Quick check
Guess first, no penalty. You want to learn which headline gets more clicks. What should change between version A and version B?
An A/B test shows two versions to similar people and compares the results. AI can write fifty headlines in seconds, but fifty random headlines don't teach you much. What you want is a few versions that differ in one planned way.
A prompt is the message you type into the tool. So tell it what stays fixed, what changes, and the platform's limits. Google's responsive search ads, for example, take headlines of "up to 30 characters" (Google Ads Help).
A prompt for versions you can learn from
Write 6 headlines for our spring bike service ad.
Each under 30 characters.
Keep the offer the same in every one: full service, $79.
Change only the benefit: 2 about saving time, 2 about safety, 2 about price.
Don't invent numbers or promises.Now the versions differ in one planned way: the benefit. If the safety ones win your test, you've learned something about your customers, not just about one ad.
Fill the gaps
Fill the gaps so the versions test one idea at a time.
Prompt
Write 6 headlines for our spring bike service ad. Each under . Keep : full service, $79. Change only : 2 about saving time, 2 about safety, 2 about price.
Count the characters yourself before you upload. Don't rely on the tool's own count.
Try it
Your steps, about 2 minutes
- 1.Pick one ad you run now. Write a prompt that keeps the offer fixed and changes one thing, with your platform's limit. Count the characters in what comes back. Add the prompt to your AI brief kit. Your kit now holds a brief, a claim check, a voice card and a test prompt, ready for your next campaign.2 min