AI creative
Ad copy formulas that still work
A practical guide to choosing, writing and testing ad copy formulas, with hypothetical examples, claim checks and a measurable approval workflow.
OppRise Editorial · Reviewed · 8 min read
The short answer
AIDA, PAS, feature–advantage–benefit, before–after–bridge and objection–proof–action remain useful drafting structures, not proven winners for every campaign. Choose one around the buyer's main barrier, use verified claims and test the argument against a relevant business outcome.
What to take away
- Match the formula to buying intent and the barrier to action.
- Connect every benefit to evidence and a consistent landing page.
- Adapt the argument to the placement without losing qualifications.
- Measure approved-output efficiency, including review and rework.
1. Start with the buying decision, not the acronym
A formula organises an argument. It cannot rescue an irrelevant offer, create evidence or make a landing page fulfil an unsupported promise. Choose the structure after identifying what the buyer needs to understand before taking the next step.
For someone searching for a replacement dishwasher seal, compatibility and availability may matter more than a story. For someone discovering a new bookkeeping service, explaining the problem and process may be necessary. Use fewer persuasive steps when intent is already clear.
Every advertising example in this guide is hypothetical. Treat its product details as assumed, not verified claims or suggested wording for your business.
- Name one audience, one buying situation and one desired action.
- List verified offer details, exclusions and the landing-page destination.
- Identify the main barrier: relevance, understanding, trust, price or effort.
- Choose a formula that addresses that barrier without adding unnecessary claims.
2. AIDA: build understanding for an unfamiliar offer
Attention, interest, desire, action works as a drafting sequence when buyers need a short explanation. Attention should establish relevance, not merely interrupt. Interest explains the mechanism; desire connects it to a useful outcome; action offers a proportionate next step.
Hypothetical ad for a beginner pottery course: “Your first bowl starts here. Learn hand-building in a small-group Saturday class, with clay and firing included. Make something for your own kitchen. View upcoming classes.” The copy moves from recognition to process, tangible purpose and booking information.
The tradeoff is space. Compressing all four stages into a search headline can produce vague fragments. Put relevance in the headline, practical details in supporting copy and the fuller explanation on the destination page. Avoid empty attention lines such as “This changes everything”, which consume space without explaining the offer.
- Can the reader identify the product before reaching the CTA?
- Does the desire line describe a plausible benefit rather than a guaranteed transformation?
3. PAS: make the problem concrete without manufacturing anxiety
Problem, agitation, solution is useful when the buyer recognises a recurring frustration. Interpret agitation as explaining the practical consequence, not amplifying fear or shame.
Hypothetical ad for a quote-management tool: “Still searching your inbox for the latest quote? Scattered versions make the next follow-up harder to organise. Keep quotes and follow-up notes together. See the workflow.” The consequence follows logically from the problem without claiming lost revenue.
PAS becomes manipulative when it invents damage: “Every missed follow-up is costing you thousands.” Unless you can substantiate that claim in the relevant context, remove it. The formula also fits poorly when the purchase is mainly pleasurable. A restaurant experience rarely needs a manufactured crisis to become attractive.
- Use customer language for the problem, with permission where necessary.
- Describe an observable consequence, not an assumed emotional state.
- Show how the solution addresses the named problem specifically.
4. FAB: translate features without outrunning the evidence
Feature, advantage, benefit connects what a product has with why someone might care. It is especially useful when drafts contain specifications but no reason to choose.
Hypothetical commuter bag example: “A separate shoe compartment keeps footwear apart from your work essentials, so you can pack for the office and training in one bag.” The compartment is the feature, separation is the advantage and combined packing is the benefit.
Keep the inference modest. A separate compartment does not automatically mean odour-proof, waterproof or hygienic. Each additional claim needs its own support. Conversely, removing every specification can make the benefit generic: “Make your commute easier” tells buyers little.
- Trace each benefit back to a verified feature.
- Retain specifications buyers need to assess suitability, such as dimensions.
- Check that the pictured variant actually includes the advertised feature.
5. Before, after, bridge: sell the process, not a miracle
Before, after, bridge frames a change and explains how it happens. Its strength is making a service tangible. Its danger is presenting an ideal outcome as inevitable.
Hypothetical document-organisation service: “Files spread across folders? Build a shared structure with clear names and owners. Our guided setup session helps your team agree the system and organise a starting set of documents.” The bridge is a defined session, not an unexplained leap from disorder to perfection.
