Instagram Insights Guide: how to read your analytics and act on them

Start with one number: your engagement rate

If you only ever look at one figure, look at this one. Engagement rate is the share of people who did something — liked, saved, commented, shared — rather than merely scrolled past. It is the closest thing Instagram gives you to a measure of whether anyone cares.

The trap is that engagement rate is meaningless without context, and almost every article you will read quotes a single universal target. That is wrong, and it is wrong in a predictable direction: engagement rate falls as you grow. Not because your content got worse, but because Instagram starts showing you to strangers, and strangers engage less than the people who chose to follow you. A 100,000-follower account holding 3% is doing better than a 2,000-follower account holding 5%.

So your report grades you against a benchmark for your tier. These are the actual thresholds it uses:

Your account health score, and exactly how it's calculated

A single 0–100 number sits at the top of your report. Scores like this are usually marketing theatre, so here is the entire calculation. It is the average of up to four components, each scored out of 100 and each capped at 100:

Two design decisions worth knowing about. First, components we cannot measure are dropped from the average, not scored as zero. A week-old account with no growth history is graded on the components it does have, rather than being told it is failing at something it has not had time to do. Second, every component is capped. A viral month that produces a 40% growth rate does not paper over a dead engagement rate.

Above 70 reads “Healthy & growing.” Between 40 and 69, “Doing okay.” Below 40, “Needs attention.” If you are below 40 and cannot tell why, look at which component is dragging: the score is an average, so one component near zero costs you roughly 25 points on its own.

The leaderboard: find your best posts in five seconds

Instagram makes you scroll your own grid and remember numbers. The leaderboard ranks every post you have published, on the metric you choose, over the window you choose. Top three go on a podium — second, first, third, with first raised and highlighted, the way a podium actually looks — and ranks four through ten sit underneath.

The four metrics, and what each is really asking

• Most viewed. Total plays and displays, repeats included. Where Instagram reports no view count — which it often doesn't outside Reels — we fall back to reach, so the column is never mysteriously empty.
• Reach. Unique accounts. The honest measure of distribution. Use this one when comparing a Reel against a carousel, because views flatter video.
• Engagement. Total interactions where Instagram gives us the figure, and otherwise likes + comments + saves + shares added up ourselves.
• Followers. Accounts that followed directly from that post. Expect this tab to look empty — Instagram reports a follow count on very few posts and returns zero for the rest. That is Instagram's data, not a bug in the leaderboard, and we would rather show you an honest empty state than invent numbers.

Windows

Seven, fourteen, thirty days, or all time. Posts are filtered by age and by having a non-zero value on the chosen metric, then sorted. A post published yesterday will not appear in a leaderboard sorted by follows if Instagram has not attributed any follows to it yet — which, again, is most posts.

What moved the needle: viral posts and spike days

“Viral” is a word that means nothing unless it is measured against something. We measure it against you. Every post carries a reach multiple — how many times your normal reach it achieved. A post that comfortably beats your baseline is tagged overperformed. One that runs far past it is tagged viral, and the badge shows the multiple: 11.7× VIRAL.

This is why a post with 900 reach can genuinely be viral. If your typical post reaches 80 people, 900 is an eleven-fold event, and it is telling you something specific about what worked. An account with a million followers gets no credit for the same 900.

Alongside the individual posts, the report finds spike days — days where your account-wide reach jumped well above its own trend line, with the percentage above normal and the post that most likely caused it. A spike day with no obvious post behind it usually means someone else shared you, and that is worth ten minutes of investigation.

Open any individual post and you also get “Why this post performed” — a short, plain-English read on what actually distinguished it, generated from that post's own numbers rather than from a template.

When to post, and how much it's actually worth

Instagram has never told you when to post. Every “best time to post” article you have read is a global average across millions of accounts that are not yours, which makes it worth approximately nothing.

Your report groups your own posts into time bands — Morning, Afternoon, Evening, Night, each with its hour range — and shows the average reach each band earned for you. It does the same across the days of the week. The best band and the best day are called out, and when the best band genuinely beats the worst, the report tells you by how much.

Separately, the report compares your posting times against when your audience is actually active, and where posting nearer their peak would have measurably lifted your reach, it quantifies the lift. When it has nothing useful to say, it stays quiet rather than inventing a recommendation.

Which format earns your reach

Reels, feed posts, carousels, stories. The report splits your total reach across the formats you actually publish and names the winner with its share. This settles arguments that would otherwise run for months: if Reels earn 96% of your reach, the carousel you spent four hours on is a hobby, not a strategy.

A small technical note, because it explains something you might otherwise notice and distrust. Instagram's format breakdown returns a few junk keys — internal placeholders with names like DEFAULT_DO_NOT_USE. We filter those out rather than render them. Your percentages are computed across the real formats only, which is why they will not always match a naive sum of everything the API returned.

Are you reaching strangers, or the same people every week?

This is the single most under-appreciated number in Instagram analytics, and the app buries it. Discovery is the share of your reach that came from people who do not follow you.

A low discovery percentage means you are talking to the same room every week. Your engagement rate might look wonderful — of course it does, everyone in the room already likes you — while the account quietly stops growing. A high discovery percentage means Instagram is pushing you outward, which is the only mechanism by which accounts get bigger.

