The 5-beat structure of a direct-response video ad

"I have been drinking this skinny little can every day for a week and my pants fit again."

That is a real opening line from a scaled OLIPOP UGC ad on Meta. No logo. No music swell. No "hey guys." Just a claim, a hand holding the can, and a face that knows exactly what it is doing. By second three you have either swiped or you have not, and the rest of the ad is built around the assumption that you have not.

Direct-response video is not storytelling. It is a sales call compressed into thirty seconds, and like a sales call it has a script. Five beats. Each one has a job, a time budget, and a way it usually fails.

Here is the structure. Then the contrarian take. Then the annotated example.

The 5 beats

Beat 1 — Hook (0–3s)

Function: Stop the scroll. Earn the next two seconds.

Time budget: 3 seconds, hard cap. Motion's creative analytics platform tracks "hook rate" — 3-second views over impressions — as the single most-used input to creative iteration, with 30–40% as the baseline for a strong ad.

Common failure mode: Branded intro card. Logo first. "Are you tired of…" Anything that signals "this is an ad" before the claim lands. Strong creative teams literally tag ads by hook type to find which patterns are unique enough to break the pattern interrupt — see this hook-stacking writeup on weaving multiple hooks through a UGC script.

Real example: Liquid Death's "Don't Be Scared, It's Just Water" Super Bowl spot opens on children chugging tallboys at a party scored to Judas Priest. You cannot place the ad genre. You watch.

Beat 2 — Problem (3–7s)

Function: Name the specific pain. Make the viewer feel seen.

Time budget: 4 seconds. You are not educating; you are confirming.

Common failure mode: Too generic. "Bloating is uncomfortable" is a billboard. "I literally couldn't button my jeans by 4pm every day" is an ad. Specificity is the whole game here.

Real example: A Saie sunscreen ad: "Every other sunscreen leaves a white cast on my skin and makes my makeup pill by lunch." Two pains, one sentence, named at a level of detail only an actual user would say.

Beat 3 — Solution (7–15s)

Function: Introduce the product as the specific mechanism that resolves the named problem.

Time budget: 8 seconds — the longest beat besides proof.

Common failure mode: Feature dumping. Listing what the product is instead of what it does to the problem. This is where Hormozi's framing in $100M Leads is sharp: stop pitching the category, pitch the specific mechanism. Not "loans" — "we approve based on revenue, not credit score, in 48 hours." Not "razor" — "ceramic blade that doesn't get hot against your balls."

Real example: Manscaped's Lawn Mower 4.0 ads do not say "trimmer." They say "SkinSafe ceramic blades" and demonstrate the no-nick claim with the actual product moving against skin. Mechanism, not category.

Beat 4 — Proof (15–25s)

Function: Make the claim believable to a skeptic.

Time budget: 10 seconds. The longest beat in the ad. There is a reason.

Common failure mode: Skipping it, or replacing it with a logo wall. "As seen in Forbes" is not proof. Proof is before/after, screen-recorded review counts, a side-by-side, a sales ranking, a stitch of three different creators saying the same thing unprompted.

Real example: OLIPOP's best-performing creative on Meta in 2025 was a creator stitch — three different women, three different kitchens, all opening their fridge to a row of OLIPOP cans. No script alignment, just visual repetition. That is proof at the format level.

Beat 5 — CTA (25–30s)

Function: Tell the viewer the specific next click.

Time budget: 5 seconds.

Common failure mode: Vague. "Check it out." "Link in bio." A real CTA names the page, the offer, and the urgency: "Use code SUMMER for 20% off your first 4-pack at drinkolipop.com — first-time customers only."

The contrarian: the proof beat is doing 60% of the conversion work and most ads skip it

Hooks get all the attention because they are easy to measure. Hook rate, thumbstop rate, three-second view — Motion's whole creative reporting product is built around them. So media buyers iterate on openings forever and ship ads where beats four and five are an afterthought.

This is upside-down. The hook decides whether the ad gets watched. The proof decides whether the watcher converts. In every direct-response account I have seen tear-downs from — including the LeadEnforce breakdowns and the UGC frameworks Foreplay teams reverse-engineer from saved swipe files — the gap between a 1x and a 3x ROAS creative is almost never the hook. Two ads with the same hook rate routinely produce 2–3x different CPAs, and the variable is whether beat four landed.

The CTA is overrated for the same reason the proof beat is underrated. People who reach second 25 of a vertical video have already decided. The CTA is a tiebreaker, not a closer.

If you have one hour to improve an ad, do not write five new hooks. Reshoot the proof.

Annotated example: a 30-second Saie ad

0:00–0:03  HOOK     "I'm 34 and people keep asking if I'm 25. It's the SPF."
                    [No logo. Direct-to-camera. Tight crop.]

0:03–0:07  PROBLEM  "Every sunscreen I tried left a white cast or pilled
                    under my foundation by lunchtime."
                    [Specific. Two named pains, not one general one.]

0:07–0:15  SOLUTION "Saie's Sunvisor is a tinted mineral SPF — the tint
                    cancels the white cast, the formula is gel-based so it
                    doesn't pill."
                    [Mechanism, not category. Names the *why* it solves
                    each named pain.]

0:15–0:25  PROOF    [Screen-rec of Sephora reviews scrolling — 4.6 stars,
                    "12,000+ reviews."]
                    [Cut to creator's own face, no makeup, then with
                    foundation over the SPF — no pilling, no cast.]
                    [Cut to a stitch of two other creators saying the same
                    thing in their own words.]
                    
0:25–0:30  CTA      "Sunvisor at sephora.com — use SAIE15 for 15% off
                    your first one."
                    [Named site. Named code. Named offer. Named order.]

Notice that the proof beat is one-third of the runtime and contains three different proof types — review volume, before/after demonstration, and creator stitch. That stack is doing the conversion work. The hook just bought the right to show it.

One specific takeaway

Open Motion or Meta Ads Manager. Sort your top creative by 3-second view rate and your bottom creative by purchase ROAS. The ads in the overlap — high hook rate, low ROAS — are the ones with broken proof beats. Reshoot beat four on those three ads this week. That is your highest-leverage hour.