243 leads from $1,493 of ads
The writer behind the Growth Computer newsletter gave a Claude Code agent $1,500 and full control of a Meta Ads account, then walked away for 31 days. Each morning the only human input was one command; the agent read its own logs, pulled fresh performance data, decided what to create, pause or scale, generated the ad images, built landing page variants, and wrote down its reasoning, more than 5,500 lines over the month.
Start with one job only. Fill in the three lines. Keep the bot on that job.
The one job this bot does: ______________________________
What it may do on its own: ______________________________
What it must ask me first: ______________________________
Two lines from this case, side by side. No new figures.
The writer behind the Growth Computer newsletter gave a Claude Code agent $1,500 and full control of a Meta Ads account, then walked away for 31 days. Each morning the only human input was one command; the agent read its own logs, pulled fresh performance data, decided what to create, pause or scale, generated the ad images, built landing page variants, and wrote down its reasoning, more than 5,500 lines over the month.
It spent $1,493 of the $1,500 budget and brought in 243 leads at $6.14 each, against a $2.50 target, so by its own goal it missed. Its best ad hit $1.29 per lead on day 12. The single manual change the owner made, adding business-email validation, spiked cost per lead above $50 and the account never fully recovered. Human time was about 2 minutes a day, versus the 1 to 2 hours a day a media buyer would usually spend.
The after line is the source's claim, not our measurement.
smalloffice.space · Get a case like this in your inbox each morning. Subscribe free
Read this case on your phone now, and keep it for later. Type your email. We send this case to that address, and you get the next cases each morning too.