Threads is now big enough that measuring what works on it is a real job, and in September 2026 the tools for that job changed shape. Meta rebuilt the in-app Insights panel around AI-written summaries, added a follower-versus-non-follower reach split, started scoring every post against your own recent baseline — and quietly removed link clicks. This guide is the practitioner's read: what the redesign actually measures, which numbers a creator should steer by (discovery reach, profile visits, replies and reposts, communities) versus the ones that just flatter, where to find the panel and who can see demographics, how to cover the link-click blind spot the update introduced, and the honest limit every analytics guide skips — that a dashboard scored against your baseline only means something once you post enough to have a baseline worth beating.
Threads crossed into "big enough that measuring it is a real job" territory in 2026 — Meta reported the platform past 500 million monthly active users in mid-2026 — and in September the analytics for that job were rebuilt. On September 8, 2026, Threads redesigned its in-app Insights panel so that Meta AI interprets the numbers for you instead of just listing them, added a split of each post's reach into followers versus people who do not follow you, and started scoring every metric as higher, lower, or typical against your own recent posts. In the same update it removed link clicks. The full change is covered in the report on Threads rebuilding its profile analytics around AI summaries; this guide is the practitioner's version — what to actually steer by.
Two things trip creators up with the new panel. First, richer analytics tempt you to over-read a single post, when the whole design is built to be read across posts against a baseline. Second, the AI summary feels authoritative, so it is easy to treat a machine-written "this topic is working" as a verdict rather than a hypothesis you still have to verify. This guide maps what the redesign measures, separates the metrics that change decisions from the ones that only feel good, covers the link-click blind spot the update introduced, and ends on the honest boundary between measuring your content and producing enough of it for the measurement to mean anything.
The redesign is not a coat of paint. It changes what you see, how it is framed, and — in one case — what you no longer get at all.
The headline change is that Meta AI now writes a plain-language summary at the top of both individual posts and your account as a whole. Instead of a table of counts, you get a sentence: a post might be described as getting several times more views than your recent posts, or a month flagged as "one of your quieter months" with views down a stated percentage. Threads' framing is that this helps you see how each post did and what drove the result. The right way to use it: treat the summary as a lead that points you at what to investigate, then confirm the claim against the underlying numbers in the same view. It lowers the skill floor for newer creators, but a summary is a fast triage layer, not ground truth.
The single most useful number the rebuild exposes is the split between reach from your existing followers and reach from people who do not follow you (in-stream discovery). This is the difference between engagement and growth. A high non-follower share means the algorithm is surfacing you to strangers — the post has viral shape and the topic can grow your account. A high follower share means you are mostly reaching the people you already have, which is fine for retention but will not expand your audience on its own. Almost every "what should I make more of" decision on Threads now starts here.
Each per-post metric — views, viewers, likes, reposts — is now labeled relative to your other recent posts rather than shown as a bare absolute number. This is a genuine improvement, because it rewards consistency over one-off spikes and makes the numbers legible without you having to remember what "normal" looks like. It has one prerequisite, though: the baseline only means something if you post regularly enough to have one. A creator who publishes twice a month gives the system almost nothing to compare against, so the labels get noisy. Baseline-scored analytics quietly assume a cadence.
The notable removal is link-click data, which the panel previously surfaced and no longer does; Meta said only that it is "exploring" the metric. For any creator who routes traffic off Threads — to a site, a shop, a newsletter — this is a real blind spot introduced in the same update that enriched everything else. The rest of the picture got better while the one number that maps directly to off-platform business disappeared. That does not mean you stop driving clicks; it means you have to measure them yourself, which the link-click section below covers.
The rebuilt panel shows a lot. Here is the short list a creator should actually steer by, in rough order of how directly each one connects to growth.
Covered above, and it sits at the top of the list on purpose. When you review a week of posts, the first sort is by non-follower reach share — that is where the algorithm is telling you which topics and formats it is willing to push into discovery. A post that went wide with strangers is a candidate to make more of; a post that only reached your existing followers is telling you the idea resonates with your base but is not currently a growth lever.
