Does Threads Have an Algorithm? How Reach Is Decided in 2026

If you've been wondering how to post on Threads in a way that actually gets seen — not just published into the void — the first thing to accept is that Threads has an algorithm, and a fairly aggressive one. Adam Mosseri has said it publicly, Meta's transparency center documents it, and anyone who's watched a post hit 50 replies in 30 minutes and another get crickets has felt it.

This guide pulls together what Meta has officially said, what the platform's behavior reveals in practice, and a 14-day plan for digging out if your reach has flatlined.

What Has Meta Publicly Said About the Threads Algorithm?

Meta has confirmed in writing that Threads uses an AI ranking system to decide what shows up in your feed and in what order. According to Meta's Transparency Center documentation on the Instagram Threads feed, the system gathers public content plus posts from accounts you follow, scores each candidate against signals about your past behavior and interests, and then orders them by how much "value" the model predicts each one will provide. That's the official framing — and it tells us three useful things. First, Threads is not chronological by default. Second, recommended (non-follower) content is part of the mix, which is why a brand-new account can occasionally land a viral post. Third, "value" is a learned prediction, not a hand-tuned rule, which is why the platform's behavior shifts every few months. Adam Mosseri said publicly in November 2024 that the team is rebalancing ranking to prioritize content from people you follow and reduce recommended content from accounts you don't, meaning unconnected reach is getting harder. This is the core architecture every other ranking signal sits on top of.

What Are the 6 Ranking Signals That Matter Most on Threads?

The six signals that move Threads reach are reply engagement, follow probability, profile clicks, like likelihood, scroll-past likelihood, and engagement velocity. Meta hasn't published a leaderboard of weighted signals, but Buffer's breakdown of the Threads algorithm — based on Meta's own transparency disclosures — identifies five core predictions the model makes for every candidate post: like likelihood, reply engagement, follow probability, profile-click likelihood, and scroll-past likelihood. Combine those with the recency/velocity signal Meta has repeatedly emphasized, and you get the six factors that actually move reach in 2026. The relative ordering matters: reply engagement and follow probability sit at the top because they're the hardest actions to fake and the strongest predictors of long-term value to a viewer. Profile clicks and likes are middle-weight confirmations that interest is real. Scroll-past likelihood is the negative signal that quietly throttles posts that don't earn attention in the first second. Engagement velocity is the multiplier across all of it — the model checks how fast the early signals arrive before deciding whether to widen the audience or kill distribution. Below is what each one actually measures:

1. Reply engagement — how likely you are to write a real reply, weighted by reply frequency and recent activity.
2. Follow probability — whether viewers tend to follow the author after seeing similar content, including your Instagram interactions.
3. Profile clicks — taps from feed into the author's profile, a high-effort signal that interest is real.
4. Like likelihood — the baseline "this matches your taste" prediction.
5. Scroll-past likelihood — a negative signal; if people thumb past your post fast, distribution narrows.
6. Engagement velocity — how quickly the first wave of interaction lands after publishing.

Anything that isn't one of these (hashtags, post length, emoji counts) is at best a weak indirect input.

Why Do Replies Outweigh Likes on Threads?

Replies outweigh likes on Threads because the platform is explicitly designed around conversation, and the algorithm rewards the actions that are hardest to fake. Meta's transparency materials repeatedly point at the same idea: actions that require more effort carry more weight. A like is one tap. A reply is reading, thinking, typing, and posting — and the model treats it accordingly. Mosseri has said that for creators trying to grow, "the sum of all your replies is about as valuable as the sum of all your posts," which is unusual phrasing from a platform exec and a real signal of how the team thinks about it. In practice this means a post with 10 substantive reply chains will out-distribute a post with 100 silent likes, and a post with reply chains where the original author keeps responding tends to keep distributing for hours rather than minutes. The corollary matters too: engagement bait that farms low-effort replies ("comment YES if you agree") is now actively penalized — Mosseri publicly acknowledged the engagement-bait problem in October 2024 and said Meta was "working to get it under control," and the algorithm has been getting more confident at detecting it since.

How Does Threads Treat Cross-Posts From Instagram?

Cross-posts from Instagram aren't punished by the algorithm in the way a lot of creators assume — but they almost always underperform native Threads posts for structural reasons. Meta has actually built features that encourage cross-posting in both directions, so there's no hidden demotion flag for "this came from IG." What hurts cross-posts is everything else. Instagram captions are written for an image-first feed and can run up to 2,200 characters; on Threads the cross-post gets cut to the 500-character post limit with the rest hidden behind a "more" expander almost nobody opens. Hashtags get tucked behind a "Show hashtags" toggle on cross-posted content. Image-led posts arrive in a text-led feed where the hook lives in the first line. And if your followers use both apps, they see the same post twice within minutes, which kills the second impression and trains the algorithm that your content is repetitive. The fix isn't to stop cross-posting — it's to rewrite the post for Threads. Lead with the text hook, cut the caption to 250 characters of actual punch, and stagger publishing by a few hours. If your strategy depends on getting both platforms to lift each other, the broader patterns in what works on Threads in 2026 cover what to keep and what to drop.

