The Threads algorithm is an AI ranking system that predicts how likely you are to like, reply to, click on, spend time with or scroll past each post, then orders your feed by predicted value. It draws on your activity from the last 15 minutes up to the last 84 days. Replies matter more here than on most networks.
That is the short answer, and unusually for a social platform, most of it comes straight from Meta. Threads publishes a plain description of its feed ranking on the Meta Transparency Center, including the exact predictions it makes. What Meta does not publish is how much each prediction weighs, which is where independent data from Buffer, Metricool and Sprout Social helps. This guide puts the two side by side: what Threads says, what the numbers show, and how to plan and schedule posts that the system can confidently recommend.
How the Threads Algorithm Works: Key Findings
- Eight official predictions. Meta's description of the Threads feed system lists predictions such as how likely you are to like a post, create a reply, click the author's profile, or scroll past without engaging.
- Signals span 15 minutes to 84 days. The input signals are measured over windows from the last 15 minutes up to 84 days, so both fresh behaviour and long-term habits shape what you see.
- Replying lifts engagement by about 42%. Buffer's analysis of more than 128,000 Threads posts found posts where the author replied to comments earned around 42% more engagement, the biggest lift of six networks studied.
- Median engagement fell to about 3.6%. Buffer's 2026 benchmark puts Threads' 2025 median engagement rate at about 3.6%, down from about 4.4% in 2024.
- 500 million monthly users. Meta announced 500 million monthly active users in June 2026, about 10 months after passing 400 million.
- Ranking changes moved time spent by 10%. Meta's CFO said ranking optimisations in Q3 2025 alone drove a 10% increase in time spent on Threads.
- People can now steer their own feed. Dear Algo (February 2026) and Your Algo (June 2026) let users ask for more or less of a topic for one to seven days.
What Has Changed on Threads in 2026
Threads changes faster than any other Meta app, so if you read an algorithm guide from 2024, much of it is out of date. These are the changes that matter for reach as of October 2026.
Users can talk to the algorithm directly
In February 2026 Threads launched Dear Algo. According to TechCrunch, you write a public post starting with "Dear Algo" and say what you want more or less of, and your feed adjusts for three days. Because the request is public, other people can repost it to apply the same preference to their own feed. In June, Meta added Your Algo, a private version where you choose whether a request lasts one, three or seven days, rolling out first in the US, Canada, the UK, Australia and New Zealand.
For anyone posting on Threads, this is a new kind of signal. A viewer who asks for "more posts about specialty coffee" this week is explicitly telling the system to widen distribution for that topic. Posts that are clearly about one subject are easier to match to those requests than posts that mix three ideas.
Communities are out of beta
Threads introduced Communities in October 2025 across more than 100 popular topics, and by a December 2025 update you could explore over 200. Meta said it was building ranking within communities and in the For You feed so the most relevant community posts appear first. In June 2026 Communities left beta with a Communities Hub in the main menu, community icons, and a progress view showing when a topic is close to becoming a community.
The audience is bigger and returns more often
TechCrunch reported that Threads reached 500 million monthly active users in June 2026, having announced 400 million on 12 August 2025. On daily use, Mark Zuckerberg said on Meta's Q3 2025 earnings call that Threads had "recently passed 150 million daily actives", and CFO Susan Li said ranking optimisations made in that quarter "drove a 10% increase in time spent on Threads". She also said ads were now running globally in the Threads feed, which means organic posts share the feed with paid ones.
Insights now separate followers from non-followers
In September 2026 Threads rebuilt its creator Insights in the US first, adding a views-over-time breakdown of followers versus non-followers, plus labels that show whether views, profile visits and follows are "higher", "typical" or "lower" than your own norm. That split is the single most useful number for understanding the algorithm, because non-follower views are recommendation reach.
