Engagement rate, and the number underneath it

Three legitimate answers for one post, and the reason two agencies can report very different figures without either lying.

There is no such thing as the engagement rate

Take one post: 500 engagements, 10,000 followers, 6,000 accounts reached, 9,000 impressions.

• Divided by followers: 5.0%
• Divided by reach: 8.3%
• Divided by impressions: 5.6%

All three are correct. All three get called “engagement rate”. The highest is 1.7 times the lowest, and on other posts the spread is far wider — which is how two agencies can report very different numbers for identical performance and both be telling the truth.

The figure alone is meaningless. The figure with its denominator is a fact.

What each one is good for

Of followers

The figure in almost every media kit, and the least informative of the three. It ignores whether anybody actually saw the post, so a post shown to nobody and a post shown to everybody are treated the same. It flatters small accounts, and — the part people miss — it falls as an account grows even when the posting is performing exactly as well, because the denominator gets bigger while reach does not keep pace.

It is quoted constantly because follower counts are public, so it is the only one an outsider can calculate.

Of reach

Accounts that saw the post. This is usually the fairest, because it answers the question you actually meant: of the people who saw this, how many did something? It is also the figure quoted least often, largely because it is usually the one that needs explaining.

Of impressions

Every time the post was shown, including the same person seeing it three times. Always the lowest of the three and always the most conservative. Useful for paid work, where you are being charged by the impression.

Which actions count

Likes, comments, shares and saves is the common set. Clicks, profile visits and follows are sometimes added. There is no standard, so what matters is choosing once and staying consistent — adding saves for one comparison and dropping them for the next produces a change that has nothing to do with performance.

Be careful with a platform that does not report an action at all. A missing figure is not a zero: if saves are unavailable, the honest thing is to exclude saves from the calculation, not to count them as none.

Typical and overall are different questions

For a set of posts there are two defensible aggregate figures.

Every post counted equally — work out each post's rate, then average those. This answers “what does a typical post of mine do?”

Totals divided by totals — add up all engagements, add up all reach, divide once. This answers “what did the account do overall?”

They agree while your posts perform similarly and part company the moment one does not. And the direction they part is worth understanding: a post that travels escapes your own audience and reaches strangers, who engage far less than followers do. Its rate is lower, and because its reach dwarfs everything else it drags the overall figure down.

So a genuinely successful month can produce an overall engagement rate well below your typical post, and it is not a decline. Reporting only whichever of the two figures suits the story is a standard trick, which is why both should always be shown.

Use the median

When one post has run away with the numbers, the mean sits somewhere no individual post ever was. The median — half your posts above, half below — is a much better description of normal, and it is the number to quote when someone asks what you usually get.

The spread is worth showing too. A median of 3% with a range from 2.5% to 3.5% is a consistent account. The same median with a range from 0.4% to 40% is a completely different account, and the median alone hides that.

On benchmarks

Published “good engagement rate” figures are almost always quoted without the denominator, without the platform, without the account size and without saying which actions counted. Treat any single number offered as a universal benchmark as unusable.

The comparison that means something is against your own posts over time, with the same denominator and the same actions each time.

Frequently asked questions

What is a good engagement rate?

There is no answer without knowing the denominator, the platform, the account size and which actions were counted — and benchmark figures are almost always quoted without those. Compare against your own posts over time instead, keeping the denominator and the actions the same each time.

Why do I get three different numbers for one post?

Because followers, reach and impressions are three different denominators. All three rates are legitimate; quoting one without saying which is what makes it meaningless.

Which should I put in a media kit?

Whichever you choose, label it. Rate against reach is the fairest and the most defensible if anyone checks; rate against followers is the most commonly expected. Giving both, labelled, is better than giving one unlabelled.

Why did my engagement rate fall when a post did really well?

Because a post that travels reaches people who are not your audience, and they engage far less. Its own rate is lower, and its reach is large enough to pull the overall figure down. Your typical post has not changed — which is why both figures are worth looking at.

Should saves and shares count?

That is your decision. What matters is making it once and keeping it consistent, or comparisons measure your choice rather than your posts.

Does my data leave my browser?

No. Everything is calculated on your device and nothing is sent anywhere.