Separate deliverables from outcomes. A consultant may promise a documented naming convention if that is in scope, but not that nobody will misfile anything again. For services requiring customer participation, state the effort involved. This can reduce superficial appeal while making the offer easier to assess accurately.
- Describe the starting situation without belittling the buyer.
- Make the bridge a concrete mechanism, deliverable or sequence.
- Disclose material prerequisites and avoid implied result guarantees.
6. Objection, proof, action: answer the reason for hesitation
When buyers already understand the offer, another benefit headline may miss the issue. Use an objection, relevant evidence and a next step instead. Evidence can be a specific policy or demonstrable process; it does not have to be a testimonial.
Hypothetical furniture ad: “Unsure it will fit? Check the full dimensions and downloadable room-planning sheet before you choose. Explore the two-seat range.” This addresses sizing uncertainty without promising universal suitability.
Match proof to the objection. A customer review praising delivery does not prove durability. If using testimonials, preserve their meaning, obtain permission and avoid turning one person's experience into a typical-result claim. Discounts also need care: they answer price concerns, not uncertainty about quality or compatibility.
- Use actual enquiry themes rather than imagined objections.
- Confirm the evidence is accessible from the destination.
- Choose a CTA that resolves uncertainty, such as “Check compatibility”.
7. Adapt the formula to how the ad is encountered
A formula is not a requirement to show every sentence in every placement. In search, lead with the product or service relevant to the query. Where platforms combine assets, ensure individual headlines remain accurate without depending on a particular preceding line.
For short video, distribute the argument across the opening visual, spoken explanation and closing action. Hypothetical bike-repair creative might open with a loose brake lever, explain the available inspection service and close with “Check workshop availability”. Do not imply immediate appointments unless verified.
Captions should convey relevant speech and sound, not just repeat a promotional headline. W3C's caption guidance explains this accessibility requirement. Keep captions readable alongside offer conditions and interface overlays. Accessibility and claim clarity should survive the edit, not be squeezed in after approval.
- Review copy in the actual crop and placement.
- Check the message with sound off and captions on.
- Keep important qualifications legible and close to the claim.
8. Test a persuasion hypothesis, not an acronym contest
Define the difference you want to learn about. For the same hypothetical bag, compare a compartment-feature argument with an office-to-training benefit argument while holding the offer, audience, destination and visual stable where possible. This tests emphasis, not whether FAB universally beats AIDA.
Google Ads experiments provide a way to test campaign changes against an original campaign. Choose an appropriate supported design, primary outcome, budget and decision rule before launch. Automated asset mixing can complicate attribution to individual lines.
Use qualified enquiries or purchases when those match the objective. Click-through rate is diagnostic, not proof of commercial success. GA4 key events and Google Ads conversions differ; document the source, event definition and reporting window rather than treating their counts as interchangeable.
- Allow for conversion delay and acknowledge insufficient data.
- Check lead quality, order value and misleading-click risks.
- Do not infer causation from a simple before-and-after comparison.
9. Turn formulas into a controlled production workflow
Give writers or AI a verified offer sheet, one audience barrier and two chosen structures. Request genuinely different arguments, not twenty synonym swaps. Review claims first, then message, tone and placement. Keep the approved wording beside its supporting evidence.
Measure efficiency as total briefing, drafting, review and rework minutes per approved variant. Hypothetically, reducing drafting from 40 to 20 minutes while review rises from 10 to 35 makes the workflow slower. Track approval rounds and post-approval corrections too; more output alone is not an efficiency gain.
Brand Vault stores brand context; RiseStudio prepares grounded creative for human refinement and approval. RiseCampaigns is designed around explicit approval and paused campaign creation. Connected execution depends on provider permissions and availability: TikTok and LinkedIn approvals are pending, and Microsoft Advertising is planned. These workflows do not guarantee better ads or time savings.
- Approve factual accuracy before generating placement variants.
- Record the hypothesis, owner and destination with each final ad.
- Compare production time only across work of similar complexity.
Frequently asked questions
Which formula should a small team start with?
Start with feature, advantage, benefit for a concrete product, or objection, proof, action when enquiries reveal a recurring concern. Neither requires an elaborate story.
Can AI choose the formula?
AI can suggest structures from a supplied brief. A person should verify the audience assumption, supporting evidence, offer conditions and final wording.
How many variants should we test?
Use the fewest needed to test a clear hypothesis within your available traffic and budget. More variants can spread evidence too thinly.
Should the formula be visible in the final copy?
No. It is a drafting scaffold. Remove repetitive transitions and keep natural language, while preserving the reasoning that makes the offer understandable.
Put this into practice with