Every so often I review which marketing terms get searched the most on Google in Chile, and engagement consistently sits near the top: around 3,900 searches a month, more than "sales funnel" and several times more than "email marketing". The fun part is what the results page shows when you search it: half the results are articles explaining what engagement is, and the other half are calculators to measure it on Instagram or TikTok. In other words, people want more than the concept, they want a number that tells them whether they're doing well or not.
In this guide I'll cover both: what engagement is, how to calculate it (with the formulas and a worked example), what counts as a good rate, and how to improve it with actions you can execute this week. I'll also tell you why this metric started to matter outside social media, which is the part that rarely gets explained.
What is engagement?
Engagement is the level of interaction your audience has with your content: likes, comments, shares, saves, clicks. The word points at the idea of commitment, and that helps you understand what's really being measured: while reach counts how many people saw what you published, engagement counts how many people did something about it.
The difference matters because reach can be bought and engagement can't (well, it can, but we'll get to why that's a bad idea). If 50,000 people saw your post, that says something about the algorithm and your budget. If 2,000 commented on it or saved it, that says something about your content.
The analogy I use with clients is a talk: reach is how many people walked into the auditorium, engagement is how many raised their hand, took notes or stayed to chat afterwards. You can fill the auditorium by giving away tickets, but you can't force anyone to take notes.
How do you calculate engagement?
The base formula is simple: add up the interactions on a post and divide them by the people who saw it, multiplied by 100 to get a percentage.
Engagement rate = (total interactions / reach) x 100
A worked example with round numbers so it's easy to follow: if a post reached 10,000 people and gathered 300 likes, 80 comments, 15 shares and 25 saves, that's 420 interactions over a reach of 10,000. That gives you an engagement rate of 4.2%.
Now, the detail that creates confusion: there's more than one version of the formula, and all three are in use.
Engagement over reach: interactions divided by the people who saw the content. It's the most honest way to evaluate the quality of a post, because it compares interactions against people who actually had the chance to interact.
Engagement over followers: interactions divided by your total follower count. It's the one most online calculators use, and the one brands use to evaluate influencers, because follower counts are public and reach is not. Its weakness is that it punishes accounts whose content reaches lots of people who don't follow them.
Engagement over impressions: interactions divided by the total number of times the content was displayed (the same person can see it three times). It produces lower rates than the other two and is mostly useful for comparing formats within your own account.
Which one should you use? Whichever you can keep constant over time. Problems show up when you compare one month calculated over reach against another month calculated over followers and draw conclusions from that difference.
What is a good engagement rate?
This is the most asked question and the one with the least serious answer. You'll find tables saying that between 1% and 3% is normal and anything above 5% is excellent, and as a rough reference they're fine, but they carry a statistical trap worth understanding before you celebrate or despair.
Engagement rate drops as the audience grows. An account with 2,000 followers where half are acquaintances, clients and genuine fans can hit 8% without breaking a sweat. An account with 500,000 followers sitting at 3% is doing remarkable work. And not because the small account does it better: the relationship between audience size and strength of the bond simply works that way, the more massive the reach, the more diluted the average commitment.
On top of that, the rate changes by network (TikTok and Instagram tend to run higher than Facebook or LinkedIn), by format and by industry. A football meme and a pension fund's post compete in different leagues.
That's why my recommendation is to stop chasing the universal benchmark and compare against your own history: calculate your average rate over the last three months and evaluate every new post against that baseline. That comparison does control all the variables, because the only thing changing is the content.
Why does engagement matter beyond social media?
Here's the part that interests me as a search consultant, because engagement stopped being an exclusively social media topic a while ago.
First, inside the platforms: algorithms use a post's early interactions to decide how many more people to show it to. Engagement, besides measuring the outcome, feeds it: a post that interacts well in its first hour travels further, which in turn gives it more chances to interact, a loop that rewards content that generates conversation.
Second, outside the platforms: search has a sister metric that plays the same role, CTR, which measures how many people click your result when they see it. I wrote a full guide on CTR and the logic runs parallel: you can appear in front of thousands of people in the results and still lose the visit if nothing invites the click.
And third, the new layer: the AI systems that recommend brands form their opinion from public signals, including the conversation that exists around your brand on social networks and communities. I wrote about how AI models choose which brands to recommend, and the short version is that a brand generating real conversation leaves traces these systems read. Engagement went from vanity metric to one of the signals in a complete brand profile, alongside the organic metrics growth teams report.
How do you improve engagement?
Here's the complete recipe, ordered by impact according to what I've seen in content projects.
Reply to everything during the first hours. Every reply of yours is one more interaction, it invites the thread to continue, and platforms read that conversation as a signal of living content. It's the cheapest lever there is and the one most accounts ignore. If there's a community manager on the team, this is their court.
Ask questions that can be answered in one line. "What do you think about the change?" performs worse than "team A or team B?". The friction of replying defines how many people reply, and lowering that friction is content design, not manipulation.
Prioritize the formats your audience already rewarded. This connects with the previous section: if you measured your rate per post for three months, you already know whether your carousels outperform your videos or the other way around. Double down on what works before experimenting with what's new.
Post when your audience is online, not when it suits you. Each platform's native stats show you your specific audience's hours. Given the effect of early interactions, posting at the right time can change a post's entire trajectory.
Write for a segment, not for everyone. Generic content generates polite scrolling. Content that speaks to a specific group generates comments like "this happened to me yesterday", which are the ones that pull more people into the conversation.
And the promised warning: buying followers or interactions destroys exactly what you're trying to build. Bought followers don't interact, so your rate drops, the algorithm shows you less, and any brand or client who checks your account with one of those calculators from Google notices in 30 seconds. Same with engagement bait, those posts that ask for interaction without giving anything back: platforms have been explicitly penalizing it for years.
Wrapping up
The exercise I suggest needs no budget and no paid tools: take your last 30 posts, calculate the engagement rate for each one with the same formula (any of them, but the same one), and rank them from highest to lowest. Then look at the top five and the bottom five, and hunt for the pattern: which topic, which format, which type of opening repeats in each group.
That ranking is your audience study built with your own data, and it answers with evidence the question generic benchmarks can't: what the people who already chose to follow you want to see. Publishing more of that is the most direct way to raise your engagement, and along the way it turns a metric many teams report out of habit into a concrete editorial decision.