What you finish reading decides what news you see next
Yahoo rebuilt its news app around reading time after acquiring Artifact, and says longer reads teach the system to surface more stories worth finishing.
Newsroom Robots · Kat Downs Mulder: Inside Yahoo’s AI Strategy for the Future of NewsMost news apps keep score with clicks. Click a headline, and the system assumes you wanted it and finds you more like it.
Yahoo News now keeps score with staying power. Kat Downs Mulder, who leads Yahoo News as senior vice president and general manager, says the company rebuilt its app around how long people actually spend reading, how many times they come back in a month, and how long each visit lasts.
The idea came from Artifact, an AI powered news app built by the founders of Instagram. Artifact personalized its feed based on reading time instead of clicks, and added personalized alerts built around topics people followed closely. It won a devoted following but shut down in January 2024, after its founders concluded the market was not big enough. Yahoo, which had been following the project, acquired the technology and, within a couple of months, used it to relaunch the Yahoo News app.
Here is how the new loop works. The system watches for deep engagement, not a tap. If you linger on investigations into housing costs, cook through three recipes, or follow every update on a hurricane, it learns that you wanted to know and understand, not just peek.
That changes what gets surfaced. Stories that satisfy curiosity and hold attention rise, while headlines that trigger a quick, curious click but disappoint fall away. Yahoo says the shift dramatically increased users, engagement time, and page views consumed in the app.
Then comes the second turn of the wheel. When longer, more satisfying reads are what the algorithm rewards, publishers have a reason to produce more of them. Readers who liked what they found stay longer, the system learns more about what they value, and the recommendations get better. Mulder calls it a positive flywheel, something she said is "really rooted in a desire to know and understand" rather than an instinct to click, and one that encourages quality content because quality is what gets rewarded.
Questions people ask
Why are clicks such a misleading signal?
A click only shows that a headline got your attention for a second. It does not show whether the story answered your question, held your interest, or left you feeling misled. Time spent, return visits, and session length are harder to fake and closer to satisfaction.
Does personalizing by reading time trap people in a filter bubble?
It can, if it only shows you what you already like. Yahoo says it mixes widely important, editorially chosen stories with personalized picks, so readers still see what the world is talking about alongside their own interests.
Do other apps use time spent the same way?
Yes. TikTok, YouTube, Instagram, and Spotify all weight watch time, listen time, or completion rate heavily. The logic is the same: what people stick with is a better guide than what they tap once.