Instead of staring at Ads Manager trying to figure out what's working, you're gonna export your data, drop it into Claude Code, and let your agent tell you exactly what to kill, scale, and do next. Takes 2 minutes once you get the flow down.
📊 The Process
This is your weekly (or daily) routine. Three steps, no setup, no tools to install.
- Open Meta Ads Manager
- Select your date range — last 7 days is a good default
- Make sure these columns are showing: Campaign name, Ad set name, Ad name, Amount spent, Results, Cost per result, CTR, CPM
- Click Reports → Export Table Data → CSV
- Save the file
2Drop It Into Claude Code
Drag the CSV file into your Claude Code workspace folder. That's it. Your agent can now read the file.
If you named it something like ads-march-9.csv, your agent will see it automatically.
3Ask Your Agent to Analyze
Open Claude Code and tell it what you want to know. Here's a solid starting prompt:
read ads-march-9.csv and give me a full breakdown.
my target CPL is $25.
tell me:
- which ads are profitable and which are bleeding money
- which ad sets i should kill immediately
- which ads are winning and should get more budget
- what i should do tomorrow based on this data
use the actual numbers. be specific.
💬 What You'll Get Back
Your agent reads the entire file and gives you a breakdown like this:
"Campaign 'Local Dentist - Broad' spent $142 with 3 leads at $47.33 each — that's almost 2x your target. Ad set 'Interest Stack Cold' spent $89 with zero results, kill it. Your best performer is 'Retargeting - Video Viewers' at $12.50 per lead, well under your $25 target. I'd scale that one 20% and pause everything in the cold interest stack. For tomorrow, focus your budget on retargeting and test 3-4 new hooks in a fresh ad set."
💡 The key: Always tell the agent your target CPL. Without a target, it can't tell you if something is good or bad — it's just numbers without context.
🔥 Other Things You Can Ask
Once your data is in, try these:
"compare this week's data to last week. am I trending up or down?"
"my client wants to know why CPM went up. explain it in a way I can tell them."
"if I had to cut my budget in half tomorrow, which campaigns would you keep?"
"write me a weekly report for my client based on this data."
"which creative angles are working best? what should my next batch of ads focus on?"
The agent does the math and the analysis. You make the final call.
📅 How Often Should You Do This?
📆 Weekly (minimum)
Export every Monday morning. Review the past 7 days. Make your kill/scale decisions for the week. Send your client a report. This is the baseline.
📊 Daily (recommended)
Quick 2-minute check. Export, drop in, ask "anything I need to act on today?" Catches problems before they burn through budget. This is what the best media buyers do.
💡 Pro move: Keep your CSV files organized. Name them by date: ads-2026-03-09.csv. That way when you want to compare weeks, you can tell Claude "compare ads-2026-03-09.csv with ads-2026-03-02.csv" and it'll show you the trend.
⚠️ Common Mistakes
⚠️ Don't just say "analyze my data" — be specific about what you want. Include your target CPL, what you're optimizing for, and what decisions you need to make.
⚠️ Don't panic over one day — ad accounts need 3-7 days minimum to show real patterns. One bad day doesn't mean the campaign is dead. Export a full week before making big decisions.
⚠️ Don't forget the columns — if you export without CPM or CTR columns, the agent can't diagnose creative problems. Make sure all the key metrics are in your export.
🚀 What's Coming: Automatic Data
⚡
Adlevel — No More Manual Exports
Right now you're exporting CSVs manually. That works and it's what you should learn first.
But Adlevel connects directly to your ad accounts and feeds data to your agents automatically. No downloading, no dragging files. Your agents just watch your campaigns in real time and flag things as they happen.
Once Adlevel is set up, this entire lesson becomes automatic. But learning the manual way first means you actually understand what the data means — which makes you a better media buyer whether you use the automation or not.
✅ Checklist
- ☐ Exported campaign data from Ads Manager as CSV
- ☐ Dropped the CSV into your Claude Code workspace
- ☐ Asked your agent to analyze with your target CPL included
- ☐ Got a kill/scale/watch recommendation from the agent
- ☐ Decided on your frequency: daily or weekly exports
🏷️ Tags
aidata-analysisclaude-codecsv-exportads-managerreportingadlevel