How to use the new Threads Insights to improve your posts (2026 analytics update)
Threads rebuilt Insights in 2026 with AI summaries and follower-vs-non-follower reach. Learn to read the new profile analytics and turn it into better posts.
To use the new Threads Insights, announced September 8, 2026, record your baseline first, then read each post's written summary as a lead to check against the numbers. Use the split between follower and non-follower views to see whether discovery is working, rank posts by profile visits and follows, check demographics once you pass 100 followers, and log your numbers weekly.
Key takeaways
The September 2026 redesign rolled out first to creators in the U.S.
Each post gets a written summary and a higher, typical or lower call against your baseline.
Views now split into followers and non-followers. That split shows whether discovery works.
Demographics (age, gender, city, country) need at least 100 followers.
Copy your main numbers into your own sheet weekly. Native history is short.
On September 8, 2026 Threads rebuilt its in-app Insights, and the change is more than a new coat of paint. Threads now writes a plain-language summary of how each post performed and tells you whether its numbers are higher than, typical of, or lower than your own baseline. The redesign rolled out first to creators in the U.S.
The headline new metric is a reach split — how much of your audience is followers versus people who don't follow you yet — which is the single most useful number Threads has ever exposed for deciding what to make next.
This guide is the practical read-and-act loop, not a metric glossary. You'll open the panel, set a baseline so the AI's "above/below" call means something, use the follower-vs-non-follower split to tell discovery from preaching-to-the-choir, find the topics and formats that actually earn profile visits and follows, unlock demographics, and turn one week's readout into next week's plan. Treat the AI summary as a hypothesis to verify against the raw numbers, not a verdict — that habit is the difference between analytics that change your output and a dashboard you just glance at.
How do you use Threads Insights to improve your posts?
Open Insights and write down your baseline. Open Insights from your Threads profile. Before you read anything, record your current follower count and your typical per-post views and interactions from the last couple of weeks. The rebuilt panel judges each post against your recent baseline, so the AI's "above/below" call is only meaningful if you know what your normal actually is.
Read the AI summary as a lead, not a verdict. The new written summary tells you in plain language whether a post beat or missed your baseline. Use it to point you at what to investigate, then confirm the claim against the underlying numbers in the same view. AI-written performance notes are a fast triage layer, not ground truth — a summary that says "this topic is working" is a hypothesis you verify across several posts, not one.
Split your reach into followers vs non-followers. This is the marquee 2026 metric. A high non-follower share means the algorithm is pushing your post into the feeds of people who don't follow you — discovery is working and the post has viral shape. A high follower share means you're mostly reaching people you already have. Neither is inherently good: growth needs non-follower reach that then converts, which is why the next step matters.
Rank posts by profile visits and follows, not just views. Views are the vanity number; profile visits and follower growth are the ones that compound. Pull your recent posts and sort them by the profile visits and new follows each drove, then look at what the top few share — a topic, a hook style, a format (text vs. image vs. clip). That shared pattern, not the single most-viewed post, is your signal for what to make more of.
Unlock and use audience demographics. Insights shows age, gender, and location (cities and countries) for your followers once your profile has at least 100 followers. Use it concretely: post timing against your audience's main time zones, and reference points, examples, and language that match who's actually reading — not who you assumed.
Turn the readout into a two-week plan. Close the loop or the analytics were wasted. Pick the two or three patterns that repeat across your top posts — a winning topic, a format, an audience the discovery split favors — and commit next week's posts to doubling down on them. Cut or rework the formats that sit consistently below baseline. Then check Insights again in a week and see whether the numbers moved.
Log your numbers weekly for real trends. Native Insights shows a rolling window, so long-run trends and any metric Threads later trims aren't guaranteed to stick around. Once a week, copy your key numbers — follower count, median views, the non-follower reach share, top post — into your own sheet. That log is what turns a snapshot into a growth curve you can actually reason about, and it survives any future panel redesign.
What mistakes do people make reading Threads Insights?
Don't rebuild your strategy off one post. A single post plus the AI summary is noise — the rebuilt panel is designed to be read across posts, so verify any "this works" claim over several before acting on it.
High non-follower reach isn't automatically a win. If a post reaches many non-followers but drives no profile visits or follows, it went wide without converting — reach that doesn't compound into audience is a dead end.
Demographics are gated. A profile needs at least 100 followers before age, gender, and location data unlock.
