Your buyer can now tell the feed what to show, in plain words. On Facebook and Instagram that control is called Your Algorithm. On YouTube it is called Custom Feeds. The short answer for a service business is this: make each post easy to place in one clear topic, name the buyer and the problem in the first lines, and give that buyer one simple next step. Vague content was already hard to sell from. Now a viewer can actively ask for less of it.
This is not the end of the algorithm. It is an extra layer on top of it. Viewing behaviour still matters, and Meta describes these controls as an added signal for its ranking systems, not a replacement for them. What has changed is who gets a say. The viewer no longer only trains the feed by scrolling. The viewer can now state a preference directly, and on Facebook Meta says recommendations can adapt within the same session.
What changed in October 2026?
On 13 October 2026, Meta announced Your Algorithm for Facebook Feed and Reels. A viewer can add and manage the topics they want to see, through the three-dot menu on a post. After tapping Interested or Not interested, Facebook may ask follow-up questions, and the viewer can explain in their own words what they liked or disliked. Meta says large language models help its ranking systems understand the interests, context and intent behind those words, and it calls the result an algorithm you can actually talk to.
Instagram got there first. Your Algorithm launched for Reels in the United States on 10 December 2025, reached Explore on 15 April 2026, and became available across Feed, Reels and Explore, globally in English, on 10 June 2026. It shows an AI-generated summary of the topics shaping a person's recommendations, and recommended posts can carry a label such as From Running. A viewer can ask for more or less of a topic, or add a new one.
YouTube announced Custom Feeds at Made on YouTube on 23 September 2026. A viewer describes a feed in a conversational prompt, for example witty podcasts for a daily 30-minute commute, and YouTube builds a homepage tab around that request. YouTube said several feeds can be saved, and that the feature is coming soon for users in the United States on web, mobile and TVs. One honest limit to note: the experimental help documentation describes one custom feed at a time, while the announcement describes saving several, so treat the exact limit as still settling and check your own account before promising a client a number.
TikTok already has its own version. Its Manage Topics control, live since June 2025, lets a viewer slide how often more than ten broad topics are recommended, and its Smart Keyword Filters remove videos containing a flagged keyword, plus synonyms found by AI. The pattern is the same across all four platforms. The viewer is moving from silent scrolling to stated preference.
How does Your Algorithm actually work?
Think of it as two signals working together. The first signal is behaviour: what a person watches, likes, comments on and sends. The second signal is now a stated preference: a topic added, a topic removed, or a sentence typed in plain language. Meta says the stated preference sits alongside behaviour, and that recommendations can update in real time on Facebook.
That distinction matters for how you read your own results. A post can be good and still miss a viewer who turned its topic down. A post can also reach a stranger who asked for exactly that topic, even if that stranger has never followed you. Reach is becoming less about shouting to everyone and more about matching a stated request.
One more limit deserves honesty. Meta's Facebook announcement describes viewer controls. It does not publish a creator playbook for Your Algorithm, and it gives no guaranteed reach for any topic. The practical advice in this article is therefore a careful reading of what the controls reward, not an official rule from Meta. If a course promises a trick that forces your post into someone's chosen topic, treat that promise with suspicion.
Why does a typed request change the job of a post?
Before these controls, a post mainly had to earn attention. Now it also has to earn placement. A system that reads plain language has to understand, in seconds, what your post is about, who it is for, and what problem it solves. If the first line could belong to any coach, consultant or agency, the system has little to match it against.
This connects directly to earlier shifts on the same platforms. Instagram already rewards original posts over reposts, and it already weighs sends as a sign that a post is worth passing to a specific person. A clear topic strengthens both effects. A viewer who asked for your topic is more likely to watch, send and remember the post, and a send to one specific friend is still one of the strongest signs a post earned its place. If you want the detail on sends, read the earlier article on Instagram sends and originality.
The same logic applies to short video search. A buyer who types a question into search is already stating intent, and a video that carries that question in clear words stands a better chance of being opened. Feed controls extend that idea from search to the feed itself. The buyer states the topic once, and the feed keeps bringing matching posts. short videos
What does a clear topic look like in practice?
