What to do when your hook rate is below benchmark

01Is the benchmark you are missing a real number?

You are looking at one column, probably three-second video plays over impressions, sitting under a figure somebody told you was normal. Before you change a single frame: most circulating hook rate benchmarks are not measurements. They are one vendor's blog quoting another vendor's blog. We traced them in where hook rate and hold rate benchmarks actually came from, and the paper trail runs out fast.

Two things follow. Your own account's median over the last ninety days beats any published number, because it holds your product, your price and your audience still. And hook rate has a denominator you do not control - a broad cheap audience hands you impressions that were never going to stop for anything.

So build the baseline before you build a fix. Pull every video ad that ran in the last quarter, drop anything with fewer impressions than a normal day of delivery, and take the median. Now you have a number that belongs to your account. An ad in the bottom third of that list has a real problem. An ad two points under a figure you read somewhere may have nothing wrong with it.

02Triage in cost order, not in blame order

Six things can produce a low hook rate. They are not equally likely and they are wildly unequal in what they cost to change, so the order matters more than the list. Start at the top. Do not skip a step because a later one feels more like the real problem - that instinct is what produces a re-shoot that fixes nothing.

03What is actually wrong with your first three seconds?

There are four separable jobs in the opening, and a low hook rate means one of them is missing, not that the whole thing is bad. Sound at frame zero. A first frame that is a moment rather than a photograph. A first cut that arrives when the picture needs it. And one legible idea before frame twelve. Diagnose them one at a time and you will usually find three are fine. The full anatomy sits in how a short ad is actually built, shot by shot.

A hook rate does not measure how much your first frame moves. It measures whether a stranger understood something.

The finding that made us withdraw our own motion gate

We built a gate on frame-to-frame change in the first two seconds because it looked like the obvious hook metric. It ranked our accepted films below our rejected ones and was withdrawn, and the numbers are printed in where hook rate and hold rate benchmarks came from. What did the work was the row underneath: whether a proposition was legible early. Not motion. Information.

The check that catches all four at once

Before delivery we watch the whole cut cold, at 360 pixels, and answer five questions in writing. What question does this pose by second three. When is it answered. Does any earlier frame leak the answer. What does each beat do to the question. And what combined claim do the words and the picture make together. We call it the Stranger Test, and it takes about four minutes.

The last question is the one that saves money. A film of ours paired the line you will probably kill it with a shot of the product kit. Read the way a customer reads, words and picture made one claim, and the claim was do not buy this. It was killed for that, and the rule it produced is that fear aims at the status quo, never at the product experience.

04Polish is legible in the first second, and it can cost you

The uncomfortable finding, and the reason a bigger budget sometimes lowers a hook rate: a viewer classifies your frame as an advertisement before they read anything in it. Studio lighting, a clean sweep, a rack of product and a logo are all category signals, and the category they signal is the one the thumb is trained to skip.

The rest of that study is worth reading before you book anything: direct-to-camera was worth about +50% hooking power, and casting a celebrity took about 13% off it. The resolution we run on is a register split. Rawness belongs to the performance - handheld, unscripted, imperfectly framed. Polish belongs to the object. A properly photographed product inside a scruffy human moment is the combination we build toward. That is a reading of the VidMob casting data, not a measured comparison of the three.

05Should you fix this hook or replace it?

Three questions, answered on a phone with the sound on, get you to a specific change. The honest default is that you replace the opening rather than repair it: a hook is a mechanism, and mechanisms do not improve by degrees. What repairs well is everything around the mechanism: the caption, the in-point, the export, and the sound, which on a working split of ours is worth 40% of the ad and usually treated as 5%.

06Score the replacement before you build it

The expensive mistake is replacing one weak hook with another and finding out a week later. Pick the family the new mechanism belongs to, then tick the gates honestly. Cheaper still: build only the first 48 frames of three candidates, with their real first second of audio, export at 360 pixels and watch them on a phone among other content. A mechanism survives a stand-in, so you can test it before the film exists.

One caution about what a fixed hook buys you. Across 578,750 creatives and 6,015 advertiser accounts, Motion found roughly 5% of creatives become winners, and that the rate barely moves with account size. A better opening raises your odds on one swing. It does not change the fact that most swings miss, which is the argument in how many creatives your budget can actually teach you.

The fixes at the top of this list cost an afternoon. The fixes at the bottom cost a week, and the industry's standing instinct is to start at the bottom. If you only take one thing from the page: play it cold, on a phone, with the sound on, and listen to the first tenth of a second.