Engagement rate is interactions divided by reach. It is the most quoted number in social measurement and it has a structural flaw that shows up at exactly the moment you most want a reliable reading.
The denominator problem
Reach is on the bottom of the fraction. So anything that reduces reach increases engagement rate, all else equal. A post shown to five hundred committed followers will almost always post a better rate than a post shown to forty thousand strangers, because the five hundred already chose you.
Which means the metric is at its most flattering when a post fails to travel, and at its harshest when a post succeeds in reaching beyond your audience. It is inverted relative to the outcome most accounts actually want.
A metric that improves when a post reaches fewer people is not measuring what you think it is.
Why this is not a technicality
Because it produces confident, wrong decisions. An account looks at two posts, sees 8% against 3%, concludes the first is the better template, and makes more of it. The first reached five hundred people. The second reached forty thousand and brought in followers. The rate ranked them backwards and the ranking was acted on.
This pattern is common precisely in accounts that have started to break out of their existing audience — which is to say, in accounts at the moment where good decisions matter most.
When engagement rate is fine
Comparing posts that reached similar audiences of similar composition. Within that constraint the denominator is roughly held constant and the ratio does what it claims. That is a real use, and it is narrower than how the number is normally deployed.
It is also reasonable as a health check on a following: if rate collapses across everything over months while reach holds, that is worth investigating.
What to use instead
Split the question. Reach and engagement are two different outcomes and compressing them into one number destroys information:
- Did it travel? Reach, measured against your own baseline rather than against other accounts.
- Did it land? Watch time and completion among the people it reached.
- Did it move anyone? Saves and sends, and follows per thousand reached.
Three numbers that answer three questions beats one number that answers a blend of them badly. Instagram's own guidance describes ranking in terms of how many and how quickly people interact — a set of signals, not a single ratio, which is a reasonable hint about how to read them.
The general principle
Be suspicious of any ratio whose denominator is the thing you are trying to grow. Cost per acquisition, engagement rate, conversion rate on a shrinking audience: they all share the property that you can improve the number by doing less of what you want.