How hooks earn the next three seconds

What the hook must do#

The hook has one job: earn the next three seconds. It does that by showing who this is for, what problem it solves, and which payoff direction the watch promises. Brand mark visible, human face or product in frame, payoff stated plainly. Teases and slow logos lose thumbs that plain promises keep. Start at the Meta Ads hub for the family view, and read Meta own diversification research for why varied openings matter.

Sound off is the default test: if the muted opening confuses, captions merely decorate confusion. Our captions guide carries the legibility rules, and video lengths set the structure the hook opens.

Five opener types that repeat#

Five structures repeat across winning accounts: the payoff first (result in frame one, method after), the problem callout (name the pain the viewer recognises), the proof face (customer states the outcome), the pattern interrupt (unexpected visual that still relates), and the offer plainly (price, trial or bundle stated, not teased). Each suits different briefs, and each needs captions from frame one.

Rotate types across concepts rather than repeating one winner until fatigue: the auction rewards range. Diversification shows the evidence, and UGC versus studio helps pick the face for each type.

Writing hooks that survive sound off#

Write the opening as three short caption lines: line one names the viewer or problem, line two states the payoff, line three points to the proof coming. Keep each line under about six words, sized for phones, high contrast, clear of Reels interface. Burn captions in or upload en_GB SRT so nothing depends on audio. Meta own video specs list caption support per placement.

Read every opening muted at arm's length: viewer, payoff and next step must all land without volume. Safe areas decide where those lines can live.

Testing hooks before concepts#

Cut three openings per concept with identical middles and endings, then read thumb stop rate and 3 second views first. Openings decide, endings merely confirm, so promote the winning hook before judging the story. Kill openings below the account median ruthlessly and log each hook with its rate in a shared sheet.

Hold the middle constant during hook tests so the read stays clean; changing story and opening together teaches nothing. Our A/B guide sets the clean read rules, and the 3-2-2 method shows the launch structure.

Hook mistakes that cost attention#

Slow brand idents, lifestyle montages with no promise, jokes that need context, and offers hidden until second twelve all spend budget teaching the auction to skip you. A second failure is testing one hook per concept, which confuses a weak opening with a weak idea and kills concepts that deserved a better first frame.

Refresh hooks on the schedule in our fatigue guide, and ask our team for a hook audit when thumb stop rate trails the account median for two weeks running. Keep a hook library with frames, lines and rates so each new brief starts from evidence rather than memory.

Brief creators with winning frames, not adjectives: show the three frames, the caption lines and the rate each earned. Concrete references beat mood words every time. Archive winning openings with project files so autumn rematches reuse spring evidence.

Logging hooks into a library#

Every hook test should leave a library row behind: concept name, opening frames, caption lines, spend, thumb stop rate and 3 second view rate. Over a quarter this sheet becomes the most honest brief in the business, because it records what thumbs actually paused for rather than what the room liked. Review it monthly and brief the next round from its top quartile.

Share the library with creators and media buyers alike so both sides plan from the same evidence. Retire the bottom quartile of hook types each quarter, double down on the top two types with fresh scripts, and protect novelty by testing one unfamiliar opener per round. The method in diversification explains why range plus memory compounds, and A/B discipline keeps each new entry trustworthy.

Frequently asked questions#