Below are twelve complete ad scripts, written out line by line at 6, 15 and 30 seconds, each with a note on the structural choice it makes and the variable it is testing. They are not taglines. Every one has a hook, a stated offer before the halfway point, some form of proof and a call to action that names an action, because a script missing any of those cannot be diagnosed when it underperforms. Read them as a menu of structures to steal, not lines to copy.
Format below: the script as spoken or captioned, then what it is doing. Brand names are invented.
| # | Length | Structure | Variable it tests |
|---|---|---|---|
| 1 | 6s | Price reveal | Price-led open |
| 2 | 6s | Objection first | Blocker named in hook |
| 3 | 6s | Captions only | Silent-first construction |
| 4 | 15s | Problem-solution | Recognisable versus staged problem |
| 5 | 15s | Founder direct | Founder face versus user-generated |
| 6 | 15s | Testimonial | Outcome number versus habit change |
| 7 | 15s | Demo, one shot | Mechanism shown versus benefit claimed |
| 8 | 15s | Comparison | Named category versus named rival |
| 9 | 15s | Qualifying question | Audience filtering in the hook |
| 10 | 15s | Listicle | Enumeration versus narrative |
| 11 | 30s | Case study | Named metric versus client outcome |
| 12 | 30s | Unboxing to objections | Shipping objection in the proof slot |
Short scripts: 6 seconds
At six seconds you get one idea. Not one idea plus a brand line.
1. Price reveal, 6s, DTC (testing: price-led open)
"Forty-two pounds of skincare, replaced by one bottle at twenty-two. Tap through, first one ships free."
One sentence of arithmetic and one instruction. There is no proof slot at this length, so the offer has to be self-evidently good. Runs as a retargeting cut of a longer ad.
2. Objection first, 6s, SaaS (testing: naming the blocker in the hook)
"You think migration takes a quarter. Two ledgers, one afternoon, no card. ledgerly.com."
The hook is the objection, said before the viewer can raise it. Best used on audiences who already visited the pricing page.
3. Single visual claim, 6s, hardware (testing: silent-first construction)
Captions only, no voiceover. "SAME BAG." / "SIX DAYS OF CLOTHES." / "PACKS IN 40 SECONDS." / "SHOP NOW."
Written for muted playback, four words maximum per card. If it works silently it works voiced, and this is the cheapest way to find out.
Mid scripts: 15 seconds
The workhorse. Hook, stakes, offer by second 7, proof, CTA.
4. Problem-solution, 15s, home goods (testing: recognisable versus staged problem)
"Every mug in this cupboard is a different height." / "So nothing stacks, and the door never shuts properly." / "Ours nest. Four for thirty-two pounds." / "Cupboard door shut on the first try, first time in two years." / "Order before Thursday for weekend delivery."
The problem is mundane and visible, which is the whole trick. Nobody spills coffee in slow motion. Proof is the cupboard door, filmed once.
5. Founder direct, 15s, DTC (testing: founder face against user-generated)
"I priced this wrong for eleven months." / "We were charging thirty-eight and losing money on shipping." / "It is twenty-nine now, shipping included, because the volume finally works." / "Same formula, same factory, nothing else changed." / "Link in the profile if you have been waiting."
Confession as hook. It buys the credibility the price claim then spends. Works once per account.
6. Testimonial, 15s, fitness app (testing: outcome number versus habit change)
"I have started and quit six training apps." / "This one just told me what to do on Tuesday." / "Eleven weeks in, still going. It is free for the first month." / "I run three times a week now. I did not before." / "Search Cadence in the App Store."
The proof is the behaviour change, not a transformation figure. It is more believable and it does not need a before photo.
7. Demo, 15s, tool (testing: showing the mechanism versus claiming the benefit)
Voiceover over an unbroken screen recording: "Paste the two exports." / "That is the whole reconciliation." / "Free for two ledgers." / "Nine days of close became three." / "Two busiest ledgers, ledgerly.com, no call."
One continuous shot, no cuts, because the cut is what people suspect. If a demo needs editing to look fast, it is not fast yet.