Longer posts without leaving the app
Since September 2025, Threads lets you attach up to 10,000 characters of text to a post, on top of the 500-character main post. Meta said it added this because people were screenshotting longer writing when 500 characters was not enough.
| Date | Change | What it means for reach |
|---|---|---|
| Nov 2024 | For You rebalanced towards accounts you follow | Connected reach up, unconnected reach down |
| Mid 2025 | Meta says links are now "ranked properly" | Link posts no longer a reason to avoid Threads |
| Sep 2025 | Text attachments up to 10,000 characters | Long-form ideas can live natively on Threads |
| Oct 2025 | Communities launch | New topic-based surface beyond For You |
| Feb 2026 | Dear Algo | Viewers can request topics publicly for three days |
| Jun 2026 | Your Algo, Communities out of beta, 500M monthly users | Private topic controls; communities easier to find |
| Sep 2026 | Insights redesign (US first) | Follower vs non-follower views visible |
What Meta Officially Says About Threads Ranking
Meta calls the system the "Instagram Threads Feed AI system". Its Transparency Center page (last updated 7 March 2025 when we read it) says the feed is "composed of content from people you follow and recommended content", and describes three steps:
- Gather inventory. The system collects "a portion of the public content available on Threads and all the content posted by accounts you follow", limited to posts that meet Meta's integrity rules.
- Leverage signals. It looks at input signals about each post and about you, such as how you have engaged with similar accounts, content and topics.
- Rank. Posts the system predicts will be more valuable to you appear higher.
Notice the asymmetry in step one. Every post from accounts you follow enters the candidate pool, but only "a portion" of everything else does. Following someone does not guarantee you see their posts near the top, but it does guarantee they are considered.
The predictions Threads makes
The same page lists the predictions the ranking models make for each viewer and each post. This is the closest thing to an official list of ranking factors any major platform publishes.
| Prediction (Meta's wording, abridged) | What it rewards | What you can do |
|---|---|---|
| How likely you are to like a post | Quick approval | Make the point land in the first line |
| How likely you are to create a reply | Posts people want to answer | Take a position or ask a specific question |
| How likely you are to click on a post | Curiosity about the full post or thread | Write openings that make people want the rest |
| How likely you are to click on a post and then like it | Posts that pay off after the click | Deliver on the opening, no bait |
| How likely you are to click on a post and then click on another post | Posts that lead to more browsing | Good replies under your post keep people reading |
| How likely you are to click the author's profile | Interest in you, not just the post | A clear bio and pinned post |
| How likely you are to click the author's profile and then another post | Profiles worth exploring | A consistent run of posts on your topic |
| How likely you are to scroll past rather than engage | Negative: posts people ignore | Cut vague announcements and filler |
Meta's summary also mentions predictions about time spent viewing a post and its permalink page. The page does not say how these predictions are combined, and it is explicit that the models and their inputs change often.
The input signals, and their time windows
The signal list is more technical, but it contains three practical details. First, signals are grouped around your past engagement on the Threads feed and on post permalink pages, including likes and time spent. Second, they include your engagement rates with specific authors, and the engagement of replies and replies to replies. Third, they are measured over a range of windows: 15 minutes, 1 hour, 6 hours, 1 day, 1 week, 28 days, a month and 84 days. One signal covers up to 100 posts a viewer liked in the last 15 minutes. Country and device platform are also inputs, as is the number of posts the viewer has created, including replies.
What the windows tell you: Threads reacts to what a person did in the last quarter of an hour and what they have done over the last twelve weeks. A post can catch someone's current mood, but it is the steady 84-day relationship between a viewer and an author that keeps you in their feed.
What users can do, and why it matters to you
The Transparency Center lists five user controls: unfollow, hide, mute, report, and switching to the Following view. Hiding a post also limits similar content. Each of these is a negative signal you want to avoid triggering. Add Dear Algo and Your Algo, and viewers now have more direct say over their Threads feed than on any other Meta app.
Breakdown 1: How Ranking Differs by Surface
There is no single Threads algorithm. There are several surfaces, and each one decides what to show differently.
