After posting its latest performance update yesterday, in which CEO Mark Zuckerberg shared his enthusiasm for the company’s coming AI-powered functions, Meta has now also outlined how its advancing AI systems are already driving performance improvements, and creating more personalized, more fun experiences in its apps.
Though I’m not sure that I would describe many of its generative AI additions as “fun,” but Meta’s also talking about algorithmic improvements, increasing relevance and engagement within the feed.
First off, on content ranking: Meta says that its AI systems have led to significant improvements across the board in relevance and value in-stream.
- On Facebook, our feed and video ranking improvements in Q4 2025 delivered a 7% lift in views of organic feed and video posts, with video time spent growing double-digits year-over-year in the US.
- Facebook is now surfacing over 25% more same-day Reels compared to Q3 2025, and on Instagram we increased the prevalence of original content in the US by 10 percentage points in Q4, with 75% of recommendations now coming from original posts.
- Threads is benefiting too, with Q4 optimizations driving a 20% lift in time spent.
So, performance improvements across all of its apps, based on its evolving AI-powered relevance engine. The data here also points to the importance of video content on Facebook, and the increasing emphasis on timeliness on IG, while Threads is also steadily rising as a viable, valuable option for marketers.
Meta also points to the growth of AI dubbing, which is expanding the reach of many videos from different regions, while it also notes that content being generated via its Meta AI app tripled year-over-year in Q4 2025.
That last stat is a little skewed, because Meta released its AI content “Vibes” feed in the Meta AI app in September, which would have led to a big increase in video generation in Q4 as people checked out this option.
But even so, more people are using Meta AI to generate content, which could be good, could be bad, depending on your perspective.
Meta also highlights the value of its AI tools for advertisers, with more marketers now using its AI options to power their campaigns.
“In Q4 2025, the combined revenue run-rate of our video generation tools hit $10 billion, with quarter-over-quarter growth nearly three times faster than overall ads revenue. And our incremental attribution feature is gaining real momentum – our latest Q4 model rollout drove a 24% increase in incremental conversions compared to our standard attribution model, and the product reached a multi-billion-dollar annual run-rate just seven months after launch.”
To be clear, that “run rate” is how much Meta is making from the use of these tools, not how much they’re generating for advertisers, which is an important distinction.
Meta also says that it’s doubled the processing capacity of its ad ranking model, while it’s also launched a new run-time model across Instagram Feed, Stories, and Reels that’s increased conversion rates by 3%.
Finally, Meta also says that its click-to-message ads have continued to gain more traction with brands, as more users look to lean into direct messaging over social platform interactions, while its Business AIs, which can respond to customer queries for you, are now being used by over a million people per week in Mexico and the Philippines.
“This year, we’ll expand to more markets and add capabilities so Business AIs can not only answer questions like product availability, but also help people get things done directly in WhatsApp.”
Really, this mostly sounds like Meta’s trying to justify the billions that it’s spending on AI development to its investors, though there are some interesting points of note here in regards to how Meta is re-aligning its systems around its AI recommender systems, and where it’s focused, which could also help to guide your planning.
And this is where AI tools should provide benefit, in analyzing large data sets, like Meta’s user interactions database, then correlating behaviors to help improve performance results.
AI tools can assess more data points faster, enabling them to get down to more specific behavioral quirks and leanings, which should mean that they will be better at driving more relevance. Which is what Meta’s saying, though whether this is enough to calm investors who are unsure about the company’s expensive future vision is another matter.
About the Author
