Six scored versions, a ranking, and the top 2 in two sizes
Edits copies of your ad in your library. Posts and spends nothing.
One static ad you already like
Using Hermoso: here is my best-performing static ad: [paste its link]. Score it and tell me the single biggest fix. Then make 5 versions that differ only in the headline: one outcome-led, one price-led, one a question, one that leans on social proof, one with a deadline. Keep the picture and every other line exactly as they are. Score all six and rank them. Give me the top 2 in 1:1 and 4:5.
This one has a blank to fill in. ChatGPT sends a prompt the moment it opens, so copy it and fill the brackets first, or use Claude, which drops it in the box for you to edit. Hermoso has to be connected for either one to run it.
The plan, step by step
Score the original
Rates the ad on scroll-stop, hook, clarity, brand fit and call to action, and names the one change that would help most.
score_adTest five headlines
Writes five headlines in your brand voice and edits the ad once per headline. The picture, product and layout stay put, so the headline is the only thing being tested.
headline_variantsRank all six
Scores each version the same way so you can see which headline moved the needle before you spend on any of them.
score_adSize the finalists
Lays out the top two for square and portrait feeds.
resize_ad
What it will not do
- A score is a prediction, not a result. Run the top two as a real test and let your campaign numbers decide.
- Your original ad is never changed. Every version is a new copy.
Before you start
- Hermoso connected to ChatGPT or Claude. It is one link and takes about a minute: see how to connect. Or skip that and paste the prompt into the Studio chat in the Hermoso app.
- Your brand saved in Hermoso. No brand yet? Paste your website into the chat first and ask your AI to set one up.
- A Hermoso account. The free plan is enough to try it; anything that renders uses credits (see pricing).
Make it yours
Give it your own five headlines to test, or tell it what to test ("shorter", "speak to busy parents"). For a video ad, ask for new opening hooks instead.
Run it again
When the test ends, paste the winner back in and run it again. Each round starts from a stronger ad.
Related
- AI image ads
- Workflow: headline variants
- Workflow: new hooks for a video
- Competitor ad watch (blueprint)
- Customer-voice angle bank (blueprint)
- A week of posts, review first (blueprint)
- All blueprints
Not connected yet? Add Hermoso to ChatGPT or Claude with one link, then come back and run this. Or open the app and paste it into the Studio.
Connect Hermoso → Or start in the app