Instagram Analytics: How to Read Your Own Numbers

Most people open Instagram Insights, look at a chart that goes up or down, feel briefly good or briefly bad, and close it again. That is not analysis. Analysis is deciding, before you look, which numbers you are going to write down — and then writing them down every month whether they flatter you or not.

This page is the whole method: what each figure counts, how to turn it into a rate that survives a change in follower count, and what the numbers genuinely cannot tell you. Every calculator linked here runs in your browser and will never ask for your Instagram login.

Where Instagram’s numbers live and what you can get

Account-level insights are only available on professional accounts — the Creator and Business account types. A personal account shows you likes and comments on a post and nothing else. Switching is free and reversible, and it is the single change that makes everything on this page possible.

The numbers then arrive in two places. Per-post insights sit behind the “View insights” link under an individual post and show reach, interactions, profile activity and, for Reels, watch time. Account-level insights aggregate that across a date range and add audience data: where your followers are, when they are online, and how the total moved.

Two limits are worth knowing up front. The account-level date range is finite and has changed more than once over the years, so any long history has to be copied out by hand each month. And audience breakdowns are suppressed below a follower threshold, because the sample is too small to anonymise — if your audience tab is mostly empty, that is why, and it is not a bug.

Reach, impressions and views count three different things

This is the most commonly confused trio on the platform, and getting it wrong quietly corrupts every rate you calculate afterwards.

Reach is people — unique accounts that saw the post at least once. If the same person scrolls past your carousel four times, reach counts one.

Impressions are appearances. Every rendering on someone’s screen adds one, so that same person adds four. Impressions are therefore always equal to or larger than reach, and the ratio between them tells you how often the average viewer came back.

Views (also labelled plays) count video starts, and Instagram has redefined the boundary between plays, views and replays more than once. It is the least stable of the three and the one to be most cautious about comparing across years of your own data.

Our guide to reach vs impressions vs views works through the edge cases, including carousels, Story frames and re-shares. The short version: use reach as your denominator wherever one is needed, because it is the only one of the three that counts humans.

Engagement rate: the one number worth tracking

Raw likes are useless for comparison because they scale with audience. A post with 900 likes from 60,000 followers is doing considerably worse than a post with 90 likes from 2,000 followers. Converting to a rate removes the audience from the comparison, which is exactly what you want.

There are two defensible formulas, and they answer different questions.

Rate by followers answers “how responsive is my audience?” It is the version brands quote, because followers are the only input they can verify from the outside. Rate by reach answers “how good was this post?” — it removes distribution from the equation and measures the content on its own terms. Reach-based rates are always higher, often by a factor of two or three, because reach is a fraction of followers for most accounts.

The engagement rate calculator computes both. It also takes saves and shares as optional inputs, because those are frequently the more meaningful signals and most published formulas ignore them entirely.

Averaging across posts, not cherry-picking one

A single post tells you almost nothing. One reel catching a wave will double your monthly average, and one dud will halve it. The calculator has a multi-post mode that takes up to twelve posts — one line each, likes and comments — and returns the average, the best, the worst and the spread between them.

The spread is the number people ignore and shouldn’t. A tight spread means your audience responds consistently, which makes your average a real forecast. A wide spread means your average is a fiction sitting between two clusters of very different posts, and the interesting question is what separates them.

Benchmarks by follower tier, and how far to trust them

Published creator-marketing reporting consistently finds that engagement rate falls as follower count rises. Small accounts have audiences who chose them deliberately; large accounts accumulate passive followers. The tiers below are the ones our calculators use, and they are rules of thumb rather than thresholds Instagram enforces or publishes.

The “around typical” column is the band from 60% of the benchmark up to the benchmark itself — the range the calculator treats as unremarkable rather than weak.

The niche adjustment on top of the tier

Follower tier is only half the comparison. Some verticals are simply chattier than others: people save recipes and comment on animals, and they read finance posts in silence. The calculator therefore multiplies the tier baseline by a niche factor before comparing:

Factors run from ×1.20 for pets and animals, through ×1.15 for beauty and fitness, ×1.10 for food and art, ×1.00 for travel, fashion and general lifestyle, down to ×0.85 for business, ×0.80 for tech and ×0.75 for finance. A nano-tier finance account is therefore measured against 3.0%, not 4.0%. The factor is our own rule of thumb about how interaction-heavy a vertical tends to be, not measured Instagram data — which is why the tool shows the full derivation and lets you divide it back out. It adjusts only the comparison, never your calculated rate.

One further caveat matters more than the table does: any published benchmark is a snapshot of whichever set of accounts the publisher happened to sample. What counts as a good engagement rate goes further into where these numbers come from and why the honest answer is always “compared to what?”

Reach rate: distribution and content are different problems

Reach rate is the share of your followers a typical post actually reaches:

The calculator treats anything under 20% as low and 45% or above as strong, with the ordinary band in between. Those are rules of thumb rather than thresholds Instagram publishes, and they are strongly format-dependent — reels routinely reach people who do not follow you, so a reels-heavy account can exceed 100% without anything unusual happening.

The reason to track it separately is that it splits one vague worry into two specific ones. A low reach rate is a distribution problem: the post is not being shown, and the fix is upstream of the caption — format, cadence, timing, or an audience that has drifted. A normal reach rate with a low engagement rate is a content problem: people saw it and scrolled past, and no amount of better timing repairs that. Diagnosing the wrong one wastes months.

