How the Threads Algorithm Works in 2026: For You, Topics, Dear Algo
Meta has published what the Threads feed predicts for every post: likes, replies, profile taps, time spent, and whether you'll scroll past. Read as a creator, that list is a fairly direct brief. Here's how we use it.
Part of the guide: Threads for Creators in 2026: The Complete GuideThe Threads algorithm ranks a For You feed that blends posts from accounts you follow with recommended posts, ordered by what Meta's AI predicts each person will find most valuable. Meta's own system card lists the predictions: whether you'll like a post, reply to it, tap the author's profile, spend time reading it or scroll straight past. Topic tags and communities tell the system what a post is about, and users can steer their own feeds with Dear Algo and Your Algo. For creators, that adds up to a clear brief: write posts people stop on and answer.
How does the Threads algorithm decide what you see?
The Threads feed is chosen and ordered by an AI system in three steps, according to Meta's system card for the Threads feed:
- Gather inventory. The system collects all content from accounts you follow and "a portion of the public content available on Threads", limited to posts that follow Meta's quality and integrity rules.
- Leverage signals. It considers signals such as "how you've engaged with similar accounts, content or your interests."
- Rank content. Posts the system predicts "will provide more value for you" go higher.
The For You feed is the blend of steps one to three: followed and recommended posts mixed together. That's the feed most people open by default, and it's where non-followers find you.
This article is the deep dive on ranking from our Threads for creators guide, which covers the rest of running a profile.
What does the Threads algorithm predict?
Meta's system card lists the predictions the Threads feed uses, each with its own input signals. Read from a creator's side, every prediction is a thing your post either earns or doesn't.
| What Threads predicts | Example signals Meta lists | What it means for your posts |
|---|---|---|
| Will you like it? | Your likes and viewing time in the past week; how often you like this author | Repeat readers matter; consistency builds it |
| Will you scroll past it? | How long you spent on posts over the past day, week and month | A weak first line is actively penalized |
| Will you reply? | How much you post and reply yourself; replies to the post in the last hour | Posts that invite an answer get shown to people who answer |
| Will you tap the author's profile, then another post? | Profile taps on the post over the past month | Your profile has to reward the tap |
| Will you click the post, then another post? | Engagement with replies and replies-to-replies in the past 1 to 6 hours | Active reply threads keep a post alive |
| How long will you spend on the permalink page? | Views and taps on the post's replies in the last hour | Reply sections people read are a ranking asset |
| How long will you view the post? | Impressions; time spent by media type over 84 days | Format matters per viewer |
Two details stand out. First, several signals look at the last hour or six hours, so early activity on a post weighs heavily. Second, Meta counts engagement "within level 2 (i.e. replies to replies)": a post that sparks a conversation between other people gets credit for it.
What's the difference between For You and Following?
For You is ranked and includes recommendations; Following is chronological and doesn't. Meta's system card describes the switch: tap the Threads logo on the home screen and choose "Following" to "see content only from accounts you follow, in reverse chronological order," and "you won't see recommended content."
| For You | Following | Custom feeds and communities | |
|---|---|---|---|
| Who appears | Followed accounts plus recommended posts | Followed accounts only | Posts on a topic or from chosen profiles |
| Order | Ranked by predicted value | Newest first | Topic-based |
| How you reach new people | Here, mostly | Not at all | Through the topic or community |
| What helps a creator | Engagement predictions, topics | Posting when followers are online | Relevant tags, membership |
People can also reorder their feeds and make a custom feed their default, which Meta announced in its 2025 feature update. For a creator, that means some followers never see the ranked feed at all. Posting at hours when your audience is awake helps with them in a way no ranking trick will.
Does Threads favor people you follow now?
Yes, more than it did at launch. In November 2024 Adam Mosseri posted that Threads was "rebalancing ranking to prioritize content from people you follow, which will mean less recommended content from accounts you don't follow." His advice to creators: "you should see unconnected reach go down and connected reach go up." He called it a work in progress.