Reach that does not convert is a dead end. The metrics that compound are profile visits and new follows, because they turn a moment of attention into a lasting audience. The most useful read on Threads is a two-step one: which posts drove the most non-follower reach, and of those, which converted that reach into profile visits and follows. A post that reaches many strangers but sends nobody to your profile went viral without growing you — a common and easily-missed failure mode.
Threads is a conversation-shaped platform, and replies and reposts are what keep a post circulating in the feed. A view spike with few replies rarely compounds; a post with strong reply and repost activity keeps getting resurfaced. Weight interaction over impression when you decide what worked. This is also where the platform differs from a pure video feed — a text post that starts an argument-in-good-faith can outperform a slicker one that nobody responds to.
The panel now surfaces which in-app communities and topics drove a post's engagement. Treat this as a targeting hint, not just a description: if a specific community carried a post, that is a signal about which framings and subjects to lean into next, and where your account has traction you might not have noticed. It is one of the more actionable additions, because it tells you not only that something worked but with whom.
Audience demographics (age, gender, top locations) and peak posting times and days are the slow-moving inputs that set your schedule and your references. Check them roughly monthly rather than weekly — they do not move fast — and use them concretely: publish against your audience's main time zones, and match examples and language to who is actually reading rather than who you assumed. One gate to know: personal accounts unlock demographics at 100 followers, while business accounts see them immediately.
Raw view count and total follower count feel good and steer almost nothing. The test is the one from the Facebook analytics guide: for any number you are about to celebrate, ask what you would do differently if it were half or double. If the answer is "nothing," it is vanity. Views pass that test rarely; non-follower reach, profile visits, and reply activity pass it almost always.
Insights lives at the top-left of your profile in the mobile app (the insights icon) or at threads.com/insights on the web, and it is open to every account type — not just creator or business profiles. The only thing gated behind account type is demographics: a personal account needs at least 100 followers before age, gender, and location appear, whereas a business account sees them from the start. Everything else — views, interactions, follower growth, the reach split, the AI summaries — is available to anyone. If you are new to the panel itself, the step-by-step how to use Threads Insights walks the read-and-act loop; this guide is the strategy behind it.
Because native link clicks are gone, any creator who cares about off-platform traffic has to rebuild that measurement outside Threads. The practical fix is to own the destination and tag the path. Put a tagged link (UTM parameters, or a short-link service that reports clicks) on the links you share from Threads, and read the click and conversion data from your destination's own analytics — your site, your email platform, your store — rather than from Threads. That gives you back the number the panel dropped, and it is arguably more trustworthy because it counts an actual arrival, not a tap.
There is a strategic version of this too. The safest response to a platform removing the metric that measured your off-platform goal is to route more of that goal through surfaces you control — an email list and a blog you own outright, where the analytics can never be taken away in an update. That is a distribution decision as much as a measurement one, and it is the thread that connects Threads analytics to the wider point below: the platforms give and take away the numbers, but owned destinations do not.
Three habits separate analytics that change your output from a dashboard you glance at. First, know your baseline before you read anything — record your typical per-post views, interactions, and follower count so the "higher/lower/typical" labels mean something against your reality, not a vibe. Second, never rebuild a strategy off one post; the panel is designed to be read across posts, so a pattern only counts once it repeats over several. Third, treat every AI summary as a hypothesis to verify, not a verdict — a machine-written "this topic is working" is a prompt to check the underlying numbers across your recent posts, and often the claim holds for one post and not the trend.
The discipline is the same one behind any honest measurement: a winner is only meaningful once you have enough at-bats to compare. That is why cadence is not separate from analytics — it is the precondition for it. A panel that scores you against your own baseline is only as trustworthy as the number of posts feeding that baseline.
Here is the limit no analytics guide likes to admit. The rebuilt Insights is a rear-view mirror: it tells you, now in plain language, what already worked. It does nothing about the actual bottleneck for most creators, which is producing enough good, on-brand posts to have a meaningful sample — and then producing more of the winners fast enough to matter before the next weekly review. A panel that proves a specific topic pulls non-follower reach is worthless if you can only make one post about it a month. Worse, the new baseline scoring makes low volume actively misleading: with too few posts, the "typical" the AI compares against is barely defined, so the labels wobble. Analytics rewards consistency, and consistency is a production problem.
