How to Calculate Engagement Rate on YouTube, TikTok and Instagram

Ask three tools for a creator’s engagement rate and you can get three different numbers, all of them correct. They aren’t measuring the same thing. One divides likes and comments by followers, another divides every interaction by views, and a third averages a different set of posts. Without the formula, an engagement rate is just a percentage with no unit.

This guide is for creators putting numbers in a media kit, brands and agencies comparing creators, marketers reporting on campaigns, and students who need the formulas written out. It covers each formula exactly, which interactions YouTube, Instagram and TikTok actually show you, worked examples, a spreadsheet recipe, and what “good” looks like according to studies that publish their method.

The definitions come from the platforms’ own help centers and developer documentation. The benchmarks come from named, dated reports by Iqfluence, CreatorDB, Quid and Dash Social, each with its method stated, because a benchmark without a method can’t be compared with anything.

What is engagement rate?

Engagement rate is the share of an audience that interacted with a piece of content, expressed as a percentage: interactions divided by an audience number, multiplied by 100. Everything depends on which interactions go on top and which audience number goes underneath.

The inputs, defined the way the platforms use them:

• Engagements (or interactions) are the actions people take on a post: likes, comments, shares, saves and, on some platforms, reposts. Meta’s developer documentation defines Instagram’s “total interactions” as likes, saves, comments and shares, minus unlikes, unsaves and deleted comments.
• Followers (subscribers on YouTube) are the accounts that follow the creator. It’s the only audience number anyone can see from outside.
• Reach is the number of unique accounts that saw a post at least once. Meta labels Instagram reach as an estimated metric.
• Views count plays rather than people, and each platform sets its own rules for what counts as a play. Meta defines Instagram views as the number of times the media “has been played on Instagram”.
• Impressions count how many times content was displayed, whether or not anyone stopped. On YouTube, impressions are thumbnail impressions: how many times your thumbnails were shown to viewers.

None of the three platforms gives you one official engagement rate for every post. YouTube Analytics reports likes, comments and shares as counts, and its Engagement tab leads with watch time and average view duration. Instagram’s API gives you total interactions but no rate. In some regions, TikTok Studio shows an engagement rate when you look at insights on other creators’ posts, but its help page doesn’t publish the formula. In practice, you calculate it yourself.

Our own methodology judges creators on average views of their last 10 posts against their size, and on whether the comments contain real conversations, rather than on one engagement percentage.

What are the engagement rate formulas?

There are four common denominators, and two ways to summarise them across posts. Every formula below multiplies by 100 to give a percentage.

ERF (engagement rate by followers) is the formula behind most of the public benchmarks in this guide, because followers, likes and comments are public. ERR (engagement rate by reach) is the fairer measure of content, because it only counts people who actually saw the post. ERV (engagement rate by views) suits short video, where most viewers may not follow the account at all.

Per post or account level?

A per-post engagement rate measures one post. An account-level engagement rate summarises a creator’s recent posts, usually the last 10 or 12. Quid’s 2026 benchmark report explains why it reports engagement per post: it “controls for post volume and audience size”. If you instead add up a whole month of interactions and divide by followers, an account that posts daily will look far more engaging than one that posts weekly, even when each post performs the same.

Median or mean?

Use the median, the middle value when you sort the posts, rather than the mean, the arithmetic average. One viral post can drag the average far above what a typical post does. Iqfluence’s September 2026 study shows how big the gap gets: for Instagram accounts with 1,000 to 5,000 followers, the median engagement rate is 3.10% but the average is 6.16%.

Here’s a hypothetical example. Say a creator’s last 10 posts scored 1.7%, 1.8%, 1.9%, 2.0%, 2.1%, 2.1%, 2.2%, 2.3%, 2.4% and, after one went viral, 14.5%. The mean is 3.3%. The median is 2.1%, much closer to what a typical post does.

Which interactions should you count on each platform?

Count only what you can count consistently. If you compare creators from the outside, that usually means likes and comments, because they’re the interactions every platform shows publicly. If you have the account’s own analytics, you can add shares and saves, as long as you do the same for every account you compare.

YouTube

The YouTube Data API’s public statistics for a video are its views, likes and comment count. YouTube made dislike counts private in November 2021; creators still see their exact dislikes in YouTube Studio, and the YouTube Data API only returns them to the video’s owner. Shares are in YouTube Analytics only. YouTube’s Help Center says its engagement metrics (views, likes, dislikes and subscriptions) are checked for legitimacy, and that counts can be slowed, frozen or adjusted while that happens.

