Guide: Building Meta Ads Performance Dashboards with AI

Traditional Meta Ads performance analysis involves manually exporting data, building spreadsheets, calculating metrics like ROAS and CPA, identifying underperforming campaigns, and determining which ads to scale or kill. This process typically requires advanced Excel skills, deep media buying knowledge, and hours of manual work to transform raw campaign data into actionable decisions.

AI assistants like Claude and Gemini can automate this entire workflow by analyzing your Meta Ads exports, calculating advanced performance metrics (Hook Rate, Hold Rate, Conversion Rate), diagnosing issues using media buying logic trees, and providing specific campaign recommendations—all in under 5 minutes.

This guide provides complete instructions for transforming your Meta Ads data into executive-ready dashboards using either Claude or Gemini.

What You'll Need

Required:

• Meta Ads data export (CSV format)

• Export from Meta Ads Manager → Reports → Export

• Include columns: Campaign Name, Ad Set Name, Ad Name, Spend, Revenue, Impressions, Link Clicks, Purchases, 3-Second Video Plays, ThruPlays

Optional Context:

• Breakeven ROAS target (defaults to 1.6 if not specified)

• Time period being analyzed (e.g., "Last 30 days", "Q4 2024")

Time Investment: 5-10 minutes per complete analysis

Option 1: Using Claude (Recommended for Complex Analysis)

Access and Setup

1. Navigate to claude.ai

2. Log in to your account

3. Start a new conversation

Upload Your Data

Preferred Method - File Upload:

1. Click the paperclip icon (📎) or drag-and-drop your CSV file

2. Claude automatically reads file contents

Alternative - Direct Paste: For files under 100 rows, paste data directly as text

Provide Claude with this system instruction that defines its role and diagnostic framework:

You are an expert Senior Media Buyer and Data Analyst specializing in E-commerce Meta Ads (Facebook/Instagram). Your goal is to audit ad data, build a dashboard, and provide high-level strategic recommendations.

You do not just report numbers; you diagnose the "why" behind performance and prescribe the "what next."

User Context:
I am an e-commerce brand owner. I will provide you with my Meta Ads campaign data (either as an uploaded CSV file or pasted text).

Logic Framework:

1. Core Metric Definitions & Formulas (Calculate these internally):
   - Breakeven ROAS (BE-ROAS): Assume 1.6 unless specified otherwise.
   - Hook Rate (Thumbstop Ratio): (3-Second Video Plays / Impressions) × 100. Benchmark: >25% is good.
   - Hold Rate: (ThruPlays / 3-Second Video Plays) × 100. Benchmark: >30% is good.
   - Conversion Rate (CR): (Purchases / Link Clicks) × 100. Benchmark: >2% is healthy.
   - Click-to-Purchase Ratio: Link Clicks / Purchases.

2. Diagnostic Logic Tree (Used to determine recommendations):
   - IF ROAS > Target AND Spend is Low → SCALE (Increase budget by 20%).
   - IF ROAS < Target AND Spend is High → KILL (Pause immediately to save budget).
   - IF CTR is Low (<0.8%) AND CPM is Normal → Creative Fatigue/Boring Ad. → Test new creative visuals.
   - IF CTR is High (>1.5%) AND Conversion Rate <1% → Landing Page/Offer Mismatch. → Audit product page speed and offer clarity.
   - IF CPM > $40 → Audience Saturation or Poor Creative Ranking. → Broaden audience or completely refresh creative.

Output Formatting:

Please organize your response into the following four sections using Markdown:

SECTION 1: EXECUTIVE HEALTH CHECK
- Total Spend: $
- Total Revenue: $
- Blended ROAS: (Revenue / Spend)
- CPA (Cost Per Acquisition): $
- Global Sentiment: (e.g., Bullish, Concerned, Stable) — Provide a 2-sentence summary of the account health.

SECTION 2: CAMPAIGN DASHBOARD (The "Meta Report")
Create a visual dashboard with insightful graphs. Also create a Markdown table ranked by Amount Spent (High → Low):

| Campaign Name | Spend | ROAS | CPA | CTR | CPC | Hook Rate | Verdict (Scale/Kill/Optimize) |
|---------------|-------|------|-----|-----|-----|-----------|-------------------------------|
| [Row Data]    | ...   | ...  | ... | ... | ... | ...       | ...                           |

SECTION 3: CREATIVE DEEP DIVE
Identify the single Best Performer and Worst Performer.
- The Winner: Ad Name — Why it won (Hook Rate vs. Conversion Rate insights).
- The Loser: Ad Name — Why it failed (Was it the click or the conversion?).

SECTION 4: STRATEGIC RECOMMENDATIONS
Provide three specific, actionable bullet points describing exactly what to do inside Ads Manager right now.

Examples:
- "Turn off Ad Set X."
- "Launch a new angle for Product Y."
- "Duplicate Campaign Z with 20% higher budget."

Trigger:
Please analyze the attached data/text following the logic above.
If no data has been uploaded yet, ask the user to upload their export file.

After providing this instruction, type:

Please analyze this Meta Ads data following the diagnostic framework.

What You'll Get

Claude will generate:

• Executive Health Check with blended metrics and sentiment analysis

• Campaign Dashboard with sortable table showing all key metrics

• Creative Deep Dive identifying top winner and worst performer with diagnostic reasoning

• Strategic Recommendations with specific Ads Manager actions

For the complete SOP including Gemini setup, refinement prompts, and advanced dashboard features, click here.

• 74% of marketing emails have unusable link text for screen readers, so cleaner copy, fewer emojis, shorter sentences, better buttons, and real ALT text can lift accessibility and conversions fast.

• A Claude Code + Opus 4.5 walkthrough shows how agents work, how Skills plug in, and how to build automations like posting to X from inside your workflow.

• Kapwing’s research suggests AI slop and brainrot are flooding YouTube, with some countries and channels racking up billions of views and millions in estimated ad revenue.

• A contrarian view says ChatGPT is basically a search engine with a multi-model retrieval pipeline, so winning means being fast, fetchable, and easy to ground.

• OpenAI’s “Project Gumdrop” may land as a smart pen or pocket audio gadget that ships notes straight into ChatGPT, with Foxconn shifting production away from China and a 2026 or 2027 target.

• Where AI is headed in 2026

• Google’s “Nano Banana” moment helped Gemini surge, powered by DeepMind talent, heavy TPU investment, and faster shipping, as Alphabet fights to keep Search monetised while closing the gap on ChatGPT.

• CES 2026 wrapped with a wave of practical wow

• Claude skills is a must learn!

  

• DeepMind and Boston Dynamics are wiring Gemini Robotics into Atlas and Spot, aiming to turn flashy demos into factory ready dexterity, starting with Hyundai plants and a big focus on real world safety.

• CES buzz for marketers was all about agentic AI buying media, retail media going off site, and creators scaling into full media businesses, while Reddit’s new Max Campaigns leans into transparency to win ad trust.

📊 CapGo AI: Make market research effortlessly fast. Seamlessly gather vast web information into spreadsheets in seconds.

✒️ Dittto - Fix your hero copy with an AI trained on top SaaS websites.

✨ Chat Ur Data GPT - Boost productivity and unlock data insights with AI-driven assistance

🌄 Before Sunset: An AI personal assistant that plans your day by automatically generating daily schedules.

🪄 InMagic - Uncover your personality based on your Instagram profile

💰 TaxGPT: A tax assistant that makes the boring stuff simple.

😍It was great, all I needed

🙂Good, not great

🙄It sucked