Real Agent Cases

Audit 90 days of Meta ads from a CSV export with Claude Code

Posted
by Prop (@propialis)
Last checked
Method:
How we verify

Evidence: Described by posterThe person describes what happened in words, without screenshots or video.

Outcome

Claude Code flagged 14 money-losing creatives, found that 7 of the top 10 ads opened with a question, and spotted 3 ad sets cannibalizing each other.

Poster says:14 money-losing creatives flagged

This is the source's own number. We haven't measured it.

Agent

Claude Code

Prompt Original prompt

Exact prompt as published in the source.

give me every ad that spent over $50 with a CPM above $20 and a hook rate below 25%

Try this in:

ChatGPT Gemini Grok

Sends the Original prompt shown above. ChatGPT and Grok open with it filled in, and ChatGPT may send it straight away. Gemini opens empty, so we copy the prompt for you to paste. Every button also copies it.

The source used Claude Code, which can take actions on your computer or in a browser. These are general chat tools, so results may differ, and a normal chat may only give you a plan.

Opens in a new tab. We don't see what you send.

These links are only a shortcut. We're not affiliated with OpenAI, Google or xAI, and your results may not match the source.

Optional stronger version

Our rewrite with extra checks. The poster didn't test it.

Goal: From my Meta ads export, list every ad with spend over $50, CPM above $20 and hook rate under 25%.

Inputs (fill in or confirm):
- CSV export
- How hook rate is calculated in my data
- Date range
- Columns for spend, impressions, 3-second views

Rules:
- Compute CPM and hook rate explicitly and show formulas.
- Filter by all three conditions.
- Sort by spend.
- Keep ad IDs and names.
- Define hook rate as 3-second video views ÷ impressions unless my data defines it differently; say which you used.
- Exclude non-video ads from the hook-rate test and list them separately.
- Add a column showing which threshold each ad fails by how much.
- Export the result as CSV.
- Check row counts match the export.
- Rank ads by hook rate and cost per result.
- Flag ads with too little spend to judge.
- Give 3 concrete recommendations with numbers.
- Recalculate hook rate yourself from the raw columns and say if it differs from the export's figure.
- Group results by creative type or angle if the names show it.

If you get stuck: If a column is missing, ask. If a login, CAPTCHA or two-step code appears, stop and ask me to handle it; never guess or store passwords.

Verify before you finish: Check a few rows by hand.

Output: Filtered table with metrics.

Done when: the list matches the criteria exactly.

Source

https://x.com/propialis/status/2041264181106692318

Related cases

  • Manage Google Ads with Claude Code connected to analytics and the websiteRecreated promptEvidence: Described by poster

    With Google Ads, Google Analytics and the Webflow site connected to a Claude Code project, the poster adds negative keywords quickly, tracks ad performance, and found that mobile wasn't converting because the page showed nothing relevant above the fold.

    Claude Code · Marketing · Source date:

  • Find car brands without a web app and email them a pitch with ManusRecreated promptEvidence: Screenshots or logs

    Manus searched for car brands without a web app, wrote an email and sent it to them; some replied and the poster built one an app.

    Manus · Marketing · Source date:

  • Qualify scraped leads against your ideal customer before cold email with Claude CodeRecreated promptEvidence: Described by poster

    Claude Code reads each scraped company's website and checks it against the poster's ideal customer profile, dropping about 30% of leads before outreach; the poster says reply rates roughly doubled and they booked 100+ meetings in 3 months.

    Poster says:100+ meetings booked in 3 months

    Claude Code · Marketing · Source date: