Instagram engagement rate: formulas, which one to use and what it hides

Engagement rate exists to answer a question the raw number cannot: is a post with 400 likes good? It depends on how many people saw it and how many follow you. Dividing interaction by a base makes posts of different sizes comparable.

The problem is that more than one base is possible, and the platform does not hand you the rate. Every tool picks a formula, which is why two dashboards show different rates for the same profile without either being broken.

What counts as an "interaction" varies too. The fullest reading adds likes, comments, saves and shares. Many public calculators only add likes and comments, because those are the two numbers visible from outside the profile.

By follower is the formula anyone can compute from outside, because follower count is public. It is what almost every online calculator uses and what agencies use to assess creators. It carries a known bias: the larger the account, the lower the rate, because a growing share of the base is inactive and never sees the content.

By reach can only be computed from inside, with access to insights. It is the more honest one for judging content, because it removes from the calculation everyone who never saw the post. If a post reached 2,000 accounts and got 200 interactions, 10% of the people who saw it reacted, and that is information about the piece, not about the size of the profile.

It depends on the decision you are about to make.

• Judging a specific piece: use by reach. It is the only one that isolates content quality from audience size.
• Judging profile health over time: use by follower. It captures the effect of a base that grew without engaging, which the reach formula hides.
• Comparing with another profile or negotiating a sponsorship: use by follower, because it is the only one computable from outside, and say so explicitly.

The expensive mistake is mixing them. Comparing your reach-based rate against a market benchmark someone measured by follower produces a wrong conclusion wearing the costume of data.

There is no universal number, and any table promising one is an average of a sample that is probably not yours. The rate varies with account size, niche, format and the goal of the content.

The benchmark that works is your own. Take the last thirty posts, compute the rate for each with the same formula and use the median as your baseline. Median, not mean, because a single viral drags the average up and makes everything else look like failure.

Two reading corrections almost nobody applies: compare format with format, because carousels and Reels have different interaction patterns, and give the content time before judging, because video accumulates interaction for weeks while a static post concentrates almost everything in the first 48 hours.

• Inflated base. Bought or giveaway followers sit in the denominator and never interact, which drags the follower-based rate down without anything getting worse.
• Off-niche viral. A post that reaches a lot of the wrong people has high reach and a low reach-based rate. That is not bad content, it is content delivered to people who were not the audience.
• Likes as a proxy for value. A like is the cheapest interaction for the viewer. Saves and shares cost more and are worth more as a signal, both to the algorithm and to the business.
• Bought engagement. Bot interaction raises the rate and moves no revenue. If the rate grows and profile visits do not follow, be suspicious.

• I stated which formula I am using before comparing any number.
• I added all four interactions, not just likes and comments.
• I used the median of the last thirty posts as a baseline, not the mean.
• I compared carousels with carousels and Reels with Reels.
• I cross-checked the rate against profile visits, to separate interaction from intent.
• I waited at least 72 hours before judging video content.