Hook Rate Benchmarks by Vertical: What 'Good' Actually Looks Like in 2026

The definition problem, first

Hook rate is three-second views divided by impressions. The trouble is that platforms count a three-second view differently, some sources use two seconds, some use ThruPlay, and almost nobody says which. A benchmark quoted without its definition is close to useless.

So treat any number here — including ours — as a range for orientation, and benchmark yourself against your own account's history rather than against a blog post.

The ranges worth working from

• Below 20% on cold traffic: the creative isn't working. Not the targeting, not the offer — the first three seconds.
• 20–25%: functional, unremarkable. Most accounts live here.
• 25–35%: good. This is what a well-written hook on a well-matched audience produces.
• 35–45%: strong. Sustained performance in this band usually means the hook is doing something structurally different from the category norm.
• Above 45%: check your definition and your audience. Warm traffic and retargeting inflate this substantially.

Billo publishes a fuller benchmark set with their own methodology stated (Billo), which is worth reading alongside this.

Why it moves by vertical

Three factors, roughly in order of impact:

• Problem visibility. Categories with an immediately legible problem — cleaning, storage, pain — hook higher, because the first frame can state it.
• Category saturation. Skincare and supplements have trained viewers to recognise and skip the conventions. Hooks that would be strong elsewhere read as ads there.
• Purchase consideration. High-consideration products hook lower and convert better on the traffic they do hold. A low hook rate with a strong hold rate is a different situation from a low hook rate with a weak one.

What to do with a bad number

Hook rate is the cheapest metric to fix because it's the most isolated. You can change the first three seconds without touching the rest of the asset, and you can test three openings against one body for the cost of a re-render.

That's the practical technique in how to write a hook that survives three seconds.

A note on where these numbers come from

The ranges above are synthesised from published third-party benchmarks, not from our own client data. When we have enough of our own to publish something defensible by vertical, we'll publish it and say how it was measured. Until then we'd rather cite other people's work than invent authority — the same reason there are no ROAS figures on our pricing page.