Follower growth: gross follows vs net growth

Here is a distinction almost every analytics tool blurs, and blurring it will make you feel much better about a shrinking account.

• New follows is a gross number: everyone who hit Follow this month.
• Net growth is what actually happened to your follower count once the people who left are subtracted.

Gain 500 and lose 400 and your gross follows say 500 while your account grew by 100. We show both, labelled, and we never present the gross figure as growth. The gap between them is your churn, and when more than roughly a third of your new followers are leaving again, that is the number to fix — not your posting frequency.

Growth rate and steadiness

Your growth rate is monthly net growth as a percentage of your following. Around 2% a month is a genuinely healthy clip; sustained, it doubles an account in under three years. Your steadiness factor asks a different question: is the growth reliable, or is it one viral spike and eleven flat months? Above 60 is steady. A low steadiness with a high growth rate is a warning, not a victory — it means you got lucky once and have not worked out how to do it again.

Reels retention: the eight-second line

Reels get two numbers that no other format gets, and they answer different questions. Average watch time is the mean seconds watched per play — it tells you whether your hook works. Total watch time is every second anyone watched, added together — it tells you what the algorithm sees, because total attention is what Instagram is optimising for when it decides whether to keep pushing a Reel.

The rough line is eight seconds of average watch time. Below it, people are bouncing before your point lands and the problem is almost always the first two seconds. Above it, the hook is doing its job and any remaining problem is in the payoff.

Skip rate completes the picture: the share of viewers who moved on almost immediately. A Reel with high reach and a high skip rate was served widely and rejected quickly, which is a more useful thing to know than the reach number on its own.

Which posts actually drove DMs, clicks and sales

This is the part Instagram structurally cannot do, because it does not know what happened after someone left the post. InstantDM does, because InstantDM is what happened next.

For every post, the report can show how many DMs the automation sent, how many links were opened, how many were clicked, and — where you have told us what a sale looks like — how many sales followed. Chained together, that is a real funnel with real numbers at each step:

The report only draws the steps it has data for, so a funnel with three links is not a broken funnel — it is an honest one. And there is a specific, common situation it exists to catch: the post with mediocre reach that quietly produced a third of your revenue. Ranked by reach it is invisible. Ranked by what it did, it is the most important thing you posted, and it tells you exactly what to make next.

On the Sales view, the report also reads the intent of the conversations your automation handled, separating people asking to buy from people complaining. Buying demand sitting unanswered in your DMs is worth knowing about today rather than at the end of the quarter.

Four ways to read the same report

The same data answers different questions depending on who is asking. Rather than bury every chart in one endless page, the report reorganises itself around the question you actually have. Switch between them freely — nothing is recalculated, only rearranged.

The Collab Report is the one worth a closer look if you work with brands. It renders as a media kit: your headline stats, your audience demographics, your proven viral reach — plus three things you control. You can list your rates and packages, add testimonials from brands you have worked with, and hand-pick the posts you want featured. Then you send the whole thing as a link. No PDF, no screenshots, no deck that was out of date the day you made it.

Sharing a live link instead of a screenshot

Instagram Insights cannot be shared at all. You screenshot it, and whoever receives it has to take your word for it. Two things change that here.

Share the whole report

One link, openable by anyone, no login. Send it to your agency, your client, or your manager. You can set a password if the numbers are sensitive, and the recipient can set an auto-refresh interval — off, every 15 or 30 minutes, hourly, two-hourly, or a custom number of hours — so a report left open on a second monitor keeps itself current.

Share a single Reel with a brand

You ran a collab. The brand wants proof. Instead of a screenshot, send a live link to that one post: the media, the caption, and a grid of eight real numbers — reach, views, likes, comments, saves, shares, interactions, engagement rate — plus the read on why it performed. The brand sees the post's actual performance, and nothing else about your account.

Exports, and putting your report into an AI

Instagram gives you numbers you can look at. The point of this section is numbers you can use.

• PDF — the whole report, print-formatted. Hand it to a client.
• Excel — a spreadsheet with one row per post and columns for date, type, caption, reach, views, likes, comments, saves, shares, follows, interactions and the permalink. It downloads as a .csv that Excel and Sheets both open directly.
• JSON — every field behind every chart, for a developer or a script.

Chat with your report

One button copies your full report to your clipboard and opens ChatGPT or Claude in a new tab. Paste, and ask it anything: why did this post work, what should I make next, what is the pattern in my best five posts. It is a deliberate copy-and-paste handoff rather than an AI bolted into the product, and that is the feature, not a limitation — your data goes to your clipboard, not to our servers.

Point your own agent at it

Share your report and it becomes a URL that returns the whole thing as JSON, no key, no login. Paste that into your own AI agent, a scheduled job, a notebook, a spreadsheet formula. It is your data, and it should be reachable from wherever you keep your other data.

If your report is locked

If you connected Instagram before granting insights access, the report appears blurred with an “Update permissions” button over it. Nothing is broken. Instagram scopes permissions individually, and read-only insights access is a separate approval from the messaging permissions InstantDM already has.

The button sends you to Instagram, Instagram asks what you want to approve, you approve insights, and the report fills in. It takes about thirty seconds, it is read-only, and it does not grant anything that could post, comment or message on your behalf. You can revoke it from Instagram's own settings whenever you like.

Frequently asked questions