65 Claude Prompts for Paid Media
Audits · Search terms · Creative · PMax · Meta · Post-click · Measurement · Margin · Reporting
Most prompt packs give you five in full and keep the rest behind a comment or a DM. This one does not. All sixty-five are written out, with the input each one needs and the thresholds that decide whether a finding is real.
The prompts were never the scarce part — anyone can write a prompt. What is scarce is knowing which check to run when, what number counts as a problem, and when to stop optimising and go fix the landing page instead. That sits in the "use when" lines and the volume floors, and it only comes from having run a lot of accounts.
It assumes you can already pull a report and act on what it says. If you have never opened your own ad account, this will not help you — and the document says so on page three rather than selling you something it cannot deliver.
Ten rules that decide whether any of it works
The document spends a page on operating rules before the first prompt, and they matter more than the prompts do. Paste real data rather than descriptions of it. Rank by money, because twelve small problems and one large one is not thirteen problems. Give the model your constraint — CAC ceiling, budget cap, inventory position — before asking for a recommendation, or you get advice for a business that is not yours.
The line that appears in almost every prompt is an instruction to say "insufficient data" rather than guess. A fluent wrong answer is worse than no answer, and language models are built to be fluent.
And the rule that does not move: these prepare and rank, you approve and implement. Anything in the profitability cluster produces a number that will drive spend — verify it in a spreadsheet, and ask for the formula alongside the result.
Read it here
The whole document, in the page. Nothing to fill in, and no need to leave the site — though you can download it if you prefer.
1 / 33
Free to read, quote and share with attribution. Not for resale.
What's inside
Nine clusters, sixty-five prompts, each with the input it needs and the moment to run it.
- 01
Cluster 1 — Account Audit & Waste
8 prompts. Where the money is draining out right now. Run this cluster first; it tells you which one to run second.
- 02
Cluster 2 — Search Terms & Keywords
8 prompts. Reading the report that shows what you actually paid for.
- 03
Cluster 3 — Ad Copy & Creative
8 prompts. Testing frameworks and copy that matches the query.
- 04
Cluster 4 — Performance Max & Shopping
7 prompts. Feed quality, asset groups, and seeing inside PMax.
- 05
Cluster 5 — Meta & Paid Social
7 prompts. Audience structure and creative diagnosis off search.
- 06
Cluster 6 — Landing Page & Post-Click
5 prompts. The half of the result that happens after the click.
- 07
Cluster 7 — Measurement & Tracking
6 prompts. Making the platform's conversion count reconcile with what the business banked.
- 08
Cluster 8 — Profitability & Budget
8 prompts. Margin maths, CAC ceilings and budget models. Every output here drives spend, so each is written to show its formula.
- 09
Cluster 9 — Reporting, Strategy & Competitor Intel
8 prompts. Including prompt 60, written to be run on the last report your agency sent you.
Who it's for
- People who run the ads themselves
- Marketing managers overseeing whoever does
- Founders paying an agency for paid media
- In-house performance teams
Free · no email required
33 pages, 424 KB. Free to read, quote and share with attribution. Not for resale.
Download the PDFI'd rather you argue with it than agree quietly
If something in here is wrong, or you have run these checks on your own account and found something different, I want to hear it. And if you want the eighth pillar run against your current agency's reporting, that is a conversation worth having.