Which Instagram metrics actually matter?

Follows per Reel, the share of reach from non-followers, saves per post, engagement measured on reach, profile visit to follow conversion, and revenue per post. Four to six is the workable range, since more becomes an admin task people abandon.

How do I run a proper test on my content?

Change one variable, run it across five posts, compare medians measured on reach, and give each post a full week before counting since Reels keep accumulating. Changing several things at once means the result cannot be attributed to anything.

What does high reach but low follows mean?

The content is working and the profile is not converting. People arrive and cannot tell what they would get by following. Fixing the name field, bio and top row of the grid improves every post you have already published, not just future ones.

Should I track averages or medians?

Medians. A single viral post distorts an average enough that a poor month reads as a good one. The median describes what a typical post does, which is the thing you are trying to improve. Track the outlier separately to understand why it worked.

How often should I check Instagram analytics?

Monthly for decisions, weekly at most for a glance, and never daily. Reels accumulate reach over days or weeks, so early readings mislead, and daily checking produces emotional decisions such as abandoning a format shortly before it would have worked.

What numbers does Instagram not tell me?

The ones that decide whether this is a business: revenue per post, how many pitches produce a deal, how long the cycle runs from first contact to payment, how much is currently unpaid, and what share of income comes from your largest client.

Instagram analytics playbook

Instagram gives you dozens of numbers and no guidance on which should change your behaviour. Most creators respond by checking daily, feeling something, and acting on none of it.

Measurement is only worth doing if it ends in a decision. This is the whole loop: instrument, read, test, decide.

Part one: the instrument

Four to six metrics, tracked monthly, in one sheet. More than that becomes an admin task you abandon.

The build is covered in analytics dashboard: a monthly row per metric and a second tab listing individual posts, sorted by follows gained. Twenty minutes to set up, ten minutes a month to maintain.

The distinction that matters is leading versus lagging. Follower count and revenue tell you what already happened. Follows per Reel, non-follower reach share and saves respond to what you change and predict where the lagging numbers go. Decide from the leading ones.

Part two: measure engagement honestly

Engagement rate is the number brands check, and it is the one creators most often calculate in a way that flatters them, because three accepted formulas produce wildly different answers for the same post.

The engagement rate guide covers which formula to use for your own judgement, which one brands will quote at you, and the benchmark bands by follower tier. Measure on reach for your own decisions.

Track the median rather than the average. One viral post distorts an average enough to make a bad month look good, and the median is what describes a typical post.

Part three: reading the patterns

Each combination points at a different problem, and the fixes are unrelated.

High reach, low follows. The content works, the profile does not convert. Fix the profile, which improves every post you already published.
Low non-follower share. Content is circulating among people who already know you. A distribution problem.
Good reach, low saves. Entertaining but not useful. Fine unless you sell something.
Retention cliff in the first two seconds. A hook problem, fixable in the edit. See hook formulas.
Steady drain through the middle. A structure problem rather than a hook one. See the storytelling framework.
Follows falling while reach holds. Usually topic drift. The wrong people are seeing it.
Reach collapsed suddenly. Work through mistakes that kill reach, starting with Account Status.
Revenue flat while reach grows. A selling problem, not a content one.

Part four: testing properly

Most creators change format, hook style, caption length and posting time in the same week, then cannot attribute the result to anything.

Pick one variable. Carousel against Reel, or question in video against question in caption.
Run it across five posts. Fewer and one outlier decides the outcome.
Compare medians, measured on reach.
Give each post a full week, since Reels keep accumulating for days.
Keep the winner, then test the next thing.

At four posts a week that is roughly ten days per experiment. Three tests a quarter compounds into a genuinely different account within a year.

Part five: the numbers that are not on Instagram

The platform reports reach and interactions. It does not report the things that decide whether this is a business.

Revenue and revenue per post come from your own records. So does deal-cycle length, pitch-to-deal conversion, and how much money is currently sitting unpaid. Those live in your CRM, and they are usually more actionable than anything Insights shows.

Client concentration belongs here too. What share of income came from your largest client is a risk metric, and anything over about half means adding a stream matters more than adding reach.

Part six: the ritual

An instrument nobody reads is a spreadsheet. The habit is what converts it into decisions.

Ninety minutes, first working day of the month, with an agenda that ends in three written decisions: one thing to do more of, one to stop, one experiment. The full structure is in the monthly review template.

The stop is the part that matters. Creators add continuously and remove nothing, and a review that never removes anything is reporting rather than deciding.

How often to look

Monthly for decisions. Weekly at most for a glance. Never daily.

Reels accumulate reach for days and sometimes weeks, so a post judged after an afternoon tells you nothing, and daily checking produces emotional calls like abandoning a format two weeks before it would have worked. The one exception is the hour after publishing, where replying to comments genuinely affects the post.

What to ignore

Follower count as a daily metric. The least actionable number available.
Likes. Weaker than saves and shares in both signal and meaning.
Comparisons with other creators. Different niche, size and content. Compare against your own last quarter.
Single-post conclusions. One post is noise.
Any metric you have never acted on. If it has not changed a decision in three months, take it off the sheet.

The whole system answers one question every month: what should I make more of, and what should I stop. Everything that does not help answer it is decoration, and most of what Instagram shows you is decoration.