Why Do the First 30 Minutes Decide Everything?

The first 30 minutes after you post on Threads decide everything because the algorithm uses early engagement velocity as its primary signal for whether to keep distributing a post or kill it. When you publish, Threads doesn't show your post to all your followers and the wider recommendation pool at once. It seeds it to a small test group — some followers, some likely-interested strangers — and watches what happens. If replies and saves accumulate quickly, the model treats that as a confidence signal and expands the audience to a larger tier, and then a larger one after that. If the test group scrolls past, the post stays small. This is why two posts on the same account from the same author can end up orders of magnitude apart in views. It's not that Threads liked one and hated the other in any meaningful sense — it's that one cleared the early-velocity bar and the other didn't. Practical implication: post when your audience is actually online, hook hard in the first line, and reply to every comment in the first hour to keep the conversation graph alive.

How Can You Tell If You're Being Algorithmically Suppressed?

You can tell you're being algorithmically suppressed on Threads when reach drops sharply across multiple posts in a row, search visibility for your handle disappears, and your replies on bigger accounts stop surfacing in active threads. Meta gives you one official check: the Account Status dashboard, accessible inside both the Instagram and Threads apps under Settings, surfaces whether your content is currently eligible for recommendation to non-followers. If you see a "Content lowered in feed" notice, that's Meta confirming distribution has been throttled. Beyond the official dashboard, the practical signs are consistent: average post views drop by more than half compared to your trailing baseline, replies on larger accounts get hidden behind a "View more replies" expander instead of appearing in the main flow, and new followers stop trickling in even when posting cadence is unchanged. Suppression isn't the same as a full shadowban — it's usually narrower and recoverable. The full diagnostic flow lives in our deeper guide on the Threads shadowban, but if you've ruled that out, the issue is almost always ranking-level distribution loss.

A 14-Day Reach-Recovery Plan

If your Threads reach has flatlined, two weeks of disciplined behavior is usually enough to reset the signals the algorithm is reading off your account. The logic of the plan is straightforward: pause to stop feeding the model more weak data, engage on other accounts to rebuild your reply-engagement signal, publish a small number of high-quality native posts to give the algorithm fresh positive evidence, then measure against your trailing baseline to see if distribution has actually recovered. Two weeks is the right window because Threads' ranking models weight your recent activity heavily — a few days isn't enough to overwrite a bad pattern, and stretching past three weeks usually means the problem is content-fit rather than suppression. The plan below assumes you've already checked Account Status, confirmed you're not in active violation, and want to rebuild distribution. Each phase has a single job. Don't compress them — the order is what makes it work.

Days 1–3: Diagnose and pause.

• Don't post. Look at your last 10 posts and note which had the worst reply-to-view ratios. Those are the model's strongest negative training data.
• Open Account Status. Screenshot anything flagged.
• Audit who you've been replying to — if it's all the same five accounts, your reply graph has collapsed.

Days 4–7: Engage before you publish.

• Spend 20 minutes a day writing substantive replies on larger accounts in your niche. Not "great post" — real replies that add a perspective.
• Aim for 8–10 quality replies a day. This rebuilds your reply-engagement signal and gets your handle in front of audiences who actually overlap with yours.

Days 8–11: Publish 4 native posts.

• One post per day, written natively for Threads (no Instagram cross-posts during recovery).
• Lead with a one-sentence hook. Ask a real question at the end.
• Reply to every comment within the first hour. Keep each reply chain going for at least three back-and-forths.

Days 12–14: Measure and adjust.

• Compare reach on these four posts to your trailing 10-post baseline.
• If two of the four cleared the baseline, your distribution is recovering — keep the cadence.
• If none did, the issue is likely content-fit, not suppression. Test a different post type (question, opinion, observation) before assuming the algorithm hates you.

Most accounts recover meaningful reach within this window. The ones that don't usually have a content problem the algorithm is correctly identifying.

Start Posting Smarter on Threads with Postory

Once you know what the algorithm is rewarding, the actual work is figuring out which post types in your specific niche win — and that's where guessing breaks down. Postory's analyzer surfaces the exact post types your Threads audience rewards, so you stop running blind A/B tests and start writing into a pattern that's already working for you. It's the same diagnostic approach we used in how the Twitter/X algorithm decides reach and our breakdown of the LinkedIn algorithm — the platform-specific signals are different, but the diagnostic loop is the same.

Try Postory's social media analyzer — find the Threads patterns your audience actually rewards.