← Writing

Most Instagram hook advice is made up

26 July 2026 · Joonas Soininen

Hooks are the most written-about thing in short-form video and the worst-sourced. We went looking for the origin of the numbers everyone repeats. Most do not have one.

The numbers that are fake

"You have 1.7 seconds." This one is real, and completely misapplied. It comes from a Facebook IQ study published in April 2016, which measured average dwell time across 850+ video ads in the 2016 Facebook News Feed. That is a scrolling mixed feed, four years before Reels existed, and it is a descriptive average that includes instant scroll-pasts. It is not a decision deadline in a full-screen autoplay feed, and it is now ten years old.

"Attention spans are 8 seconds, shorter than a goldfish." Fabricated. It traces to a footnote in a 2015 Microsoft Canada report, which cited Statistic Brain, which had no underlying study. The BBC took it apart in 2017.

"Hooks are 80% of performance." No study, no platform statement, no traceable origin at all.

"Get your 3-second hold rate above 60% and you get 5-10x the reach." This traces to a single blog post by a video-clipping vendor. Its entire evidence is the sentence "I've analyzed performance data across multiple accounts, and the correlation is undeniable." No dataset, no sample size, no method. Everything downstream just repeats it.

"The algorithm tests your video on 200-500 users first." No primary source anywhere. TikTok's published description of its recommender has no staged rollout. Meta's engineering blogs describe a candidate-ranking funnel, which narrows candidates per request, not audiences over time. Those are different things, and the folklore is a garbling of the second into the first.

What Instagram has actually said

Short list, because it is short.

In January 2025 Adam Mosseri said the top three ranking signals are watch time, likes and sends, and advised watching average watch time, likes per reach, and sends per reach. Hook performance is not among them.

In October 2024 he said the first three seconds are "super important" and to "try experimenting with different types of hooks to see what works best for you." Note what that is: craft advice with no data attached, and an explicit refusal to name a winning hook type. Every listicle ranking hook formulas is going further than the head of Instagram will.

Meta also demotes clickbait, which its own policy defines as "creating misleading expectations." A hook that overpromises is, by Meta's published definition, demotable. That is the part hook-maximising advice never mentions.

The metric nobody is using

Here is the useful part. In August 2025 Instagram shipped a per-Reel retention chart and a metric called Skip Rate, defined in Meta's API reference as "the percentage of views from people who skipped during the first 3 seconds of the reel."

So the thing all that advice was gesturing at now exists. Two catches. It is inverted: hold rate is 100 minus skip rate, so "above 60% hold" means "below 40% skip." And Instagram's own Help Center still does not document it, which is presumably why most advice still references ad metrics like ThruPlay and 3-second video plays, which organic creators never had.

Check your own Skip Rate before you accept anyone's threshold. You have the number now. Nobody has published a credible benchmark for it, including us.

The uncomfortable finding

Researchers at CHI 2026 built a model to predict whether someone stays or swipes, trained on millions of real viewing decisions, with access to the entire video. It tops out at roughly 70% accuracy, and removing the video's content features barely hurts it.

Read that again: knowing everything about the video gets you a little better than a coin flip plus 20 points. Most of what decides whether someone watches is not in your creative. It is who they are, what they just watched, and what the feed decided to do with you.

There is a matching result from marketing research: heavy branding early in an ad measurably increases avoidance. Trying too hard at the front can cost you.

What to do instead

Keep making a strong opening. It obviously matters, and Instagram says so. Just stop treating it as the whole game, and stop optimising against numbers nobody can source.

The signals Instagram named are watch time and sends. Both are decided by the middle and the end of your video, not the first frame. A great hook on a video with nothing behind it produces exactly what the platform punishes: a misleading expectation.

Practically: look at your Skip Rate to find out whether you have an opening problem at all. If skip rate is fine and watch time still collapses, your hook was never the issue, and every hour spent rewriting it is wasted.

That is roughly what Silma does before you post. It reads the whole video, not just the first three seconds, and tells you where it will lose people. Free plan is 25 credits a month, no card. Start free.

Sources