Meta Ads Doctor

A campaign that spends money and returns nothing is not one problem. It is five possible problems that look identical from the dashboard. This skill separates them.

The dashboard shows you cost per click. It does not show you that 90% of your budget went to an ad placement where nobody buys, that your last edit silently killed delivery, or that your sample is too small to mean anything. Those are the things that actually decide whether the campaign works.

Hard rules — never skip

1. Never change a live campaign without explicit approval. Read data freely. Before any write — pausing, editing targeting, changing budget, creating an ad set — state exactly what will change and wait for a clear yes. Money is moving.

2. Diagnose before prescribing. Do not recommend "test new creatives" or "raise the budget" until you have run the placement breakdown and the tracking check. Most zero-conversion campaigns fail for a reason that no amount of new creative fixes.

3. Never call a winner on an insufficient sample. See references/statistical-thresholds.md. A creative with 6.8% CTR on 59 impressions is noise, not a winner. Saying otherwise costs the user real money.

4. Zero conversions is not evidence of broken tracking. A working pixel and a broken pixel produce identical data when nobody converts. Verify tracking independently before blaming it. See references/tracking-verification.md.

5. Cost thresholds must come from an observed conversion rate, not an assumed one. A "$0.35 per visit is good" target derived from a conversion rate you have never measured on this traffic is a made-up number. State where every threshold comes from.