From the outside, then, a YouTube engagement rate is (likes + comments) ÷ subscribers or ÷ views. From inside Studio, you can also divide by engaged views, which YouTube defines as the times viewers stayed to watch past the initial seconds, not including loops.

Instagram

Instagram has moved from impressions to views. In Meta’s Instagram media insights documentation, the old impressions metric is deprecated for media created after July 2, 2024, and views is listed for posts, reels and stories. The same page lists reach, likes, comments, saves, shares, reposts and total interactions, and notes two details that affect your maths. Likes, comments, views and total interactions count organic activity only; interactions on ads are excluded. And data can be delayed by up to 48 hours, so don’t calculate a rate on a post published yesterday.

From the outside, the only interactions you can count are likes and comments, and some creators hide like counts. Iqfluence notes that accounts hiding likes stay in its dataset, which slightly lowers its Instagram figures.

TikTok

TikTok shows more publicly than the other two. Its Research API, which serves public data, returns each video’s views, likes, comments, shares and favorites (saves to a viewer’s Favorites). Inside TikTok Studio, creators get account and post analytics, including viewer and follower insights. We couldn’t find a TikTok help page that defines how long someone must watch for a view to count, so treat TikTok views as a platform-specific unit.

Tools don’t agree on what to count here either. Socialinsider’s 2026 benchmarks count likes and comments on Instagram but likes, comments, shares and saves on TikTok, so even one report’s numbers aren’t like for like across platforms.

How much can the formula change the result?

A lot. Take one hypothetical short video from an account with 50,000 followers. It gets 1,000 likes, 60 comments, 90 shares and 250 saves: 1,400 interactions in total. The platform says it reached 20,000 accounts, was viewed 32,000 times and was shown 70,000 times.

Same post, same audience, and the answer runs from 2.00% to 7.00%. None is wrong. A media kit that says “7% engagement” without naming the formula invites a brand to compare it with someone else’s 2% by followers, which is meaningless.

Viral posts make the gap wider. Say the next video reaches 400,000 accounts, mostly people who don’t follow the creator, and gets 12,000 interactions. By followers that’s 24%, which looks extraordinary. By reach it’s 3%, which is ordinary for a post seen mostly by strangers. Both numbers are true, and only one describes the content.

How do you calculate engagement rate in Google Sheets or Excel?

Put one post per row, newest first, with the follower count at the time of posting rather than today’s. This layout works in both Google Sheets and Excel:

Columns: A Date | B Likes | C Comments | D Shares | E Saves | F Followers | G Reach | H Views

I2  Engagements          =SUM(B2:E2)
J2  ER by followers      =I2/F2
K2  ER by reach          =IFERROR(I2/G2,"")
L2  ER by views          =IFERROR(I2/H2,"")

Fill row 2 down for every post, then format J:L as percentages.

Account-level summary (last 10 posts in rows 2 to 11):
Median ER by followers   =MEDIAN(J2:J11)
Mean ER by followers     =AVERAGE(J2:J11)
Trimmed mean (drops 20%) =TRIMMEAN(J2:J11,0.2)
Total-based ER by reach  =SUM(I2:I11)/SUM(G2:G11)

Three notes on the recipe. Percentage formatting displays the value multiplied by 100, so don’t multiply again. The trimmed mean drops the top and bottom 10% of values before averaging, a middle ground if you dislike medians. And the last line, total interactions divided by total reach, weights big posts more heavily than the median does; it’s a fair campaign figure, but say which one you used.

If you only have public data, delete the shares, saves and reach columns and calculate (likes + comments) ÷ followers. That’s the formula Iqfluence uses, and the one Dash Social and Socialinsider use for Instagram, so your result will be comparable with those benchmarks.

Why do engagement rate calculators give different numbers?

Most disagreements come from one of these choices, and a good tool tells you which it made:

1. The denominator. Followers, reach, views or impressions, as the worked example shows.
2. The interactions. Likes and comments only, or shares and saves too.
3. The posts. The last 10, the last 12, the last 30 days, or only one format.
4. Mean or median. A single viral post can double the mean.
5. The data source. Public counts, or the account’s own analytics, which can include shares, saves and reach that outsiders can’t see.
6. Organic or paid. Meta’s Instagram API excludes interactions on ads from its standard likes, comments and views metrics.
7. The view definition. On YouTube it changed on August 24, 2026, for every tool at once.

