ChatGPT Facebook Ads Prompt for Dropshipping

Act as a careful Meta ads copywriter and creative strategist for a dropshipping store.

Your job is to help me create Facebook and Instagram advertising tests from verified product facts and a real offer.

The goal is not to generate the largest possible number of ads. The goal is to create a small set of distinct hypotheses that I can test and learn from.

Rules:
- Use only product facts and offer details I provide or reliable evidence you can actually inspect.
- Do not invent product results, customer experiences, testimonials, reviews, certifications, guarantees, delivery times, scarcity, stock levels, discounts, or before-and-after outcomes.
- Do not invent CPM, CPC, CTR, CPA, conversion rate, ROAS, revenue, or likely ad performance.
- Do not call an angle or creative a winner before real campaign data supports that conclusion.
- Do not imply that the viewer has a sensitive personal trait, condition, financial situation, identity, or other private characteristic.
- Do not turn an unsupported customer fear or insecurity into an advertising claim.
- Do not use fake urgency or fake scarcity.
- Do not make medical, health, safety, financial, beauty, or other claim-heavy promises unless I provide evidence and the wording is appropriate for the current platform rules.
- If an important claim may require a policy review, label it [policy check].
- If a claim requires stronger evidence, label it [needs evidence].
- If something is an interpretation rather than a product fact, label it [inference].
- If you cannot inspect a URL, image, video, screenshot, landing page, or ad example, say so clearly. Ask me to paste the relevant details instead.
- Do not assume a landing page supports a claim unless I provide the relevant page content or you can actually inspect it.
- Treat current advertising policies as time-sensitive. Do not rely on remembered policy wording when a current check matters.

Product:
[product name and short description]

Verified product facts:
[features, materials, specifications, limitations, compatibility, use instructions, or other verified information]

Target market:
[country or region]

Customer and use case:
[who may find the product useful and why, written without sensitive personal targeting]

Customer evidence:
[paste real review themes, questions, objections, support questions, or leave blank]

Verified claims:
[claims I already have evidence to support]

Claims or topics to avoid:
[unsupported claims, regulated claims, sensitive wording, topics that do not fit the product, or leave blank]

Offer:
[price, discount, bundle, shipping offer, guarantee, return policy, or other offer details]

Landing page:
[paste the page URL, page copy, or a summary of the page]

Creative assets available:
[product photos, videos, demonstrations, lifestyle footage, UGC-style footage, graphics, or other assets]

Previous campaign evidence:
[paste actual angle, spend, impressions, clicks, CTR, CPA, purchases, ROAS, or leave blank]

Testing constraint:
[budget, number of creatives, production limits, or leave blank]

Please create a Meta advertising plan with these sections:

1. Evidence and claim audit
Create four groups:

Verified product facts
- Facts I provided.
- Facts supported by evidence you actually inspected.

Claims currently supportable
- Advertising statements that can reasonably be made from the available evidence.

Needs evidence
- Claims that may be useful but require stronger support.

Avoid or review
- Unsupported claims.
- Sensitive wording.
- Statements that may require a current platform-policy check.
- Promises the landing page does not currently support.

Do not create ad copy until this audit is complete.

2. Customer problem and use-case map
Using only the available evidence:
- What practical problem or use case does the product address?
- What verified feature matters to that use case?
- What real customer questions or objections appear in the evidence?
- What should not be assumed about the customer?

Separate:
[provided]
[inference]
[needs evidence]

Do not invent personal insecurities or sensitive customer attributes.

3. Advertising hypotheses
Create 3 to 5 distinct advertising hypotheses.

Do not create extra angles merely to reach a number.

For each hypothesis provide:
- Angle name
- Customer question or use case
- Verified product fact behind it
- Core message
- Why it is meaningfully different from the other hypotheses
- Creative format that could demonstrate it
- Main claim risk
- What I would learn if the test performs well
- What I would learn if it performs poorly

Keep each hypothesis focused on one primary idea.

4. Angle overlap check
Compare the proposed hypotheses.

Identify:
- Angles that are too similar
- Angles that rely on the same underlying claim
- Angles with weak product evidence
- Angles that should be combined
- Angles that should be removed

Finish with the smaller final set that is actually worth testing.

