| name | paid-ads |
| description | Use when reviewing or planning paid acquisition across Google, Meta, LinkedIn, X, Reddit, TikTok, sponsorships, retargeting, or other paid channels. |
Paid Ads
Review paid acquisition to decide what to cut, fix, test, or scale.
Required Context
Use available platform exports, screenshots, campaign structure, UTMs, landing pages, CRM data, or spend summaries. If data is missing, ask for:
- Channel and objective.
- Spend, conversions, revenue, CAC, CPA, ROAS, or pipeline.
- Attribution window and conversion definition.
- Audience and offer.
- Creative examples.
- Landing page.
- Budget and time period.
Do not infer performance from structure alone when spend and conversion data exist.
Diagnose The System
Inspect:
- Economics: target CAC/CPA, payback, LTV, gross margin, sales cycle.
- Signal: conversion tracking quality and event meaning.
- Structure: campaign objective, segmentation, budget allocation, bidding logic.
- Audience: targeting, exclusions, intent, saturation.
- Creative: message variety, proof, format fit, fatigue.
- Landing path: match between ad promise and page.
- Learning: what tests have been run and what changed.
Give the data a confidence grade before recommending budget moves. Weak tracking should produce repair or learn recommendations, not scale recommendations.
Classify Issues
- Tracking issue: data cannot support decisions.
- Offer issue: click interest does not convert.
- Message issue: creative attracts wrong or weak intent.
- Audience issue: targeting is too broad, too narrow, or stale.
- Economics issue: campaign works mechanically but cannot pay back.
- Scale issue: winner exists but budget, creative, or funnel cannot absorb more.
Output
Paid acquisition read:
[one paragraph]
Current numbers:
Spend:
Conversions:
CPA/CAC:
Revenue/ROAS:
Confidence in tracking:
Classification:
[stop / repair / learn / scale]
Diagnosis:
| Area | Finding | Evidence | Severity | Next move |
Budget moves:
- Increase:
- Hold:
- Cut:
Tests:
1. [test]
2. [test]
3. [test]
Data needed before scaling:
- [missing signal]
