SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsad-copy-and-hook-lab

ad-copy-and-hook-lab

Design & creation

View repository

Updated 10-16 08:04

About this skill

Ad Copy and Hook Lab

Generates, screens and tests advertising copy from an approved brief for the creative lead and media buyer: offer first, proof early, one message, headline and hook mechanisms, and an ethics filter that removes classic direct-response tricks that are now unlawful or off-brand.

Use When

  • The concept is approved and the ads now need headlines, opening hooks, primary text, descriptions, calls to action or end-card lines.
  • We need a bank of headlines with the three strongest marked for testing.
  • The offer feels weak: build the promise, proof, sweetener, risk reversal and a genuine deadline before any copy is written.
  • Our current ads are not converting; diagnose the copy for a weak hook, a buried offer, missing proof or claims we cannot back.
  • Responsive search ad assets, lead-form wording or click-to-WhatsApp opening messages need writing.

Do Not Use When

  • caption-writer for organic captions and community posts.
  • creative-brief-and-big-idea when there is no approved brief or platform idea yet.
  • direct-response-funnel-copy for long-form sales letters, VSLs or funnel sequences.
  • Stop when a claim cannot be substantiated or a regulated category is unchecked; return the claim register to the legal/market release gate.

Required Inputs

Artefact Source/provider Required? If absent
Approved brief: audience, single message, action, mandatories creative-brief-and-big-idea or client Yes Stop and request it; do not write to a guessed audience.
Offer facts: price, terms, deadline, stock or capacity, guarantee Client commercial owner Yes for offer copy Write benefit-only variants and mark the offer lines not assessed.
Proof: testimonials with consent, results with source, certifications, reviews Client records, CRM, review platforms Conditional Use no proof lines; never invent or paraphrase testimonials into quotes.
Channel and format list with current character limits and specs Media plan; register claims AD-05 and AD-08 or live platform help Yes Write to the register limits where available; otherwise mark length "check current spec".
Voice-of-customer language Reviews, DMs, sales notes, listening Recommended Draft from the brief and flag the missing customer language.

Workflow

  1. Confirm the brief, audience awareness level (unaware, problem-aware, solution-aware, product-aware, most aware) and single action; stop if any is missing.
  2. Run the offer workshop (offer, proof and guarantee kit): immense promise, believability ceiling, proof, sweetener, risk reversal, real limited offer or scarcity with its reason.
  3. Collect the raw material: top benefit, hidden benefit, main fear, news angle, strongest proof number, customer phrases.
  4. Run the 60-minute headline sprint (headline and hook families): at least five lines per mechanism; then short-video hooks and first lines of primary text.
  5. Screen every candidate: pre-flight questions, standalone test, "dull to whom?", specificity, the direct-marketing ethics filter and anti-ai-slop. Withhold any line that fails the ethics filter; correct and rerun the screen.
  6. Fit copy to formats: per-format limits from the register (checked 2026-09-23) or a "check current spec" flag; proof in the first frame or first line; one CTA.
  7. Shortlist the top three per test cell and hand them to ad-testing-and-scaling with the hypothesis for each.
  8. Record claims and their evidence in the claim register; deliver the copy deck with approval status.

Outputs

Artefact Consumer Acceptance condition
Offer block (4–6 short paragraphs or bullet form) Client commercial owner, copywriter Promise, proof, sweetener, risk reversal and reason-why are all present and true
Headline and hook bank by mechanism Creative lead, media buyer At least five per mechanism; each screened; top three per cell marked
Format-fitted ad copy set Media buyer, designer Fits the stated limits or carries a "check current spec" flag; one CTA per ad
Claim register Approver, legal/market gate Every claim links to evidence and a date, or is removed
Test hand-off note ad-testing-and-scaling Hypothesis, variable and success metric per variant

Evidence Produced

Evidence Format Acceptance condition
Screening record Table: line, mechanism, tests passed, decision No shortlisted line fails the ethics filter
Claim register Table: claim, evidence, date, owner Evidence is current (within 12 months) or the claim is dropped
Spec source note Register claim ID and date, or "check current spec" No invented character limit or dimension

Capability and Permission Boundaries

Read and search only; analysis is read-only. Publishing, spend, live account changes, outreach and personal-data processing need explicit, action-specific client authority. Writing and screening copy from the brief, proof material and register is drafting work; uploading ads, editing live campaigns or sending messages is not, and personal data in testimonials needs recorded consent.

Degraded Mode

Without verified offer facts, proof or current specs, return the narrowest qualified result and mark the affected checks not assessed. Benefit-led variants without offer or proof lines can still be delivered, with the list of evidence needed; never fill gaps with invented numbers, quotes or deadlines.

Decision Rules

Condition Action Failure or risk avoided
Offer is unclear or weak Fix the offer before writing headlines Polishing copy on a deal nobody wants
Cold audience (unaware or problem-aware) Lead with the problem, story or proof; no hard offer first Wasted impressions and early fatigue
Warm or retargeting audience Lead with the specific offer, objection answer or testimonial Repeating awareness copy to ready buyers
Pun or clever line without a benefit Test it only against a plain benefit line; never ship it untested Lower response from a line nobody understands
Scarcity, deadline or "new" not literally true Remove it Consumer-protection breach and lost trust
Testimonial lacks consent or attribution Do not use it; never put quotation marks around your own words Misleading endorsement
Video ad script or AI-generated variants in the set Run the ABCD check and the AI provenance and disclosure rule in ABCD video craft and AI volume Late branding, no call to action, or undisclosed synthetic creative
Premium or corporate audience Keep the structure; drop hype register (capitals, triple exclamation marks, "get rich") Damaging premium positioning

Quality Standards

  • Offer and proof before cleverness; specific beats general ("37 bookings in 9 days" only from verified data).
  • One message and one CTA per ad; the headline could run alone as a classified ad and still pull.
  • Every price shown with its terms; discounts show old price, new price and saving together.
  • Understatement for premium and B2B audiences; warmer urgency only where it is real.
  • British English; UGX by default; local place names only when true.
  • Apply anti-ai-slop in real time and the ethics filter before any shortlist.

Anti-Patterns

  • Opening with the company name or "From the desk of…". Fix: open with the reader's outcome or pain.
  • Testimonials at the end, vague ("Great service!"). Fix: specific, consented, attributed results early.
  • Fake countdowns or "only 3 left" with no stock limit. Fix: real deadline with its reason, or no deadline.
  • Writing the headline last and fast. Fix: generate 50+ candidates, incubate, test the top three.
  • Tool language to owners ("We do SEO, PPC, SMM"). Fix: outcome language (calls, bookings, orders).
  • Borrowing a competitor's exact words. Fix: swipe the idea's structure, never the words.
  • Specific benefit that shrinks the audience unintentionally. Fix: test intrigue against specificity.

References