The prompt
Score 10 of 10, Excellent
Use it when You run social media ads and want a set of variations built to teach you something about your customers, not a pile of random versions.
- {{platform}}
- {{business}}
- {{audience}}
- {{budget_and_time}}
- {{formats}}
- {{variant_count}}
- {{product_and_offer}}
- {{what_we_know}}
Full prompt and why it works
Prompt
Plan a creative test for our ads on {{platform}}, for {{business}}, and write the ad variations for it. The point is to learn which message works for {{audience}}, so the variations must differ in one planned way at a time, and the results must be easy to read afterward. Budget and length of the test: {{budget_and_time}} Formats we can make, and who makes them: {{formats}} How many variations: {{variant_count}} What we sell and the offer in this campaign: """ {{product_and_offer}} """ What we already know (past ads that worked or did not, with their results, customer comments, questions people ask before buying, reasons people buy): """ {{what_we_know}} """ Produce, with a heading for each part: 1. The question this test answers, in one sentence, and our hypothesis. 2. The variable: pick one (the angle, the hook, the visual or how the offer is framed) and say why it is the right first test for this budget. Everything else stays the same across the variations. 3. The variations, as a table with the columns Name, Angle, Hook or first line, Primary text, Headline, Visual direction. Draw angles from the problem, the outcome, a customer's own words from our notes, a demonstration, or a comparison with the old way of doing it. 4. A naming pattern for the ads, so the report can be sorted by the variable, for example "T3_angle-problem_v1". 5. How to read the result: the one metric that decides it, how long to wait before judging (marked as a rule of thumb, with reasons it may not hold at our budget), and what we would test next for each outcome. A good test ends with a decision we can act on, even when the answer is "no clear difference". Keep the copy to the facts above, with no prices, results, reviews or customer quotes we did not give. Avoid before-and-after or health claims that platform ad rules restrict, and fake scarcity or urgency we cannot back up. If the budget is too small to test this many variations fairly, say so and suggest fewer. If our notes do not say what has worked before, state what you are assuming.
Variables
- {{platform}}
- Where the ads run, for example "Facebook and Instagram" or "TikTok".
- {{business}}
- Your business and what this campaign promotes.
- {{audience}}
- Who the ads target, for example "new parents within 20 miles".
- {{budget_and_time}}
- Total test budget and how long it runs, for example "$600 over two weeks".
- {{formats}}
- What you can make and who makes it, for example "phone videos I film, and still images from our photos".
- {{variant_count}}
- How many variations you want to run, for example 4.
- {{product_and_offer}}
- What you sell and the exact offer, with price and terms if they are part of it.
- {{what_we_know}}
- Past ads and their results, customer comments, the questions people ask and the reasons they give for buying. Write "nothing yet" if this is your first campaign.
Why it works
- Who it's for. The variations exist to answer one question, so the prompt picks a single variable and keeps everything else the same across them.
- What good looks like. A good test ends with a decision, which is why it names the deciding metric and the next test for each outcome before any money is spent.
- If unsure. A budget too small for the number of variations is flagged with a smaller plan, and any assumption about past results is stated.