Ad creative hooks that stop the scroll

A hook is not an angle

An angle is the argument the ad makes. It answers why this particular person should care about the product at all. A hook is the opening move that gets the angle looked at: the first frame, the first line, the image that makes a thumb slow down. One angle can carry a dozen hooks.

Mixing the two wrecks testing. Angle tests are slow and expensive because you are validating whether demand exists. Hook tests are cheap because the demand is proven and you are only changing how it gets delivered. Run both at once and you get a winner you cannot repeat.

• A claim is what the product does: the dryer cuts drying time to six minutes.
• An angle is why that matters to this reader: you are late every morning because of your hair.
• A hook is what appears first: a clock reading 8:41 and someone still holding a towel.
• One angle should produce five to ten hooks before you retire it.
• If two ads differ in angle and hook at once, the result tells you nothing about either.

What the first three seconds measure

Hook rate, also called thumb stop ratio, is three second video views divided by impressions. It isolates the moment before anything else in the ad has a chance to matter.

Hold rate is the follow up metric: ThruPlays divided by three second views. It tells you whether the ad kept the people the hook won. Read together, they turn a vague "this ad underperformed" into something you can act on.

Short exposure is not wasted exposure. Nielsen research found that video impressions under two seconds still accounted for 38 percent of brand recall, 23 percent of brand awareness and 25 percent of purchase intent. Someone who does not stop can still register the brand, provided the frame they saw carried it.

• On Meta, hook rates above roughly 30 percent are treated as good and anything under 20 percent as a problem.
• A healthy hold rate floor is roughly 15 percent on Meta and 20 percent on TikTok.
• A low hook rate with a healthy hold rate means the ad is fine and the opening is not.
• A high hook rate with a collapsing hold rate means the opening promised something the ad never delivered.

Visual hooks on a static ad

A static ad does not get three seconds. It gets one glance, at whatever size the feed renders it, which is why static hooks come down to contrast and focal point rather than cleverness.

The practical test is size. Shrink the creative to about 150 pixels wide and look at it. If the subject blends into the background, or the eye has two things to land on instead of one, there is no hook at that size.

Then there is text. Meta dropped the hard 20 percent image text rule in 2020, so ads are no longer rejected over it, but Meta still recommends keeping image text under about 20 percent because those images tend to perform better. Six to ten words is usually the ceiling before a hook line stops being read as part of the image.

• Build one focal point and check that it survives at 150 pixels wide.
• Put the subject against a background of a different tone, since matching tones make products disappear at thumbnail size.
• Keep the hook line to roughly six to ten words.
• Set the reading order deliberately: focal point, then headline, then call to action, then brand mark.

Hook patterns worth keeping in rotation

Hooks are patterns rather than one off ideas. TikTok's creative guidance tells advertisers to land the hook, the key message or the strongest visual inside the first three seconds, and the openers that hit that mark fall into a small number of shapes.

The pattern interrupt is the most used and the most abused of them. An unexpected visual will stop a scroll on cold traffic, but if it has nothing to do with what you sell, hold rate drops and you have paid for attention you cannot use.

• A pattern interrupt breaks the visual rhythm of the surrounding feed with an image nobody expects to see there.
• A curiosity gap opens with a question the rest of the ad answers, which only works if the answer arrives fast.
• A direct callout names the viewer's situation in the first line and trades reach for relevance.
• Opening mid problem, before the product appears, usually holds better than opening on the product.
• Leading with price or a guarantee works on warm audiences who already know what they are buying.

Where the hook sits on the screen

A hook placed outside the visible area does not get seen. On TikTok the working canvas is 1080 by 1920 with roughly 120 pixels kept clear on each edge, because the interface covers the rest with usernames, captions and buttons.

The same asset also behaves differently across placements. A frame that reads well in feed loses its top and bottom in Stories, and a hook parked in a corner is the first thing a crop takes. Putting it in the middle third avoids that.

• Keep hook text roughly 120 pixels clear of each edge on vertical placements.
• Put the hook in the middle third, where every crop keeps it.
• Check the creative in the placement preview before launch rather than after.
• Export the idea once per ratio instead of letting the platform crop a single master.

Testing hooks without burning the month

Test two to five distinct angles in a cycle, find the one that earns attention, then generate hooks underneath it. Testing five hooks of the same angle in week one tells you which frame is prettiest and nothing about whether the argument works.

Volume is where this usually breaks down. A real hook test needs the same angle rendered several ways in every ratio the account buys, which is more production than most teams can absorb by hand. That is the step Aduarius is built for: write the brief once, pick a style, and get the variants back as square, vertical and horizontal creatives with the hook line already on the image.

Refresh on the metric rather than the calendar. When hook rate starts sliding on an ad that used to perform, the opening has been seen too often by the same people, and raising budget will not fix it.

• Keep everything except the hook identical when you test hooks.
• Watch hook rate per ad, not per campaign, because averages hide the ad doing the work.
• Retire a hook when its hook rate declines, not on a fixed schedule.