Pick two or three topics and write them down. For a leadership coach, one topic might be pricing a first executive coaching offer. For a consultant, one might be turning a discovery call into a signed proposal. For a service agency, one might be onboarding a new retainer client without a messy first month. Each post should be clearly about one of them, not all three at once.
Then name that topic in three places: the first line of the caption, the on-screen text, and the words you say on camera. Language models read all three, and a viewer who asked for that topic needs to recognise it at a glance. A post that opens with "Three thoughts on growth" gives the system, and the buyer, almost nothing to match. A post that opens with "How a consultant prices a first strategy engagement" gives both of them something concrete.
Here is a small test you can run before you publish. Cover the account name and ask: could a stranger tell, from the first two sentences, who this is for and what problem it solves? If the answer is no, rewrite those two sentences. Do not rewrite the whole post. The opening is where placement is won or lost.
How do you check whether your topic is landing?
Use your own numbers, not a borrowed benchmark. For each of your chosen topics, track four numbers over four weeks: reach to non-followers, saves, sends, and profile visits. Then ask one question per topic: is this topic reaching new people, or only the people who already follow me?
If one topic keeps reaching fewer new people while the others hold steady, that is useful information. It may mean the topic is too broad, the opening is too vague, or viewers in that area have turned the topic down. Change one thing at a time. Sharpen the opening first, because that is the cheapest change to test. If the pattern stays flat after two or three sharper openings, the topic itself may be the problem, and it may be time to replace it with a narrower one.
Also watch the labels Instagram shows on recommended posts. A label such as From Running tells you which interest brought a post to a viewer. When your own posts start appearing with a consistent, sensible label, that is a sign the system is placing you where you intended. When the label looks wrong, your opening probably needs to be clearer.
What should you stop doing this week?
Stop posting the same vague caption everywhere. A keyword that works on Instagram may be a word a TikTok viewer has filtered, and a caption copied across platforms is the easiest way to get that wrong. Write each platform's caption separately, even if the idea is the same.
Stop mixing three topics into one post to hedge your bets. Hedging makes placement harder, not safer. One post, one topic, one buyer, one problem. If a second topic matters, it deserves its own post.
Stop chasing a format before you have fixed the topic. A polished Reel about nothing clear will still be hard to place. A plain, specific post about one real problem gives the feed something to match, and gives the buyer a reason to reply. Format can improve a clear post. It cannot rescue a vague one.
What about YouTube Custom Feeds and TikTok controls?
For YouTube, the practical lesson is the same, with a longer time horizon. A viewer who builds a custom feed around a commute, a mood or a routine is asking for a steady supply of matching videos. That rewards a channel with a recognisable subject and a consistent promise, not a channel that jumps between unrelated subjects each week. Title and opening seconds need to carry the subject in plain words, because a typed prompt has to match something in the video itself. If you publish short videos, keep the searchable question in the video, not only in the description.
For TikTok, the lesson is about words to avoid as well as words to use. Because Smart Keyword Filters also catch synonyms, a caption full of fashionable jargon can be filtered along with the words a viewer actually meant to avoid. Plain, specific language is safer here too. Say what the video helps with, in the words a buyer would use, and leave out phrases that add noise.
Neither point is a reason to abandon a platform. Both are reasons to be more deliberate about the words you choose on each one.
What is the one next step?
This week, choose your two or three topics, rewrite the opening two sentences of your next four posts so a stranger can name the buyer and the problem, and track non-follower reach, saves, sends and profile visits per topic for four weeks. That is the whole test. If a topic earns new reach and real replies, keep it and build a simple next step behind it, such as a short resource, a conversation, or a booked call. The earlier guide on turning Instagram comments into booked calls shows how that step can work without chasing every commenter.
If you want help turning that test into a system, where each topic has a clear offer and a clear path from post to conversation to client, that is exactly what the Social Media Strategy service is built around. Start with the topics. The feed will do the rest of the sorting, but only if you give it something clear to sort.
Sources
Meta Facebook Your Algorithm report
