Engagement rate benchmarks for short-form video — and why published numbers disagree

Every quarter a fresh set of benchmarks circulates, and they contradict each other. One 2026 aggregation puts TikTok at 2.50–4.64%, Instagram Reels at 0.50–1.48% and YouTube Shorts at 3.4–5.91%. Another reports a TikTok median of 5.52% against Instagram's 2.34% and YouTube's 2.66%. A third finds Reels averaging 3.8% while static Instagram posts sit at 1.2%.

They are not all wrong. They are measuring different things.

The denominator is the whole argument

Engagement rate is a fraction, and the numerator is rarely the problem — likes, comments and shares are what they are. The denominator is where studies diverge:

A video that reaches ten times the creator's follower count will look spectacular per follower and ordinary per view. Neither reading is dishonest; quoting one against a benchmark built with the other is.

Three more reasons published numbers move

• Account size mix. Engagement per follower falls as accounts grow. Benchmarks in the 8–12% range are usually nano-creator samples; 1–2% figures usually come from accounts above a million followers. A "platform average" without a size breakdown is close to meaningless.
• View definitions changed. Instagram consolidated to a single views metric in 2025, and YouTube began counting every Shorts playback in March 2025. Per-view engagement rates computed before and after those changes are not on the same scale — the denominator grew.
• Niche. Entertainment, beauty and gaming behave very differently from B2B or finance. A cross-industry average sits in the middle of a wide distribution and describes almost nobody.

Build your own baseline instead

External benchmarks are useful for a sanity check and useless as a target. The number that actually drives decisions is your own, computed consistently:

1. 1Fix the formula. Engagements ÷ views, for the same window on every video. Write it in the report header.
2. 2Fix the window. Engagement earned in the first 7 or 14 days after publication, not lifetime. Lifetime favours old videos and hides recent decline.
3. 3Segment before averaging. By platform, by creator tier, by content format. One blended number across TikTok and Reels tells you nothing actionable.
4. 4Keep at least 20–30 videos per segment before treating a difference as real. Short-form performance is heavily skewed by a few outliers; a segment of five is a rumour.
5. 5Re-baseline quarterly. Platform distribution changes, and so does your content mix.

What a usable baseline looks like

After a quarter of consistent measurement you should be able to say something like: "On TikTok, our creator videos earn a median 4.1% engagement per view in the first 14 days. Below 2.5% is underperformance; above 7% we look at what the hook did differently."

That sentence survives a meeting. "Industry average is 4.7%" does not, because the first question — measured how, on which accounts — has no answer.

Where engagement rate stops being useful

Engagement rate is a quality signal, not a business metric. A campaign can hit every engagement benchmark and sell nothing. Pair it with:

• Views earned in the period, for reach delivered against the brief.
• Cost per thousand views, for efficiency — see how to calculate CPM.
• Share of views in the first 48 hours, for whether distribution is still building or already spent.

And treat unusually high engagement with the same suspicion as unusually low: coordinated comment pods and purchased likes distort the numerator directly. The creator audit checklist covers what to look for.