For You
The default feed, and the only one that is ranked. Buffer's write-up of what the Threads team has said about ranking confirms that the AI ranking applies to For You, while Following stays chronological. For You mixes accounts you follow with recommendations. In November 2024 Adam Mosseri said Threads was "rebalancing ranking to prioritize content from people you follow" and warned creators they "should see unconnected reach go down and connected reach go up." Recommendations did not disappear; the balance shifted.
Following
Reverse chronological, followed accounts only, no recommendations. People reach it by tapping the Threads logo. Nothing is ranked here, so timing is the only lever: a post published while your followers are scrolling sits near the top of this feed. Our guide to the best time to post on Threads covers how to find those hours.
Communities and topic tags
Every post can carry one topic tag. The Threads API documentation confirms only one topic is applied per post, and tags are 1 to 50 characters. Meta's Communities announcement says topics with a dedicated community are marked on the tag itself, so a well-chosen tag links your post to that community, and Meta has said it is adding ranking inside communities. Community champions are chosen from people who are highly followed by other members, post regularly in the community and keep conversations going in replies, which is a useful description of what the community surface rewards.
Search
Meta says people can find communities by searching by name or by tapping a community tag on a post. Search is intent-driven, so specific wording ("sourdough starter smells like acetone") beats clever wording ("a little kitchen chemistry").
Viewer-steered feeds: Dear Algo and Your Algo
These are not separate feeds, but they reshape For You for days at a time. They reward accounts with a clear topical identity, because a request for "more about indie game development" can only pull in posts the system can confidently label.
| Surface | Ranked? | Main lever | Who it helps most |
|---|---|---|---|
| For You | Yes, AI ranking | Predicted likes, replies, clicks, time spent | Accounts that start conversations |
| Following | No, chronological | Posting while followers are online | Accounts with an engaged existing audience |
| Communities | Increasingly | Topic tag plus active replies in the community | Niche experts and fans |
| Search and tags | Relevance | Specific words and one accurate tag | Useful, evergreen posts |
| Dear Algo / Your Algo | Adjusts For You | Clear single-topic posts | Accounts with a defined niche |
Breakdown 2: How Ranking Differs by Stage of Distribution
Meta does not describe distribution stages for Threads the way LinkedIn does. But the time windows in the signal list, combined with how replies work, give a practical model of what happens after you post.
Stage 1: The first 15 to 60 minutes
Your followers are the candidate pool that is guaranteed, so this is where the first likes and replies come from. The short signal windows mean the system is reading early behaviour quickly. If your first viewers scroll past, there is little evidence for wider recommendation. This is the main reason to schedule posts for when your followers are active, rather than whenever you finish writing.
Stage 2: The reply window, roughly the first few hours
Replies, and replies to replies, are inputs in their own right. A post where the author answers each reply produces more of both, and gives later viewers a reason to click into the permalink and keep reading, which the "click a post, then another post" prediction rewards. Buffer's data scientist found that posts where the author replied earned about 42% more engagement on average, after controlling for whether the post received comments at all, and around two thirds of accounts saw a positive effect.
Stage 3: The first day
If engagement holds, recommendations widen beyond followers to people with similar interests. The Buffer write-up notes a Threads statement that ranking emphasises timeliness, engagement and topical relevance, surfacing active conversations. Threads is a "now" app, and a post that is still generating replies at hour six is treated as an active conversation.
Stage 4: The 28 to 84 day memory
Long after an individual post has faded, its effect remains in the relationship signals: how often a viewer has engaged with you over the past weeks. This is why consistency matters more on Threads than a single viral post. An account that earns a reply from the same reader every week builds an 84-day history that a one-off hit does not.
| Window | Signal behaviour it captures | Your job in this window |
|---|---|---|
| 15 minutes to 1 hour | What the viewer just liked and replied to | Publish when followers are online; be ready to reply |
| 6 hours to 1 day | Today's interests and conversations | Keep the reply thread active; quote good replies |
| 1 week | Recent topic and author affinity | Post on the same theme two or three times |
| 28 days to 84 days | Long-term relationship with authors | Show up consistently so regular readers keep engaging |
Breakdown 3: How Ranking Differs by Content Format
Threads is text-first, but the data says text alone is not the strongest format. Two Buffer datasets give slightly different pictures, which is itself useful.
