Engagement Rate Formulas by Platform: Why the Same Post Scores 2% and 4.8%

One name, three denominators, three answers

Work the same post three ways and the disagreement stops being abstract. Twelve hundred interactions on an account with 40,000 followers gives 3.0%. The post reached 25,000 unique accounts, so on reach it gives 4.8%. It was displayed 60,000 times, so on views or impressions it gives 2.0%. Nothing about the post changed between those three numbers — only the question being asked. Followers asks what share of the people who subscribed reacted, which is a question about the health of the list. Reach asks what share of the people who actually saw it reacted, which is a question about the content. Impressions asks what share of the appearances produced a reaction, which is a question about delivery efficiency and is the only one of the three that can be dragged around by the algorithm showing a post to the same person twice.

The ordering is fixed and worth memorising, because it tells you which direction a stranger's number has been bent. Impressions are always at least as large as reach, since every unique account contributes at least one appearance. Reach is capped by neither followers nor anything else, because a post can travel far beyond the subscriber list — which is why a viral post can produce a rate below 1% on reach while sitting above 10% on followers. In the ordinary case, though, reach is smaller than followers because most posts never get shown to the whole list, and that is exactly why the reach-based version is the one that appears in media kits. It is not dishonest; it is simply the version that flatters, and nobody chooses the unflattering one by accident.

What each platform actually hands you

Only LinkedIn publishes a formula and computes the result for you: clicks plus likes plus comments plus shares plus follows, divided by impressions. Two things follow from that. The numerator is unusually wide — a click on a link counts the same as a comment, and a new follower counts too — so a LinkedIn engagement rate is structurally higher than a like-and-comment rate on any other platform, and comparing them proves nothing. The denominator is impressions, where an impression is counted when the post is at least half on screen for a moment or is clicked, so a post that scrolls past unnoticed still enters the denominator.

The others hand you counters and leave the arithmetic to you, and each has a quirk that changes the answer. Instagram reports accounts reached and, since the 2024-25 consolidation, a single Views figure that absorbed the old impressions and plays metrics — so an older benchmark quoting Instagram impressions is not measuring what Views measures today. TikTok reports video views, and because the feed autoplays and loops, views accumulate faster than attention does, which pushes TikTok rates down relative to a reach-based Instagram rate on identical content. YouTube reports views and hides dislikes, so the numerator is thinner there than anywhere else, and it uses the word impressions for something else entirely: thumbnail appearances in search and browse, most of which were never watched. A rate built on YouTube impressions is a click-through rate wearing the wrong label.

Using the number without fooling yourself, or a sponsor

Three rules make the metric usable. State the formula next to the number, every time, in the form interactions over reach or interactions over followers — an engagement rate without its denominator is not a measurement, it is a decoration. Compare only against your own history on the same formula and the same platform, because that is the only comparison where the denominator is held constant. And treat published benchmarks with suspicion unless they name their denominator, since a benchmark of 4% on reach and a benchmark of 4% on impressions describe wildly different levels of performance.

The stakes stop being academic the moment money is attached. A brand offering $2,000 for a sponsored post is usually pricing against a rate quoted in a media kit, and if that rate was computed on reach while the brand's internal benchmark was computed on impressions, the two sides have agreed a price against numbers that were never the same measurement. The honest move — and the one that survives an audit — is to quote both: interactions over reach and interactions over followers, with the raw counts beside them, so the buyer can rebuild whichever version their own reporting expects. Raw counts are the only figures nobody can argue with, and giving them away costs you nothing you were entitled to keep.

Frequently asked questions

What counts as a good engagement rate?

The question cannot be answered without the denominator, which is why every published benchmark disagrees with every other one. Four percent on reach, four percent on followers and four percent on impressions describe three different levels of performance for the same account, and most articles quoting a benchmark never say which they mean. Account size matters too: rates fall as follower counts rise, because large accounts are shown to more people with a looser interest in them, so a small account outscoring a large one on the same formula is the normal outcome rather than a triumph. The only benchmark that means anything is your own median over the last few months on one platform and one formula.

Should I divide by followers or by reach?

It depends on what you are trying to learn, and the honest answer is that you probably want both. Dividing by followers measures whether the audience you built still cares, which is the number to watch if you suspect the list has gone stale or has been padded with inactive accounts. Dividing by reach measures whether the content works on the people who saw it, with distribution taken out of the picture. Track the two side by side and a divergence is diagnostic: reach-based rate holding while follower-based rate falls means distribution shrank, not that the work got worse.

Do saves, shares and clicks belong in the numerator?

There is no standard, so the only rule that matters is consistency: pick a numerator, write it down, and never change it mid-series without recomputing the history. That said, the interactions differ in cost to the person performing them, and the cheap ones are the ones that can be bought. A like takes a fraction of a second; a save or a share is a deliberate act, and both correlate far better with the outcomes anyone actually cares about. Clicks are the awkward case — LinkedIn and most Facebook formulas count them, which is one reason those rates read high, and a click on a misleading thumbnail is not engagement in any sense worth paying for.