Closing that second gap is what Kompozy is for. It is a content generation and multi-platform publishing engine, not an analytics tool — and that is the point, because the throughput it gives you is what makes baseline-scored analytics trustworthy in the first place. When Insights points at a topic that is pulling discovery reach, Kompozy is how you turn that one signal into the volume of at-bats the panel needs: it generates across eighteen formats — talking-head and avatar video like Persona Shorts, clipped shorts, image posts, brand-exact carousels, quote graphics, text posts, blogs, and newsletters — all held to one voice by a Persona Brief and banned-word filters, so every piece is consistent enough that the analytics differences you read afterward are real signal about the idea, not noise from a wobbling execution. Then it publishes and schedules to Threads alongside the other seven social platforms plus blog and email, and can run on autopilot with a per-post review gate so the queue stays full without going unattended.
The clean way to hold it: Threads Insights is your measurement layer and you should lean on it exactly as this guide describes. Kompozy is the production layer that gives that measurement enough to measure — and, on the link-click gap specifically, it is what lets you actually build out the owned email list and blog that put your off-platform numbers back under your control. Analytics tells you which door is worth walking through; an engine is what walks you through it, at the cadence a baseline-scored panel quietly demands. If you want the step-by-step tracking workflow instead of the strategy, see how to track Threads creator analytics across platforms.
Threads creator analytics in 2026 is richer and, in one way, poorer than it was. The September 8 rebuild gives you AI-written reads, a follower-versus-non-follower reach split that is the best growth signal Threads has ever exposed, and per-post scoring against your own baseline — and it takes away native link clicks. Steer by the metrics that compound: non-follower reach, profile visits and follows, replies and reposts, and the communities that carry you. Cover the link-click gap with your own tagged links and owned destinations. Verify the AI's claims across several posts, not one. And remember the boundary the panel will not tell you: it can only measure the posts you actually make, so the volume and consistency of your output is what makes every number on the page mean something.
It is the in-app Insights panel Threads rebuilt on September 8, 2026, plus whatever tracking you add around it. The redesign puts Meta AI-written performance summaries at the top of individual posts and your account, splits each post's reach into followers versus non-followers (discovery), labels every metric higher, lower, or typical against your recent posts, and shows which in-app communities engaged, demographics, and peak times. The main removal was link clicks, so off-platform traffic is no longer measured natively.
The ones that change what you make next. Non-follower reach share tells you whether the algorithm is carrying a post into discovery — the growth signal. Profile visits and new follows tell you whether that reach compounded into audience. Replies and reposts tell you whether a post kept circulating, which matters more on Threads than raw views. Communities and topics point at what to make more of. Views and follower count on their own are close to vanity.
Yes. The September 2026 Insights redesign no longer reports the link-click metric Threads previously showed, and Meta said only that it is "exploring" the feature. If your Threads goal is traffic to a site, store, or newsletter, the native panel no longer measures your actual objective, so you have to track clicks yourself with tagged links and your destination's own analytics.
Tap the insights icon at the top-left of your profile in the mobile app, or open threads.com/insights on the web. Insights is available to every account type, not just creator or business profiles. The one gate is audience demographics: personal accounts unlock age, gender, and location at 100 followers, while business accounts see them from the start.
The old panel listed raw counts you interpreted yourself. The 2026 version hands you a first read: Meta AI translates the numbers into plain language, scores each post against your baseline instead of an absolute number, and foregrounds the discovery split. That lowers the skill floor for newer creators, but it also means you now verify an AI claim across several posts rather than trusting a single summary — and you have lost the native link-click number in the trade.
Threads creator analytics in 2026 lives in the in-app Insights panel Meta rebuilt on September 8, 2026: AI-written summaries for each post and your account, a follower-versus-non-follower reach split, per-post metrics labeled higher, lower, or typical against your recent baseline, plus communities, demographics, and peak times. Link clicks were removed. The metrics worth steering by are non-follower reach, profile visits and follows, and replies and reposts — the numbers that compound into audience rather than just flatter.
Get started → · ← All guides · Compare Kompozy vs other tools