That change shapes strategy two years later. Early Threads rewarded anyone who went viral once. Now a post's best audience is the people who already follow you, and recommendations are how you get more of them. We read it as a two-step funnel:
- Recommendations bring readers. A post with a topic tag, a strong opener and an active reply thread reaches non-followers.
- The profile turns readers into followers. Only then does connected reach take over.
An account with high views and almost no new follows is stuck at step one. The fix is usually the profile and the consistency of topics, not the posts.
How do topic tags and communities affect reach?
Topic tags tell Threads what a post is about, and communities give popular topics their own feeds. You can add one topic per post, up to 50 characters, with or without a # symbol, and tapping it shows posts on that topic "in real time." Meta says that according to its internal data, posts with tagged topics "generally receive more views than those without one," and the composer now suggests trending topics and ones related to your earlier posts.
Communities build on tags. Threads' help center says you can post to a community by adding its tag, and that "even if you don't add a tag, relevant posts may be automatically surfaced into a related community." If your profile is public, anyone can see your community posts, including non-members in For You.
So a tag does two jobs: it labels the post for ranking, and it puts the post in a topic or community feed where people are browsing on purpose. Choosing which tag and which communities is a strategy question we cover in Threads communities and topic tags.
What are Dear Algo and Your Algo?
Dear Algo and Your Algo are controls that let people tell the Threads algorithm what they want more or less of for a few days. They matter to creators because they move demand for topics quickly.
| Dear Algo | Your Algo | |
|---|---|---|
| Launched | February 2026 | June 2026 |
| How it works | A public post starting "Dear Algo", e.g. "show me more posts about podcasts" | A private setting with topics to see more or less of |
| How long it lasts | 3 days | 1, 3 or 7 days, your choice |
| Who sees the request | Everyone; others can repost it to apply it to their own feed | Only you |
| Where at launch | US, UK, Australia, New Zealand | US, Canada, UK, Australia, New Zealand |
Both are short-lived on purpose. Meta's example is someone who wants more posts during a live game or fewer about a show they haven't watched. For creators, the effect is that interest in a topic can jump for three days when a Dear Algo request spreads, then drop back. We watch for that: when a niche topic is trending, a post that day reaches people who just asked to see it.
How do hides, mutes and reports affect your posts?
They teach the feed to show you less, and some of them generalize beyond one post. According to Meta's system card, hiding a post means the person won't see it again and "also helps to limit similar content from showing up in your feed." Reporting something works the same way: "the system will try to show you less content that is similar to the content you reported." People can also mute or unfollow you.
That has a practical consequence for tactics that annoy people. Rage-bait, misleading hooks and repetitive self-promotion might get replies, but every hide and report is a signal against similar posts. One heated thread with a hundred replies and twenty hides is not the win it looks like.
Do Threads posts get recommended on Instagram and Facebook?
Yes, if the profile is public. Threads' help center says that if your profile is public, "your posts may be suggested by Meta on Facebook or Instagram so people can discover, follow, and interact with you." You can switch this off per app under Privacy, then Suggesting posts on other apps. People can still reshare your posts elsewhere either way.
Meta's Insights update lets you see where people encountered your content, including on other apps. Check it before deciding the cross-app reach is worth keeping. For most of the creator profiles we run in Viralis it's free distribution and we leave it on; a creator who keeps her Instagram audience strictly separate from her Threads voice might not.
How do you use the algorithm as a creator?
You use it by writing for the predictions Meta lists: posts people stop on, reply to, and follow up by tapping your profile. Here's what we do on the creator profiles we run:
- Write the first line for the scroll. It decides the scroll-past prediction. Lead with the opinion, the question or the surprising detail.
- End with something answerable. A real question, a choice between two options, a mild disagreement. Replies and replies-to-replies are signals.
- Reply in the first hour. Several signals look at the last one to six hours. Your replies keep the thread active while it matters.
- Tag every post with one specific topic. The most specific tag that's true, not the biggest one.
