is a quiet fix behind the screen. If streaming services adopt this better measuring stick, they can honestly compare different carousel layouts. That means they'll know whether a shorter row, a taller row, or a different ordering actually helps you find what you want faster. Less time scrolling through thumbnails you ignore, more time watching something you love.

There's a catch. This is academic research, not a feature announcement. A smarter metric won't automatically change any app overnight. Streaming services would need to adopt it, and this paper doesn't promise they will. Still, it's a promising step: a measurement tool that respects real human behavior is much more likely to lead to recommendations that feel intuitive, useful, and less wasteful of your time.