The same applies when you check a creator before a deal. Our guide on how to vet a YouTube creator recommends picking one tool and using it for every creator you compare.

What is a good engagement rate on Instagram, TikTok and YouTube?

It depends on the formula, the platform and the account’s size, so the honest answer is a range from a named study. These are the dated studies we could verify, with their methods:

Read the numbers within their rows, not across them. Look at YouTube channels with a few thousand subscribers: Iqfluence’s study puts the median at 0.74% for 1K–5K subscribers, while CreatorDB puts it at 3.77% for 1K–10K. The audiences didn’t behave five times differently. Iqfluence divides by all subscribers, whether or not they watched; CreatorDB divides by views, which only counts plays. Dash Social’s report makes the same point on one set of accounts: 0.4% by followers and 1.9% by views.

Brands and creators don’t compare either. Quid’s 0.30% for Instagram comes from company accounts; Iqfluence’s Instagram creator medians are higher in every size band, from 0.70% to 3.10%. A creator should be judged against creators.

What holds in every dataset is the shape: smaller accounts engage more per follower. Iqfluence’s YouTube median falls or holds at every step from 1K–5K to 5M+, and CreatorDB’s June report finds the same decline. That’s why benchmarks are only useful within a size band, which our guide to nano, micro, macro and mega influencers breaks down. If a number has no method, no date and no sample, leave it out, as we did with several widely repeated “average engagement rate” figures we couldn’t trace to a source.

How did YouTube’s August 2026 view change affect engagement rates?

Since August 24, 2026, YouTube counts a view the moment a video starts to play, across Shorts, long-form videos and live streams. YouTube’s Help Center says this doesn’t change Partner Program earnings, which are still based on engaged views and engaged watch hours. The YouTube Data API’s view count changed on the same date, so every third-party tool that reads public views changed with it. YouTube’s announcement, quoted in CreatorDB’s study, told creators they “will likely see their total view counts increase faster moving forward.”

More views with the same likes and comments means a lower engagement rate by views. Say a long-form video got 10,000 views, 400 likes and 50 comments under the old count: 4.5%. If the same audience now registers 13,000 views because quick bounces count, the rate falls to 3.5% though nothing about the video changed. Those numbers are hypothetical.

CreatorDB’s August 2026 study gives a sense of scale. Measured six days before the switch, the median long-form engagement rate was 3.67%, against 2.59% for Shorts, which already counted views from the first frame. Within the same channels, the median Shorts rate was 78% of the long-form rate. CreatorDB is careful to say the gap reflects format differences too, not counting alone; likes per view were nearly identical, and comments drove the difference. It planned to re-measure the same channels after the switch; we couldn’t find that follow-up on its research page as of October 12, 2026.

Three rules follow. Don’t compare a YouTube engagement rate by views from before August 24 with one from after it. Label any figure in a media kit or report with its measurement date. And if you have Studio access, divide by engaged views, which keep the stricter standard.

What inflates or deflates engagement rate?

Check for these before you trust a number; some are innocent and some aren’t:

• Giveaways. “Comment to enter” posts pile up comments from people who came for the prize. Leave giveaway posts out of the average, or calculate it with and without them.
• Engagement pods. Groups of creators who agree to like and comment on each other’s posts raise engagement without adding customers. The tell is the same handful of accounts commenting on everything, quickly.
• Viral outliers. One post seen by millions of non-followers makes the follower-based rate meaningless for that post. Use the median.
• Bought likes and comments. YouTube’s fake engagement policy bans anything that artificially increases views, likes or comments, and its help pages say like and dislike counts can change as invalid ones are removed. A sudden drop in likes on old posts can be a clean-up, not a collapse.
• Hidden like counts. On Instagram, a creator who hides likes will look less engaged to every outside tool.
• Paid boosts. Promoted posts reach people a creator’s organic audience never would. Ask whether a post was boosted before comparing it with organic ones.
• Follower count drift. Dividing an old post’s interactions by today’s follower count understates past performance for a growing account. Use the count at the time of posting where you can.

What to do next

If you’re a creator, put your median engagement rate into your media kit with its formula written next to it; our guide on what to put in an influencer media kit shows where it goes. Engagement also shapes what you can charge, which we cover in how much to charge for a sponsored video. If you’re a brand, use the median-of-recent-posts method from this guide to find micro-influencers whose audiences actually respond, then run the rest of the checks in our vetting guide before you sign anything. And whichever side you’re on, keep a dated record: from August 2026 onwards, a YouTube engagement rate by views without a date is a number without a unit.