- Keep a theme. Profile taps lead to "another post"; if the profile is about the same thing as the post that brought them, they stay.
- Ride short spikes. When your niche is trending, post that day.
- Check Insights weekly. Views by post, interactions, where content was discovered, link visits.
Here's a worked example of reading the numbers, with invented figures. Say a model's profile posted 12 times last week. Eight tagged posts averaged 4,000 views and four untagged ones 1,100. Two posts got over 60 replies each, and both ended with a question. Profile visits were high but follows were low. The plan for next week writes itself: tag everything, end more posts with a question, and rewrite the bio and pinned post, because readers are arriving and leaving.
Mistakes that work against the Threads algorithm
Broadcasting without replying. You lose the early-hours signals and the reply depth that keep posts in feeds.
Pasting Instagram captions. They're written to sit under a photo, not to start a conversation, and people scroll past them.
Using a broad tag on every post. A tag nobody in your audience browses wastes the one slot you have.
Engagement bait. Hides and reports generalize to "similar content." The bait costs reach on your next posts.
Judging by views alone. Since the 2024 rebalancing, follows are what make reach durable.
Posting the same text across profiles. Repetition is the first thing Meta's spam rules name. If you manage several creators, each profile needs its own words; a warm-up for new profiles is in how to warm up a Threads profile.
Start with your last ten posts: which got replies, which got skipped, which brought follows. The pattern is usually visible in one sitting.
Frequently asked questions
How does the Threads algorithm work?
Threads' For You feed blends posts from accounts you follow with recommended posts. Meta's system card says the AI gathers eligible content, scores it with signals about how you've engaged with similar accounts and topics, and ranks it by predicted value to you. It predicts things like whether you'll like, reply, tap the author's profile, spend time on the post or scroll past it.
Does Threads show posts from people you follow?
Yes. The For You feed mixes followed and recommended content, and in November 2024 Adam Mosseri said Threads was rebalancing ranking to prioritize people you follow, so creators should expect connected reach to go up and unconnected reach to go down. The Following feed shows only accounts you follow, in reverse chronological order, with no recommendations.
What is Dear Algo on Threads?
Dear Algo is a Threads feature launched in February 2026. You write a public post starting with "Dear Algo" and say what you want to see more or less of, and Threads adjusts your feed for three days. Others can repost your request to apply it to their own feed. It launched in the US, UK, Australia and New Zealand.
What is Your Algo on Threads?
Your Algo, introduced in June 2026, is the private version of Dear Algo. You tell Threads which topics you want more or less of and choose whether the request lasts one, three or seven days. Only you can see it, and Dear Algo and Your Algo requests are managed in one hub. It rolled out in the US, Canada, UK, Australia and New Zealand.
Do topic tags help reach on Threads?
Yes. Meta says that according to its internal data, posts with a topic tag generally receive more views than posts without one. You can add one topic per post, up to 50 characters. Tapping a tag shows posts on that topic in real time, and popular tags can become communities with their own feeds.
Can Threads posts be recommended on Instagram?
Yes. If your Threads profile is public, Meta may suggest your posts on Instagram and Facebook so people can discover and follow you. You can turn this off per app under Privacy, then Suggesting posts on other apps. Threads Insights shows where people encountered your content, including on other apps.
Sources
- Meta Transparency Center: Instagram Threads Feed AI systemtransparency.meta.com
- Adam Mosseri on Threads: rebalancing ranking toward people you follow (November 2024)threads.com
- Meta Newsroom: Control your Threads feed with Dear Algoabout.fb.com
- Meta Newsroom: New features to celebrate 500 million monthly users on Threads (Your Algo)about.fb.com
- Meta Newsroom: New Threads features (topics, Insights, cross-app recommendations)about.fb.com
- Threads Help: Tag a topic in your posthelp.instagram.com
- Threads Help: About communities on Threadshelp.instagram.com
- Threads Help: Manage suggestions of your Threads posts on other appshelp.instagram.com
- Meta Community Standards: Spamtransparency.meta.com