Insights is a rolling window, not an archive. Native history is short and panels get redesigned — if you want month-over-month trends, you have to log the numbers yourself.
Views are not engagement. A view spike with few replies or reposts rarely compounds on Threads, where replies and reposts are what keep a post circulating — weight the interaction metrics, not just the impression count.
The panel only covers your own profile. Threads' native Insights shows your account, not competitors' — third-party tools estimate other profiles' growth, but treat those as estimates, not Threads' own numbers.
How do you act on what Threads Insights tells you?
Insights is now good enough to tell you exactly what to make next — a topic that spikes non-follower reach, a format that earns profile visits, an audience the discovery split keeps favoring. The gap it can't close is the one between that finding and shipping enough of it, on-brand, to move the numbers again by the next weekly review. That production gap is what Kompozy fills. It's a full AI content generation and multi-platform publishing engine, not an analytics tool — so the loop runs analyze in Threads, produce and publish in Kompozy, then re-read Insights. Take the winning pattern the reach split surfaced and feed it in: Kompozy generates the on-brand set — a Persona Short for the topic that's pulling non-follower reach, Text Posts and Quote Graphics restating the line that landed, a Carousel on your exact brand template — all governed by one Persona Brief so voice and pillars stay consistent across every piece.
The concrete win is turning a single insight into a fanned-out, scheduled slate instead of one manual post. Autopilot publishes the set to Threads through its native surface and, in the same pass, to the other seven social platforms plus blog and email — so a topic that proved itself on Threads gets tested everywhere at once, and a per-post review gate lets you approve or tweak each piece before it ships. Be clear on the boundary: Kompozy doesn't read your Threads Insights or auto-optimize from them — you still make the read-and-act call each week (that's steps five and six of this guide). What it removes is the throughput ceiling between the decision and the next data point, so you can actually act on what Insights tells you fast enough to compound. Kompozy's paid plans (see pricing) start with Starter, which fits a solo creator running the weekly loop on one or two accounts; Pro ($499/mo for 18,000 credits) fits a brand or agency turning analytics into a multi-platform slate every week; Enterprise is available through Contact Sales.
What does this look like in Kompozy?
A real persona tweet graphic made in the Kompozy workspace, shown as an example of the format. It was not made for this guide or for any customer.
Real output · Kompozy Persona tweet format · from the Kompozy workspace
Who is this for, and who should skip it?
Best for:
Threads creators who post several times a week and want a weekly review habit.
Accounts trying to grow through non-follower reach.
Brands that test a topic on Threads before taking it to other platforms.
Not ideal for:
Anyone tracking competitors. Native Insights covers only your own profile.
Accounts under 100 followers that need demographic data now.
Anyone expecting Kompozy to read Insights for them. It does not.
Frequently asked questions
Where do I find Threads analytics?
Open Insights from your Threads profile. Audience demographics need at least 100 followers before they appear.
What changed in the September 2026 Threads analytics update?
Announced September 8, 2026 and rolling out first to creators in the U.S., the rebuilt Insights adds a written summary of how each post performed, a readout of whether its numbers were higher than, typical of, or lower than your own baseline, a split of views between followers and non-followers, and a views-over-time chart.
Do I need a creator or business account to see Threads Insights?
We could not find Threads tying Insights to a special account type. The documented gate is demographics: a profile needs at least 100 followers before age, gender, and location data appears. The September 2026 redesign rolled out first to creators in the U.S.
What does the follower-vs-non-follower reach metric tell me?
It shows how much of a post's reach came from your existing followers versus people who don't follow you. A high non-follower share means Threads is surfacing the post in discovery to new audiences — a signal to make more of that topic or format. A high follower share means you're mostly reaching people you already have, which is fine for retention but won't grow the account on its own.
How often should I check Threads Insights?
Weekly is the right cadence for most creators. Checking after every post tempts you to overreact to noise; a weekly review lets you read patterns across several posts, verify the AI summary's claims, and commit to a plan for the coming week. Log your key numbers each time so you can see real trends rather than a single rolling snapshot.
How do we know this?
On 2026-10-23 we checked the update date and features against Net Influencer's report and the 100-follower demographics threshold against Meta's Threads API documentation. We removed claims about AI-written summaries naming topics and communities, the menu location and account-type rules, which we could not verify. Kompozy plan details come from kompozy.io/pricing.