This is the exact prompt from the video, plus the checks to run before you act on the report. You need the Claude desktop app and a paid Claude plan.
Set it up
You'll need the same dates later to check the totals. The report can only use what's in the file, so make sure it has Amount spent, Results and Cost per results.
WhereAds Manager › Ads tab
Claude reads the whole file. In the video, the ad names became Ad 01 to Ad 18 and every number stayed real.
The prompt below uses that file name.
In the desktop app, Claude Code is the Code tab, and a paid subscription is required.
WhereClaude desktop app › Code › project folder
Manual asks before editing files or running commands. Auto runs without routine prompts. The run in the video used Auto.
The prompt
The line 'Use only what is in the file' tells it not to fill the gaps with guesses, and to say so when the file can't answer.
This folder has meta-ads-export.csv, an export from Meta Ads Manager. Explain what you do in plain English, with no coding terms. Build me a small tool that reads the export and makes a one-page report called report.html showing where this account is wasting money. The report should show: - total spend, total leads and the average cost per lead - every ad that spent money, sorted from the worst cost per lead to the best, compared with the average - how much the ads above the average cost beyond what they would have cost at the average - the three ads I should look at first, with one plain line each saying why Use only what is in the file. Don't guess at anything it doesn't say, and say so if something can't be worked out from it. Before you build it, tell me your plan in three short lines. When it's done, check your totals against the file and tell me how I can check them myself in Ads Manager.
Check the result
In the video's run, the first check printed nothing and Claude said so, then ran it again. Don't assume a check passed until it says what it compared.
Claude can't see your account. In the video's run it said it hadn't tried its own Ads Manager steps and that the labels may differ, so this one is on you.
WhereAds tab, no filters, same dates, same attribution setting, then the totals row at the bottom of the table
Pick the biggest spender in the report. Its spend, results and cost per result should match its row in Ads Manager.
Every total in the report only means the ads in that file. A small difference in leads may just be late credit, because an ad can be credited up to 7 days after a click on a 7-day click setting.
Read it honestly
The export has no sales figures, so an expensive lead could still be a good one.
It's the gap between ads, not a forecast of what moving the money would do.
The file in the video had no dates per ad, so those ads may not have run at the same time.
In the video's file, every result was a lead. If your file mixes result types, their costs won't compare.
Use it again
In the video's run, on a Mac, Claude made a make-report.command file that you double-click. Yours may be different, so ask Claude how to rerun it if it doesn't say.
Want me to look at yours?
If you have gone through this and want a second pair of eyes, book a free call. We open the account together and look at what is actually happening. No deck, no pitch.
A new free sheet most days. Get the next one: follow @ihamza.jameel on Instagram · YouTube
More free sheets
- RETENTIONAI Chat Privacy ChecklistNew · Checklist · 21 checks · 18 Nov 2026
- FILTERSSearch Console New Filters ChecklistNew · Checklist · 14 checks · 16 Nov 2026
- HUBChatGPT Business Hub Setup ChecklistChecklist · 13 checks · 11 Nov 2026
- TRACKINGGoogle Ads Tracking Check: When Conversions DropChecklist · 21 checks · 11 Nov 2026
