More in Ads & analytics.
Let Claude Code run a $1,500 Meta Ads budget on its own for a month
FailureEvidence: Described by posterPosted Last checked
Giorgio Liapakis let Claude Code run Meta Ads alone; it ended at $6.05 per signup against a $2.50 target, a failure.
How to use this
- Tap Copy prompt.
- Open the tool the case used (or a similar one) and paste the prompt; results may differ from the original.
- Paste it into the message box.
- Send it, then read the answer before acting on it.
New to this? What's a prompt?
A prompt is the message you send to the AI tool. Copy the one on this page, put your own details where it asks, and read the answer before you act on it.
Prompt
Recreated prompt
The source didn't share its exact prompt, so we wrote one from what it describes. The result shown is the source's, not ours.
The person didn't share their exact words, so we wrote this from what they described. We haven't tested it, and they didn't use this exact version.
Each day, read yesterday's logs and learnings, pull fresh numbers from Meta Ads and our analytics, decide whether to pause, scale or create ads within the budget rules, make the changes, then write today's log and update the learnings file.
Try this in:
Sends the Recreated prompt shown above. ChatGPT and Grok open with it filled in, and ChatGPT may send it straight away. Gemini and Claude open 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, Anthropic or xAI, and your results may not match the source.
Optional stronger version
Our rewrite with extra checks. The poster didn't test it.
Before you start, ask me 1–3 short questions about the details below and wait for my answers. Once I've answered, confirm the plan in one line and run the whole task start to finish. Goal: Run a daily ad review that keeps spending within my rules, improves cost per signup and checks that the signups are real leads. Questions to ask me first: - What's the daily budget and the cost-per-signup target? - What counts as a good lead, and where can I check that? - Which ad styles or claims are off limits? Inputs (fill in or confirm): - Ad account: [ACCOUNT] - Daily budget and total budget: [AMOUNTS] - Target cost per signup: [AMOUNT] - Analytics and signup tools: [FOR EXAMPLE POSTHOG, NEWSLETTER TOOL] - Off-limits ad styles: [FOR EXAMPLE FAKE TESTIMONIALS] - Pause and scale thresholds: [RULES] Rules: - Read the earlier logs and learnings before deciding anything. - Check lead quality from day 1, not just cost per signup. - Follow the pause and scale thresholds exactly and never go over budget. - Try new ad formats each week instead of repeating one winner. - Log each decision with the reason and the result you expect. Checkpoint: After the first week, send me a short report and wait for my OK before changing the plan. If you get stuck: If results stall for three days, tell me and suggest a change instead of continuing the same way. Verify before you finish: Compare spend against the budget and signups against the analytics tool each day. Output: A daily log, an updated learnings file and a weekly summary with cost per signup and lead quality. Done when: The month ends within budget with an honest comparison against the target.
AI tool
Outcome
Giorgio Liapakis gave Claude Code full control of a fresh Meta Ads account for January 2026, with a $1,500 budget at about $50 a day and a goal of newsletter signups at $2.50 each (in Australian dollars). Once a day, a single slash command, /let-it-rip, started a session that read the earlier logs and metrics, decided whether to pause, scale or create ads and wrote its notes back to files. Liapakis spent about two minutes a day. Claude tested about 50 ad variants across 8 formats and kept coming back to whiteboard and notebook-style "ugly creative." A lead magnet offer beat the plain newsletter offer by 18%, and preset rules (pause at $6 per lead, scale at $3) prevented runaway spending. But it didn't check lead quality until Day 16, and a business-email-only rule Liapakis added by hand around Day 18 "caused carnage" the account never recovered from. The month ended at $6.05 per subscriber, which Liapakis called a failure against the original target.
Evidence: Described by posterThe person describes what happened in words, without screenshots or video.
Known limitations
FailureThe source says the AI did not get the job done.
What went wrong: The system ran smoothly and stayed in budget, but it missed the target by a wide margin: $6.05 per signup against $2.50.
Source
- Posted
- by Giorgio Liapakis (Medium)
- Last checked
- Method:
- How we verify
Context: The daily slash command is quoted, but its full instructions aren't, so the prompt here is recreated from the described daily loop. The $2.50 target was chosen loosely at the start, and Liapakis runs a growth agency, Wibci, that the post promotes.
More in Ads & analytics
Manage Google Ads with Claude Code connected to analytics and the website
With Google Ads and Analytics connected to Claude Code, the poster adds negative keywords fast and found why mobile wasn't converting.
WorkedClaude CodeEvidence: Described by posterView CaseAudit 90 days of Meta ads from a CSV export with Claude Code
Claude Code flagged 14 money-losing ads, found 7 of the top 10 opened with a question, and spotted 3 ad sets cannibalizing each other.
WorkedClaude CodeEvidence: Described by posterPoster says:14 money-losing creatives flagged
View CaseWeekly X analytics review and growth coaching with ChatGPT agent
Every Monday ChatGPT agent pulls the poster's X analytics, studies which posts worked, and suggests bio and pinned-post changes.
WorkedChatGPT agentEvidence: Screenshots or logsView CaseFix Google Ads conversion tracking with ChatGPT agent
A solo creator had ChatGPT agent log into their Google Ads account and fix a conversion-tracking issue.
WorkedChatGPT agentEvidence: Screenshots or logsView Case