Buffer's State of Social Media Engagement 2026, which analysed posts published through Buffer, with data through 3 December 2025, measured Threads engagement as likes, reposts, replies and quotes divided by reach. It found median engagement rates of 5.55% for video, 4.55% for images, 2.79% for text and 2.34% for link posts, while noting there is enough overlap that any format can do well. Buffer's earlier Threads statistics roundup found images earning 60% more median engagement than text posts and 37% more than link posts, with images and video almost level.
| Format | Median engagement rate (Buffer 2026) | Best use | Watch out for |
|---|---|---|---|
| Video | 5.55% | Short demos, reactions, behind the scenes | Threads video is capped at 5 minutes via the API |
| Image | 4.55% | Screenshots, charts, photos with a strong caption | The text still has to carry the idea |
| Text | 2.79% | Opinions, questions, quick observations | Vague posts get scrolled past |
| Link | 2.34% | Articles and resources, with context | Few people click; give the takeaway in the post |
| Carousel | Not reported separately | Step-by-step, before and after, mini-guides | 2 to 20 items per carousel |
Text posts
The main post is limited to 500 characters, confirmed in Meta's API docs. Text has the lowest median in Buffer's 2026 data, but it is also the format with the widest spread: a sharp opinion can outperform any photo. Treat text as your conversation starter, not your reach format. Our Threads character counter helps keep posts tight.
Text attachments
The 10,000-character attachment shows as a block under your main post and expands when tapped. A tap is a click, and "click on a post and then like it" is one of the published predictions, so a strong 500-character hook with a genuinely useful attachment lines up well with the system. Nobody has published engagement data on attachments yet, so test it.
Images and carousels
Carousels can hold 2 to 20 images or videos. They encourage the swipe-and-linger behaviour that time-spent predictions pick up. If you already make carousels for Instagram, our carousel best practices apply here with lighter design.
Video
Video tops Buffer's 2026 format table, but Threads is not a video-first feed. Short, captioned clips that make sense without sound work best. Remember that Threads' head, Connor Hayes, told TechCrunch that users "appreciate the quietness of the app" in a world of loud, video-centric social. Use video when it adds something words cannot.
Links
In November 2024 Mosseri said Threads placed little value on link clicks because "people don't like and comment on links much". By mid 2025 he said, as Social Media Today reported, "We've been working on making sure links are ranked properly. Links have been working much better for more than a month now." Link posts still have the lowest median engagement in Buffer's data, because people rarely reply to a bare link. And since 22 December 2025, posts published through the Threads API fail if they contain more than five unique links.
Breakdown 4: How It Plays Out by Account Type and Size
New accounts with no audience
Your followers are the only guaranteed inventory, and you have few. The fastest route in is replies: thoughtful replies on larger accounts' posts are themselves posts, they are ranked, and they put your name in front of an established audience. Buffer quotes Mosseri telling Platformer that "the sum of all your replies is about as valuable as the sum of all the value of all your posts." Our guide on how to get more Threads followers turns this into a daily routine.
Creators with an Instagram audience
Threads profiles are tied to Instagram, so an existing Instagram following converts into Threads followers easily. Those followers are guaranteed inventory, and the November 2024 rebalance favours them. Promote your Threads account from Instagram, then give those followers a reason to reply. If you also post Reels, our explainer on how the Instagram algorithm works shows where the two systems overlap.
