Facebook is giving users a more direct way to tell the recommendation system what they want to see.
On October 13, Meta introduced a feature called Your Algorithm, allowing users to customize feed and Reels recommendations by selecting topics they are interested in or not interested in.
As reported by The Verge, Meta is using large language models to better understand natural-language preference signals and adjust recommendations.
For creators, the important part is not the settings screen.
It is the growing role of explicit audience intent.
Recommendation systems are getting more conversational
Historically, platforms inferred interests from behavior:
- what you watched,
- what you clicked,
- what you liked,
- what you skipped,
- and who you followed.
Those signals still matter.
But users are increasingly being allowed to state preferences directly.
That changes discovery.
A creator may be surfaced not only because someone watched a similar video, but because the user explicitly said they want more content about a topic.
Topic clarity becomes more important
If platforms understand content and user preferences more semantically, vague content becomes harder to position.
Creators should be able to answer:
What is this piece of content actually about?
A podcast clip titled “This changed everything” gives the system less useful context than a clip clearly centered on:
- pricing a business,
- recovering from addiction,
- buying a first home,
- building a personal brand,
- or preparing for a job interview.
Creative hooks still matter, but topical clarity helps both humans and machines understand where content belongs.
This connects directly to GEO and AI discovery
The same trend is happening across the digital ecosystem.
Users increasingly describe what they want in natural language.
Search engines, AI assistants, video platforms, and social algorithms increasingly interpret meaning rather than only exact keywords.
That rewards content with:
- clear subjects,
- descriptive titles,
- strong metadata,
- transcripts,
- coherent captions,
- and focused conversations.
For podcasters, one strong episode can produce many pieces of content—but each piece should still have a clear topic.
The PHQ perspective
Creators should stop thinking of algorithms as mysterious systems that only reward tricks.
Recommendation systems are increasingly trying to solve a straightforward problem:
Which content best matches what this person appears to want right now?
The creator’s job is to make that match easier.
That means knowing the audience, creating specific content for real interests, and packaging it clearly.
AI may make the algorithm more sophisticated, but the creator strategy becomes surprisingly simple:
Be useful, be specific, and make the topic obvious.
Sources
Primary reporting and supporting material used for this article.