Saves and shares: the quiet signals

Likes are cheap. A like costs a double-tap and carries almost no information about whether the post was useful. Saves and shares cost something — a save is someone deciding they will want this later, a share is someone putting their own reputation behind it.

It is widely believed among creators that saves and shares carry more weight in distribution than likes do. That belief is plausible and consistent with what people observe, but this specific claim has not been confirmed in a way anyone can check. Treat it as a working theory, not a fact.

What is not in doubt is that they tell you something likes cannot. A high save rate means the post was reference material. A high share rate means it was social currency. Those are different content strategies, and you can only choose between them if you record both. Include them in your engagement rate — the calculator has fields for both — and track them as their own rate against reach.

Follower growth: reading a trend, not a spike

Follower count is the number everyone watches and the one that moves for the least interesting reasons. A single post travelling well adds followers who never chose you deliberately, and a meaningful share of them leave within a fortnight. What matters is the net figure over a period long enough to absorb that noise — thirty days is the shortest useful window.

Once you have a net monthly figure, the follower growth calculator turns it into a projection. It runs two models side by side and shows the gap between them:

• Linear. You keep adding the same number of followers per day. Conservative, and usually closer to reality for accounts that are not actively accelerating.
• Compounding. Your growth rate stays constant as a percentage, so the absolute gain rises with the base. Optimistic, and the model that produces the hockey-stick charts people post screenshots of.

It also takes a monthly churn percentage, because unfollows are real and ignoring them makes every projection wrong in the same direction. Projections are capped at twenty-four months, which is a deliberate limit: beyond two years the arithmetic is still valid and the forecast is worthless.

Rather than one line, it runs three scenarios — pessimistic at half your measured gain, current at exactly what you measured, and optimistic at one and a half times — and shades the band between them. Churn stays the same in all three, because churn is a property of the audience you already have rather than of how well next month goes. There is also a deadline mode: give it a target date instead of asking when you will arrive, and it returns the rate you would need to hit it.

If your target is the 10,000 mark specifically, how long it takes to reach 10,000 followers works through the arithmetic at several realistic starting points. And if you have ever been tempted to shortcut the number, the maths of buying followers shows exactly how the denominator destroys your engagement rate — which is the number anyone paying you will actually check.

Timing: your own data beats every published chart

Every “best time to post” article is an average of accounts that are not yours, in time zones that are not yours. The useful version of that question is answered inside your own insights, in the panel showing when your followers are most active.

The complication is time zones. Your followers’ active hours are recorded in their local time, but you post in yours. The best time to post calculator handles the conversion: give it up to three audience time zones with rough percentage shares and an audience type, and it returns a seven-day heatmap in your local time.

The activity curves it uses are documented on the tool page — five audience profiles, each a set of weighted local-hour windows, with weekday and weekend variants. It also weights by what you are posting, because a format keeps earning after you publish it: a feed post is credited over the following few hours, a reel over about eight, and a story over a short window that pushes the recommendation hard toward peak scrolling hours. You can black out hours you will never post in, and export the resulting slots as a calendar file.

All of that is a transparent model, not measurements of your account. The honest use of the output is as a starting hypothesis you then test against your own insights. Why every posting-time chart is wrong for you explains the reasoning in full.

A monthly routine that takes twenty minutes

Analytics only compounds if it is boring and repeated. Here is a routine that fits in one sitting on the first of the month.

1. Record the headline five. Followers at month end, net change over the month, average reach per post, reach rate, and your average engagement rate across the month’s posts. Five numbers, one row in a spreadsheet.
2. Run the multi-post average. Take up to twelve posts, put likes and comments into the engagement rate calculator, and record the average and the spread.
3. Note the top three and bottom three. By engagement rate against reach, not by likes. What the top three have in common is next month’s content strategy.
4. Check the save rate separately. Saves divided by reach. If it is climbing, your reference-style content is landing.
5. Update the growth projection. Feed the net monthly change into the growth calculator and see whether the required daily gain for your target has moved closer or further away.
6. Re-check timing twice a year, not monthly. Audience activity patterns shift slowly. Checking them every month is noise.

Six months of that spreadsheet is worth more than any benchmark table, because it is the only dataset actually about you — and it is what makes pricing conversations straightforward. The creator pricing guide covers how those figures become a rate card.

What analytics cannot tell you

Being clear about the limits is part of using the numbers well.

They cannot explain why. Insights tell you a post reached 40% fewer people. They do not tell you whether that was the hook, the format, the hour, the season or a platform-wide fluctuation. Every explanation you attach to a number is a hypothesis you invented.

They cannot see the algorithm. Instagram publishes very little about how ranking works, and what it does publish is high-level. Any confident, specific claim about “what the algorithm rewards” is inference from observed behaviour. Some of it is probably right. None of it is documented.

They cannot measure the outcome you care about. Reach is not revenue, and engagement is not trust. Plenty of accounts with modest numbers convert extremely well because the audience is exactly right. If your goal is clients, sales or a mailing list, the number that matters lives outside Instagram entirely.

They are not a scoreboard. Comparison against accounts whose niche, age, geography and history you cannot see is the fastest way to draw a wrong conclusion from a right number.