B2C brands
Brands are already on Threads in numbers. Sprout Social's Q1 2025 Marketer Pulse Survey found 57% of brands already post on Threads, rising to 64% among B2C brands. Personality and replies matter more than polish. A coffee roaster that answers "which beans for a moka pot?" in replies earns more distribution than one posting product shots with a discount code.
B2B brands
The same Sprout survey found only 19% of B2B brands posting on Threads, with 31% planning to start. That gap is an opening. Founder and employee voices tend to work better than a company logo, and short, opinionated takes on your industry fit the format. LinkedIn depth can be trimmed into a 500-character take with the long version in a text attachment.
Publishers and newsletter writers
Link posts are ranked fairly now but still earn less engagement. Put the key insight in the post, the link second, and use replies to continue the discussion. Threads' live chats, which Meta has been expanding to more communities, reward the same thing: being present in the conversation, not just dropping links.
Political and civic accounts
Meta's official political content policy says it won't "proactively recommend content about politics on recommendation surfaces across Instagram and Threads", defining it as content potentially related to laws, elections or social topics. Engadget's report notes people can opt in to political recommendations, and followers still see these posts. If growth through recommendations is the goal, expect political posts to travel mostly through your followers and reposts.
| Account type | Biggest advantage | Biggest risk | First move |
|---|---|---|---|
| New account | Replies are ranked like posts | No guaranteed inventory | 10 thoughtful replies a day in your niche |
| Instagram creator | Followers carry over | Treating Threads as a caption dump | Cross-promote, then post conversation starters |
| B2C brand | Brands are accepted on Threads | Sounding like an ad | Answer customer questions in public |
| B2B brand | Few competitors posting | Corporate tone | Founder-led opinions with one topic tag |
| Publisher | Links ranked properly since 2025 | Low engagement on bare links | Insight first, link second |
| Political account | Loyal followers | Not recommended by default | Build follows; rely on reposts |
Breakdown 5: What Meta Says Versus What the Data Shows
Official statements tell you what the system looks at. Independent data tells you what tends to win. They usually agree, but not always.
| Topic | What Meta says | What independent data shows |
|---|---|---|
| Replies | "Create a reply" is a ranking prediction; replies are valued like posts | Author replies linked to about 42% higher engagement (Buffer, 128,000+ posts) |
| Links | Links "ranked properly" since 2025 | Link posts have the lowest median engagement at 2.34% (Buffer) |
| Followed accounts | Ranking rebalanced towards followed accounts in late 2024 | Gap between top performers and median accounts is among the widest (Buffer) |
| Formats | No format preference stated | Video and images beat text on median engagement (Buffer) |
| Engagement trend | Ranking improved time spent by 10% in Q3 2025 | Median engagement rate fell about 18% in 2025 as the audience grew (Buffer) |
| Typical post | Not disclosed | About 1,536 impressions and nearly 25 interactions per post (Metricool) |
The Metricool figure comes from its 2026 Social Media Study of 1,059,949 accounts and almost 39.8 million posts published during 2025. It is an average, so a few viral posts pull it up, but it is a useful sense check: on Threads, a post that reaches a few thousand people is doing normal, healthy work.
The falling engagement rate deserves context. Buffer puts Threads' 2025 median at about 3.6%, below LinkedIn (about 6.2%), Facebook (about 5.6%), Instagram (about 5.5%), TikTok (about 4.6%) and Pinterest (about 4.0%), and above X (about 2.5%). A growing platform adds more posts and more casual readers, so rates fall even when total engagement rises. Compare your own account with itself over time, not with another network. Our engagement rate glossary entry explains the common formulas and why they give different answers.
A Step-by-Step Method to Learn Your Own Threads Algorithm
No outside guide can tell you how the system weighs your audience. This method gets you a reliable answer in six weeks.
- Pick one or two topics. Write down the two subjects you want Threads to associate with you. Every post in the test should be about one of them, with one matching topic tag.
- Record a baseline. From Threads Insights, note your last 30 days of views, the follower versus non-follower split (where available), replies and follows. That is your "typical".
