How to use the Ad Campaign Analyzer agent skill

What is an agent skill and what does this one make the agent do?

An agent skill is a folder containing a SKILL.md file with instructions that an AI agent loads when a request matches the skill. This one is named ad-campaign-analyzer and is licensed under MIT.

Once installed, the agent can work through campaign exports (CSV files, pasted tables or dashboard screenshots), normalize the data, and separate what the numbers show from what they prove. It quantifies uncertainty where the data allows and suggests small, bounded budget experiments rather than sweeping reallocations. It asks for context such as goals, target CPA or ROAS, and recent changes before drawing conclusions.

• Normalizes Google Ads, Meta Ads and LinkedIn Ads columns into one standard analysis format
• Builds a channel-level rollup when the channels' data can be compared
• Flags investigation candidates such as zero-conversion spend, high CPA outliers and low CTR ads
• Proposes bounded budget or bid tests and a significance check for A/B tests

How do you install the Ad Campaign Analyzer skill?

Download the ZIP file from the skill page on this site and unzip it. You can install the skill in one of two places. For a personal setup, place the unzipped folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.

In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/ (folder containing SKILL.md)
• Project only: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload

How do you trigger the skill with a natural request?

You do not need a special command. Describe what you want in plain language, and the agent will load the skill when your request matches its purpose. Requests like these are the ones the skill is written for:

• Analyze my Google Ads performance
• Which ads should I kill?
• Is this campaign working?
• Where am I wasting ad spend?
• Optimize my Meta Ads
• How should I split my ad budget?
• Should I spend more on Google or Meta?
• I have $X/month for ads — how should I distribute it?

What should you prepare before you ask?

The skill begins with an intake step. It will ask for the data and some context, so having these ready makes the first answer far more useful. The skill's intake asks for the campaign data, platforms, time period, monthly budget, primary goal (such as demos, trials, purchases or leads), target CPA or ROAS, any known changes to creative, budget or targeting, the channels you run, and any constraints such as minimum spend per channel.

Campaign data can be a CSV export from Google Ads, Meta Ads Manager or LinkedIn Campaign Manager, a pasted performance table, or screenshots of a dashboard. If you do not have a target CPA or ROAS, the skill asks for an approved, dated benchmark source rather than inventing one.

• Google Ads columns expected include Campaign, Ad Group, Keyword, Impressions, Clicks, CTR, CPC, Conversions, Conv Rate, Cost, Conv Value
• Meta Ads columns expected include Campaign, Ad Set, Ad, Impressions, Reach, Clicks, CTR, CPC, Conversions, Cost Per Result, Amount Spent, ROAS
• LinkedIn Ads columns expected include Campaign, Impressions, Clicks, CTR, CPC, Conversions, Cost, Leads

What examples of analysis can you expect?

For multi-channel work, the skill checks whether conversion definitions, attribution windows, timezones, currencies, date ranges and deduplication rules can be aligned. If they cannot, it presents separate channel results and marks the cross-channel comparison as non-comparable. This is one of the most useful safeguards in the skill.

For single-campaign checks, it compares CTR, CPC, conversion rate, CPA, ROAS and impression share against your target or a sourced benchmark, and labels each as above, within or below. Its suggested tests are kept small and bounded.

• Ask: 'Analyze my Google Ads performance' with a CSV export attached and your target CPA, to get a health check per campaign
• Ask: 'I have $5,000/month for ads, how should I distribute it across Google and LinkedIn?' with a channel rollup and a bounded test plan
• Ask: 'Which ads should I kill?' with an ad-level export, to get investigation candidates rather than automatic cuts

What does the skill need, and what are the safety notes?

The skill needs campaign performance exports or pasted tables from the platforms above, and an AI agent that accepts SKILL.md skills, such as Claude Code. The file does not list specific software libraries. An optional approved, dated benchmark source helps with target comparisons.

The skill does not run scripts and does not use the network. It does not connect to ad accounts or make bid or budget changes. Before analysis, it asks you to remove or mask customer names, email addresses, user IDs and other unnecessary personal data. Treat exports and screenshots as data you check yourself before sharing with an agent, and do not upload campaign data to a third party without consent.

• Runs scripts: no
• Uses network: no
• Does not change bids or budgets in any ad account
• Check exports for personal data before sharing

What tips help you get reliable results?

Give the agent the context it asks for, especially recent changes such as creative swaps or budget shifts, because these affect how far results can be trusted. Treat zero conversions or high CPA as questions to investigate, not proof of waste, since attribution lag and sample size can explain them. Use the skill's bounded tests to confirm a direction before moving real budget.

• Include the date range and any changes made during it
• Provide a target CPA or ROAS with its source and date
• Ask for a significance check before declaring an A/B test winner
• Keep your own judgement in the loop before making budget moves