PPC & Ads · Blog post · source Jun 25, 2026
What Went Wrong Pasting a Search Term Report into ChatGPT
Test whether ChatGPT can optimize PPC from a raw search term report
He says it told him to cut terms that were making money and scale ones that weren't, because it optimized ACoS without knowing his cost of goods or fees.
Per source From a vendor-affiliated source (see Worth knowing), not independently verified.
Copy-paste prompt
Here is my search term report [attach] and my per-ASIN break-even ACoS (from price, COGS and fees): [paste]. Judge every term against its ASIN's break-even ACoS, not a single target. Only consider terms with at least [15] clicks. For each term you'd harvest into exact match, also list the negative exact to add in the source campaign. Show the numbers behind every recommendation and don't apply anything.
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
What they did
A seller who also builds PPC software exported the search term report from his own Amazon account, pasted it into ChatGPT and asked what to optimize.
How it went
He says it told him to cut terms that were making money and scale ones that weren't, because it optimized ACoS without knowing his cost of goods or fees. It flagged terms with two clicks and no spend as 'wasteful', missed real money-drainers, and suggested harvesting converting terms into exact match while forgetting to negate them in the original campaign.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
The author sells PPC software, and says so in the post. His fix applies either way: give the AI your per-product profit, fixed rules and harvest-and-negate as one step, and approve every change.