PPC & Ads · from a YouTube video, Mar 18, 2026 · Added
Zero-Order Search Term Themes and Negative Phrases with ChatGPT
Find the wasted-spend patterns hidden across thousands of zero-order search terms and turn them into negatives
“$2,200 in wasted ad spend”
Per source Reported by an agency about its own client work, not independently verified.
Copy-paste prompt
I'm selling [product, e.g. women's yoga pants]. Here's a list of search terms that spent money but generated zero sales over the last 60-90 days. Analyze these terms and identify common themes or attributes (material, price point, audience, style, etc.) that aren't relevant to my product. Group them by pattern and calculate the total wasted ad spend for each pattern. Then suggest negative phrase keywords for each irrelevant pattern, and flag any pattern that might be seasonal so I can check it against past data before negating.
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
See the original YouTube video
What they did
Vadim Soin of the agency PPC Jumpstart exports 60-90 days of search terms with spend above zero and zero orders (he uses AdLabs; the Amazon Ads console report works too) and gives them to ChatGPT with a prompt naming the product, women's yoga pants in the demo, asking it to group the terms by theme or attribute and total the wasted spend per pattern. He builds negative phrase lists from those themes plus competitor brand names, then checks every negation against historical data before applying it.
How it went
In the example, $7,300 had gone to zero-order search terms over 90 days. ChatGPT grouped them into themes such as male intent ("$861"), skirts and kids, and he says that by that point they had identified $2,200 in wasted ad spend. One term he planned to negate showed $69 in historical sales on $219 of spend, so he left it off the list.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
He warns that AI lacks nuance: it may tell you to negate a seasonal term like 'gift' when you analyze February data. The prompt below is cleaned up from the wording he reads out in the video. The agency's client results are shown on screen only.