Publisher Insider

Industry news, strategies, and exclusive case studies from the world leader in publisher monetization.

BROUGHT TO YOU BY 

In today’s issue: Facebook now lets people tell it what they want to see, why that means better audiences for focused pages, what it does to engagement bait and Reels, and five stories moving the market.

Your audience can now talk to Facebook. Be what they ask for.

Last night Meta announced “Your algorithm” on Facebook, an AI-powered tool that lets people shape what they see in Feed and Reels by telling Facebook in their own words. Users can add and manage topics in a dedicated space, open it from the three-dot menu on any post, and tap “Interested” or “Not interested.” Facebook may then ask a follow-up question about what they liked or disliked, and Meta says recommendations update in real time, within the same session. Meta also says large language models now help its ranking systems read the intent behind those requests. The same idea already exists on Instagram, where Your Algorithm launched earlier this year, and on Threads, where “Dear Algo” launched in February 2026. Meta has not published markets, languages, user numbers or a completion date. It says the rollout has started.

For publishers, this is a big shift in how Facebook learns what people want. Until now it guessed from behavior. Now people can tell it directly.

The best part of this update for publishers is the quality of the audience it sends. When someone types “more vintage car restorations” or “more easy weeknight recipes,” they have told Facebook exactly what they are looking for. If your page owns that topic, you get matched with people who asked for it. Those viewers watch longer, comment with real intent, share with friends who care about the same thing, and come back. Higher-quality interactions on a page usually mean better monetization per view and a healthier page over time. Our hypothesis, which we are testing across dozens of pages, is that pages with a tight, obvious topic will see engagement rate and watch time rise as the tool spreads, even if total reach moves less.

A typed request is a topic, and Facebook has to match it to content. Pages that are easy to classify are easier to match. If a person who looks at your page for five seconds cannot say what it is about, an AI ranking system will struggle too. Look at your last 30 posts. If they cover eight unrelated subjects, you are asking Facebook to guess, and people who request specific topics will be routed to pages that make the answer simple. Your page name, captions, on-screen text and the spoken words in your Reels all tell the system what you cover. Make them say the same thing.

“Not interested” now comes with a follow-up question, and people can explain in plain language what they want less of. Expect them to name it: “less clickbait,” “fewer recycled clips,” “no more rage posts.” Every one of those answers is a strong, explicit signal that hits generic and bait content directly. A single tap was easy for the system to discount. A written reason is much harder to ignore. Pages that grew on curiosity-gap posts, reposted viral clips and broad “tag a friend” content should plan for weaker distribution as the tool reaches more users. Pages that publish original work in a clear lane have little to fear here.

Meta named Reels alongside Feed, and Reels is where most unconnected reach on Facebook comes from today. When a viewer tells Facebook “more woodworking how-tos,” the Reels that win will be the ones the system can confidently read as woodworking how-tos, and that keep the viewer watching. That favors original Reels with a clear subject in the first seconds, clean on-screen text, and a recognizable format the audience can ask for again. Recycled trend clips with no topic identity are the easiest videos for this system to skip. If you want help building a truly original Reels strategy around your niche, this is the work our team does every day inside Facebook Elite Consulting.

Meta has not said how long a typed preference lasts. On Threads, a Dear Algo request was published as lasting 72 hours. Coverage of the Facebook version reports that topics sit in a managed list, which suggests they are more durable, but Meta has not confirmed whether they decay. We also do not know the rollout markets or how much weight a written request carries against normal behavior signals. We will report what we see on the pages we manage as the feature reaches them.

DO THIS

Sources: Meta Newsroom on X (@MetaNewsroom, Oct 13, 2026) and Meta Newsroom post “Your Algorithm, Your Words: New Ways to Shape What You See on Facebook” (Oct 13, 2026); The Verge (Oct 13, 2026); Social Media Today via Yahoo Tech (Oct 13, 2026); Digital Trends; PPC Land (Oct 13, 2026); ContentGrip.

Moving the market right now.

5 stories from this week that matter if you run pages and sites.