What it is
The Mizan Skill is a file you add to Claude that teaches it how to audit your ad tracking and calculate your real ROAS. It contains the same diagnostic logic that a former Meta Ads Growth Lead would run — 14 checks across Meta and Google, plus a revenue reconciliation formula that compares what your ad platforms claim against what your Shopify store actually recorded.
You export three CSVs (Meta, Google, Shopify), upload them to Claude, and the skill does the rest. No API keys, no app installation, no permissions to grant.
What it checks
Meta Ads — 7 checks
- Pixel existence — Is a Meta Pixel associated with your ad account?
- Pixel activity — Has the pixel fired in the last 24 hours?
- CAPI purchase events — Are server-side purchase events reaching Meta?
- Attribution window — Is it set to the recommended 7-day click, 1-day view?
- Data sharing level — Is Shopify's data sharing set to Maximum?
- Event deduplication — Are browser and server events being deduplicated correctly?
- Existing Customers definition — Is the audience defined for Advantage+ Shopping?
Google Ads — 7 checks
- Conversion actions — Do active web conversion actions exist with a purchase category?
- Attribution model — Are all conversion actions using Data-Driven Attribution?
- Conversion window — Is the click-through lookback window set to 30 days?
- Product feed health — Are products approved in Google Merchant Center (<20% disapproved)?
- Feed sync status — Is the product feed syncing regularly?
- Branded search protection — Is PMax excluded from bidding on your brand terms?
- Feed quality — Are GTINs, titles, and images complete?
Revenue reconciliation
The skill compares your Shopify paid orders against what Meta and Google claim, calculates the delta, and shows your real ROAS vs reported ROAS. It flags significant gaps (>20%) as critical tracking issues that need immediate attention.
Works best with Claude
The Mizan Skill is designed for Claude — specifically Cowork mode on Claude Desktop, where skills are a native feature. In Cowork, you just drop the SKILL.md file into your skills folder and say "audit my ads." Claude knows what to do.
That said, the diagnostic logic also works if you paste the skill content into ChatGPT, Gemini, or Perplexity as a system prompt — or simply tell any AI assistant to fetch the guides from this site. The knowledge is open.
How to install
Download the skill
Click the button below to download SKILL.md — a single markdown file containing all of Mizan's diagnostic logic.
Add it to Claude
In Claude Desktop, open Settings → Claude → Skills and drop the file into your skills folder. Or, if you use Claude Code, place it in .claude/skills/mizan/SKILL.md.
Export your ad data
Follow the export guide to download CSVs from Meta Ads Manager, Google Ads, and Shopify.
Say "audit my ads"
Upload your CSVs to a Claude conversation and say "audit my ads" or "what's my real ROAS." The skill handles the rest — running the checklist, reconciling your numbers, and giving you a prioritized action plan.
Not using Claude?
You don't need the skill file to get Mizan's diagnostic logic. All of the checklists and formulas are published as open guides on this site. Tell your AI assistant:
Fetch https://usemizan.com/AGENTS.md and use its routing logic to help me audit my Meta and Google ad tracking and reconcile my revenue against Shopify orders. I'll upload my CSV exports.
Any AI assistant that can fetch URLs will be able to read the guides and apply them to your data.
What's inside
The skill is a single markdown file. It's open, readable, and auditable — no compiled code, no obfuscated logic. Here's what it contains:
- Reconciliation formula with worked examples and interpretation thresholds
- 14 audit checks with severity levels, pass/fail criteria, and remediation steps
- CSV column mapping logic for automatic header detection
- Output format templates for structured reports
- Edge case handling (currency mismatch, single platform, refunds, COD orders)
- Links to the full guides on usemizan.com for deeper context
Who built this
The diagnostic logic comes from Narain Jashanmal, who spent a decade inside Meta building the ads infrastructure that merchants now struggle with — including the Advantage+ suite, Instagram Shopping for Shopify, and Core Ads Growth globally. The checks aren't generic. They encode what Meta's own teams would recommend.