- Plan a six-week calendar. Aim for at least one post a day on weekdays. Split them across formats: text questions, image posts, short videos, one link post a week and one text attachment a week. Keep the topics constant so only the format changes.
- Schedule for your audience's hours. Buffer's 2026 data found weekday mornings, roughly 6 to 11 a.m., strongest with a Thursday 9 a.m. peak. Use that as a starting point, then move to the hours your own Insights show.
- Block reply time. Put 20 minutes on your calendar right after each scheduled post. Reply to every genuine reply in the first hour, and add a follow-up thought to keep it going.
- Reply outward daily. Leave five to ten substantive replies on other people's posts in your topic each day. Track which ones bring profile visits.
- Log results weekly. For each post, record views, non-follower share, replies per 1,000 views and follows. Replies per 1,000 views is your conversation score.
- Compare formats and times. After three weeks, drop the weakest format and the weakest time slot. Double the strongest.
- Repurpose winners. Turn a top text post into an image or a carousel and repost the idea a week later. If the theme wins twice, it is a pillar.
- Review at six weeks. Compare against your baseline. If non-follower views rose, the algorithm is recommending you. If follower views rose but non-follower views did not, your content resonates with your audience but is not yet spreading. Work on hooks and topic clarity.
Worked Examples
A specialty coffee roaster
Northside Roasters posts five times a week. Before: product photos with "New single origin, link in bio." After: a Monday question ("Moka pot owners: what grind setting have you settled on?"), a Wednesday image of three roast levels side by side, and a Friday 20-second video of a pour. The owner replies to every answer. The question posts produce the most replies, the video the most views, and the image the most profile visits. The roaster keeps all three and drops the plain promo posts.
A B2B accounting software company
The company account gets little traction, so the founder posts instead. Each week she shares one opinion ("Most small businesses should close their books monthly, not quarterly. Here is why.") with a text attachment explaining the process, tagged with a single accounting topic. She spends 15 minutes each morning replying to small business owners' posts about bookkeeping. Profile visits from non-followers become the metric that tells her it is working.
An independent bookshop
The shop joins the books community, uses one book-related topic tag per post, and runs a weekly "what are you reading this weekend?" thread. Staff reply with recommendations. Because communities reward people who keep conversations going, the shop's account becomes familiar to members who later visit the profile.
Copy-Ready Post Templates for Threads
| Template | Example | Prediction it targets |
|---|---|---|
| Specific question | "Freelancers: what is the one clause you now add to every contract?" | Create a reply |
| Position plus reason | "Unpopular opinion: weekly team meetings should be 15 minutes. Anything longer is a status report that belongs in writing." | Reply, like |
| Before and after | Two images: old checkout, new checkout. "We cut five steps to two. Conversion went up. Here is what we removed." | Click, time spent |
| Hook plus attachment | "I tracked every hour for a month. Three things surprised me." plus a 1,500-word attachment | Click, then like |
| Community prompt | "Book Threads, what is your comfort reread?" | Reply within a community |
| Quote a good reply | Quote a follower's answer and expand on it | Reply, profile click |
How to plan Threads posts around your reply time
The Threads algorithm rewards two things a busy team struggles to do at once: posting consistently at the right hours, and being present in the replies right afterwards. Planning solves the first so your time goes to the second.
A simple rhythm that works: batch-write on Monday, plan five posts for weekday mornings, and put a 20-minute reply block on your calendar after each one. Our guide on staying consistent on social media covers the batching routine in detail.
Schedule to Every Platform From One Calendar
Plan a week of posts for LinkedIn, X, TikTok, YouTube, Bluesky and Pinterest in one sitting, let Postprism publish them at your best times, and spend your time on the replies.
Common Threads Algorithm Mistakes (and Fixes)
- Posting and leaving. Replies and replies to replies are ranking inputs. Fix: schedule the post, then block 20 minutes to answer every genuine reply.