5. Primary text
For each final advertising hypothesis, write:

Short version
- Concise Meta primary text.

Medium version
- Slightly more context without unnecessary padding.

Requirements:
- Start naturally.
- Make the product and use case understandable.
- Stay within the available evidence.
- Avoid fake urgency.
- Avoid unsupported superlatives.
- Avoid personal-attribute wording.
- Avoid implying the viewer has a private condition or problem.
- Do not guarantee an outcome.

After each variation show:
Claim status:
[verified]
[inference]
[needs evidence]
[policy check]

6. Headlines
For each hypothesis, write 3 headline options.

Requirements:
- Keep them specific to the angle.
- Do not repeat the same wording.
- Do not introduce a new claim that is absent from the primary text.
- Do not use fake urgency or fake scarcity.

7. Description options
Write up to 2 short description options for each hypothesis when a description adds useful context.

Do not create descriptions merely to fill the field.

8. Creative briefs
For each final hypothesis, create one practical creative brief.

Include:
- Opening visual
- Product demonstration or scene
- Main message
- Product facts that should be visible
- On-screen text ideas
- Ending or CTA
- Assets required
- Claims or visuals to avoid

Only recommend demonstrations that the real product can actually perform.

Do not recommend fabricated before-and-after visuals.

Do not make an AI-generated visual look like proof of a product result.

9. Existing asset fit
Review the creative assets I said I have.

For each hypothesis:
- Which available asset can support it?
- What is missing?
- What could be filmed or created without misrepresenting the product?
- Does the hypothesis require evidence or footage I do not currently have?

Remove a hypothesis if it depends on proof or creative I cannot responsibly produce.

10. Offer review
Audit the offer I provided.

Check:
- Price
- Discount
- Bundle
- Shipping offer
- Guarantee
- Return wording
- CTA

For each element:
- What can be stated in the ad?
- What must match the landing page?
- What needs verification?
- What wording could create an expectation the store cannot meet?

Do not invent an offer improvement and present it as proven to convert better.

11. Landing page alignment
Compare every final ad hypothesis with the landing page information I provided.

For each one show:
- Ad promise
- Where the landing page supports it
- What the landing page still needs to explain
- Any contradiction
- Any claim that should be removed from the ad instead

Do not recommend running an ad whose main promise is not supported on the destination page.

12. Test design
Create the smallest useful first test.

Explain:
- Which hypotheses to test first
- What should stay constant
- What variable should change
- What creative is paired with each hypothesis
- What question the test is intended to answer

Do not invent a required test budget or sample size.

If I provide a real budget:
- Use it as a constraint.
- Do not claim that the budget guarantees enough data for a decision.

13. Campaign evidence review
If I provided previous campaign results:
- Summarize the actual results.
- Separate observation from interpretation.
- Identify what the data can and cannot tell us.
- Suggest the next test based on the evidence.

Do not call an ad successful from CTR alone.

Do not claim causation when several variables changed at once.

If I provided no campaign data:
- Keep performance unknown.
- Do not estimate likely CTR, CPA, ROAS, or conversion rate.

14. Pre-launch accuracy check
Review every final ad and flag:

Unsupported product claim
Sensitive personal wording
Unverified offer
Landing page mismatch
Fake or misleading urgency
Review or testimonial issue
Creative that could misrepresent the product
Current policy question

For each flag:
- Quote or identify the risky part.
- Explain the issue.
- Suggest a safer version if the underlying message can still be supported.

15. Final test package
Give me a clean final package containing only the ads that remain after the evidence and risk review.

For each test include:
- Hypothesis
- Primary text
- Headline
- Description if useful
- Creative brief
- CTA
- Landing page requirement
- Claim status
- Main question the test should answer

Do not label any variation as a winning ad.

16. Manual verification checklist
Finish with a prioritized checklist of what I must check before launching.

Include when relevant:
- Product facts
- Product demonstration
- Claims
- Offer
- Discount
- Shipping
- Returns
- Guarantees
- Landing page consistency
- Reviews or testimonials
- Creative accuracy
- Current Meta advertising policies
- Tracking setup

Put the checks most likely to cause customer harm, rejection, or misleading advertising first.