8. Comparison, 15s, subscription (testing: named category versus named competitor)
"A box of six coffees is fourteen pounds. A daily flat white is twelve a week." / "That is the whole pitch." / "Six bags, fourteen pounds, cancel any time." / "Nobody in this office has been to the cafe since March." / "First box half price this week."
Compares against the habit rather than a rival brand, which keeps it running and keeps legal quiet.
9. Question hook that qualifies, 15s, B2B (testing: audience filtering in the hook)
"Do you know your blended CAC by channel this month, or last quarter's?" / "If it is last quarter's, you are buying media blind." / "Sightline reads both ad accounts and your Stripe data. Free while you connect them." / "Two hundred teams closed their spreadsheet version." / "Connect one account at sightline.io."
A question only works as a hook when the wrong audience cannot answer it. Filtering costs hook rate and improves cost per lead, a trade covered in how hook, body and CTA get measured separately.
10. Listicle, 15s, apparel (testing: enumerated structure against narrative)
"Three things I check before buying trousers online." / "Inseam listed in centimetres. Real waist measurement. Return window over thirty days." / "These are the only ones that passed. Sixty-eight pounds." / "Third pair I have bought." / "Size guide is in the link."
Enumeration gives the viewer a reason to hold on to the end. It also gives you three obvious variations: change the number, change the criteria, change the vertical.
Long scripts: 30 seconds
Extra time goes to proof. It does not go to the hook.
11. Case study, 30s, agency service (testing: named metric against named client outcome)
"We were shipping four ad variations a month and wondering why performance kept sliding." / "Creative was not the bottleneck. Output was." / "Now it is thirty a month, from footage we already had, and the winners last." / "Same team, same budget, no extra shoot days." / "Our first month of this paid for the year." / "Book a look at the process, thirty minutes, no deck."
Confession, diagnosis, change, result, ask. The middle line does the persuading. The supporting figure stays in the brief rather than the voiceover: our 2026 fatigue benchmark puts brands shipping 15 to 50 variants a month at 3 to 5 times the campaign lifespan of quarterly refreshers.
12. Unboxing to objection handling, 30s, DTC furniture (testing: shipping objection in the proof slot)
"This is the box a two-metre shelf arrives in." / "I assumed a courier would leave it in the rain and I would be assembling it for a weekend." / "Ninety pounds, delivered to the room you name, two-person delivery included." / "Eleven minutes to assemble, one tool, in the box." / "It has held forty-two books and a record player since January." / "Thirty-day returns and they collect. Shop the shelf."
Every line after the offer answers a different reason not to buy: delivery, assembly, durability, risk. That is what a 30 second script is for.
Turning twelve scripts into a test
Notice how few variables differ between these. Nine of the twelve place the offer between seconds 5 and 8, and the differences sit in the hook and the proof. Hold the offer steady, vary one slot, and a batch of 8 to 20 live variations teaches you something reusable rather than twenty anecdotes.
Genyad, our product, builds these from footage you already own: upload the library once, and each variation is a fresh script plus its own shot selection, voiceover, captions and export rather than a re-cut of one timeline. The AI ad script generator writes into the structures above, and the hook variation generator rewrites the first three seconds while the rest holds. We have no AI avatars or synthetic presenters, so scripts 5, 6 and 12 need a real person on camera. If you do not have one, an avatar tool suits you better than we do.
Frequently asked questions
How long should an ad script be?
Match the script to the placement: 6 seconds for retargeting, 15 seconds as the feed and Reels default, 30 seconds when you have proof worth showing. Fifteen seconds is roughly 35 to 40 words of English voiceover.
What makes an ad script example worth copying?
That you can tell which line does which job. If you cannot point at the hook, the offer, the proof and the ask, you have copy rather than a script, and when it fails you will have no idea which part to change.
Should the offer be at the start or the end of the script?
Before the halfway point, then repeated in the call to action. Nine of the twelve scripts above state price or terms by second 8, because the viewer who leaves at second 9 should still have heard the trade you are offering.
How many of these should I run at once?
Pick one structure, build four to six versions with different hooks, and run them as a batch of 8 to 20 live variations. Refresh weekly on TikTok, fortnightly on Meta feed.