Meta Ads Optimizer

What it does

Reviews active campaigns, ad sets, and ads, classifies each entity by risk (ABO/CBO aware), decides to pause, trim, or scale budget based on performance rules, and applies changes through approval tiers with a verifiable log of every action.

When to use it

• You manage multiple Meta Ads accounts and need a consistent review cadence.
• You want to scale budget on winning ads without breaking the algorithm's learning phase.
• You need auditable evidence of why each ad was paused or scaled.

Steps

1. Define the scope: accounts, campaigns, and analysis period to review.
2. Pull performance metrics per ad, ad set, and campaign.
3. Classify each entity by risk level based on spend, results, and ABO/CBO structure.
4. Propose actions: pause clear losers, hold ambiguous entities under observation, scale winners.
5. Present the proposed change list with the rationale for each, grouped by required approval tier.
6. Apply only explicitly approved changes, never automatically.
7. Verify the entity reached the expected state after the change and log the result for the next review.

The skill

---
name: meta-ads-optimizer
description: Run recurring Meta Ads optimization: classify risk, pause losers, scale winners with tiered approval gates and a change log.
---

# Meta Ads Optimizer

## 1. Purpose
Maintain a disciplined optimization cadence on Meta Ads accounts, avoiding impulsive or undocumented changes.

## 2. Scope
Define which accounts, campaigns, and date range get analyzed in each run.

## 3. Risk classification
Group entities (ABO or CBO) by cumulative spend, results, and volatility to decide how much caution to apply before touching them.

## 4. Decision rules
Propose pausing clear losers, holding ambiguous ones under observation, and scaling budget gradually on proven winners.

## 5. Approval gates
Separate changes by impact level; higher-risk changes require explicit approval before execution, never applied alone.

## 6. Execution and verification
Apply only approved changes, confirm each entity's final state after the change, and save the action log.

## 7. GoHighLevel angle
Cross-check the conversion source against GHL pipelines to confirm leads Meta reports as winners actually progress in the CRM before scaling budget.

For Claude Code, save it as ~/.claude/skills/meta-ads-optimizer/SKILL.md. In Cursor, ChatGPT or another assistant, paste it as project instructions.