Turn one campaign brief into 20 Meta and Google ad variants, score each against your brand rules and limits, and queue the best for approval.
Creative testing needs volume, but writing 20 distinct headlines and primary texts for every campaign takes most of an afternoon. Under deadline, teams launch three near-identical variants and learn nothing from the test. Copy also slips through with unapproved claims or headlines that get cut off at 30 characters.
Create a Google Sheet with a Briefs tab (brief_id, product, audience, offer, proof_points, landing_url, status) and a Variants tab (brief_id, angle, headline, primary_text, description, score, issues, approved). Put your brand rules on a Rules tab: banned words, unapproved claims, tone and required disclaimers.
In Make, use Google Sheets "Watch Rows" on the Briefs tab with a filter for status = Ready. Read the Rules tab with "Get Range Values" so every run uses the latest rules.
Send the brief to the OpenAI (ChatGPT) "Create a Completion" module with the generation prompt and JSON response format turned on. Asking for four angles (pain, outcome, proof, objection) with five variants each makes the test compare ideas, not synonyms.
Run an Iterator over the variants and use the length() function to drop Google RSA headlines over 30 characters, descriptions over 90, and Meta primary text over 125 characters, where the feed truncates it. Only variants that fit go on to scoring.
Pass each remaining variant and the Rules tab to the Anthropic Claude module with the scoring prompt. A second model reviewing the copy is less likely to wave through the generator’s own habits.
Write every variant to the Variants tab with its score and issues, and add conditional formatting that highlights a score of 8 or more with no issues. The marketer ticks approved on the ones to launch and uploads them through Ads Manager or Google Ads Editor.
After 7 days, paste CTR and CPA per ad back into the Variants tab. Add the winning angle and top two headlines to the next brief’s proof points so each round starts from what worked.
You write paid social and search ads for {{brand_name}}.
Brief:
Product: {{product}}
Audience: {{audience}}
Offer: {{offer}}
Proof points: {{proof_points}}
Landing page: {{landing_url}}
Brand rules: {{brand_rules}}
Write 20 ad variants: 5 for each angle - pain, outcome, proof, objection.
For each variant return:
- "angle"
- "headline": max 30 characters
- "primary_text": max 125 characters
- "description": max 90 characters
Variants within an angle must use different hooks, not reworded copies. Use only the proof points given; do not invent statistics, reviews or awards. Return a JSON array only.You are a brand and compliance reviewer for {{brand_name}}.
Brand rules: {{brand_rules}}
Ad variant: {{variant_json}}
Score the variant from 1 to 10 on how well it follows the rules and how clearly it sells the offer to {{audience}}.
List every issue you find, quoting the exact words:
- banned words or unapproved claims
- statistics or promises that are not in the brief
- tone that breaks the rules
- a missing required disclaimer
Return only JSON: {"score": number, "issues": ["..."], "fix": "a corrected version of the weakest line, or null if there are no issues"}Give it a keyword and get a brief built from the pages that rank today: headings, gaps, questions to answer and a target length.
An hour before each discovery call, get a one-page brief on the account, the attendee and recent news, saved to Salesforce.
Score every demo request against your ICP, update HubSpot and ping the right rep in Slack within a minute of the form submit.