- Vague announcements. "Excited to share our new website!" invites a scroll, and scrolling past is a negative prediction. Fix: lead with a specific result or a position.
- Engagement bait. "Reply YES if you agree" produces low-value replies and annoys people into hiding you. Fix: ask questions with real, varied answers.
- Mixing too many topics. Dear Algo, Your Algo and communities all match posts to topics. Fix: one idea and one topic tag per post.
- Ignoring Following. The chronological feed rewards posting when followers are online. Fix: schedule for your audience's hours, not your writing hours.
- Bare link posts. Link posts have the lowest median engagement. Fix: give the takeaway in the post and the link as a bonus.
- Copying captions from Instagram or LinkedIn. Hashtag blocks and "link in bio" read as foreign on Threads. Fix: rewrite each post for Threads; our guide to adapting content for each platform shows how.
- Only posting, never replying elsewhere. Mosseri says replies are worth about as much as posts in total. Fix: five to ten substantive replies a day in your niche.
- Judging by likes. Likes are one of eight predictions. Fix: track replies per 1,000 views and non-follower views.
- Bursty posting. Ten posts one week and none the next breaks the 28 to 84 day relationship signals. Fix: a steady calendar, even if it is three posts a week.
- Comparing your rate with other networks. Threads' median engagement rate is lower than LinkedIn's or Instagram's by design of the audience. Fix: benchmark against your own history.
- Political tangents on a growth account. Political content is not recommended by default. Fix: keep growth posts on your core topics and know what to expect when you do post about politics.
How Threads Fits Your Wider Plan
Threads rarely stands alone. A useful split: Instagram for visuals, LinkedIn for professional depth, X for live news, Threads for relaxed conversation and personality. If you are weighing Threads against X, our guide to the X algorithm shows how differently the two rank replies and links. Our Threads marketing playbook covers content pillars and goals, and the overview of how social media algorithms work compares all the major networks. Sprout's survey found 52% of marketers repurpose content for Threads from other platforms and only 26% create unique content, which suggests the accounts that write natively for Threads are still in the minority.
Frequently Asked Questions
Does Threads have an algorithm?
Yes. Meta describes it as the Instagram Threads Feed AI system. It ranks the For You feed by predicting how likely you are to like, reply to, click on or scroll past each post. The Following feed is not ranked; it shows followed accounts in reverse chronological order.
What is the most important signal on Threads?
Meta does not publish weights, but replies stand out. "Create a reply" is one of the published predictions, reply engagement is an input signal, and Buffer found author replies linked to about 42% higher engagement, the largest lift across six networks it studied.
Does follower count matter on Threads?
It matters for the guaranteed part of your audience: all posts from accounts someone follows enter their candidate pool, and since late 2024 ranking favours followed accounts. But recommendations still reach non-followers, and the long-term relationship with each viewer matters more than raw follower count.
Are links penalised on Threads?
Meta says no. Adam Mosseri said in 2025 that links were now being "ranked properly" and working much better. Link posts still earn the lowest median engagement in Buffer's data, so lead with the insight and add the link second.
How many topic tags should I use?
One. Threads applies only one topic tag per post. Choose the most specific tag that matches the conversation, ideally one that has a community.
What is Dear Algo?
A feature launched in February 2026 that lets people post "Dear Algo" with a request to see more or less of something, adjusting their feed for three days. Your Algo, added in June 2026, does the same privately for one, three or seven days.
How often should I post on Threads?
Consistency beats volume. Daily on weekdays is a strong target for most accounts; three posts a week is a fine minimum if you reply well. The API allows up to 250 posts a day, but that is a technical limit, not a strategy. Our guide on how often to post on social media compares networks.
Can I schedule Threads posts?
Yes. Threads supports publishing through Meta's official API, which schedulers use to publish text, image, video and carousel posts at set times. Scheduling keeps you consistent and frees you to be in the replies after each post goes live.
