Most of the grunt work in media buying isn’t strategy — it’s repetition. Pulling the same reports, re-reading the same campaigns looking for a pattern, writing creative briefs from scratch every week. That’s the part I’ve started handing to Claude Code, and it’s changed how much a small team can actually get done.
What it’s actually good for
Claude Code sits alongside my Meta workflow and does the first pass on things a junior buyer used to do: pulling account data, flagging which creatives are actually driving results versus just spending, and drafting a testing brief based on what’s worked before. It doesn’t replace judgment — it just gets me to the decision faster.
How I set it up
I feed it real account history: winning ads, ads that flopped, audience data, offer details. The more specific the examples, the better it gets at recognizing what “good” looks like for this specific brand, instead of giving generic advice that could apply to anyone.
What it doesn’t replace
Strategy still needs a human. So does creative production and reading the room on brand voice. What it removes is the slow, repetitive analysis loop — the stuff that used to eat a whole afternoon before you even got to the interesting decisions.
Where to start
If you want to try this yourself: pull your best and worst performing campaigns from the last 90 days, export the creative and the numbers together, and use that as the training context. Then ask it to flag patterns before you ask it to generate anything new — the pattern-finding is where most of the time savings actually live.
Related reading: