AI shopping assistants can love a brand and still send its customers somewhere else. This brand shows how.
We asked ChatGPT and Google’s AI Overviews five questions nurses and students ask:
- the best scrubs for nurses
- scrub pants with lots of pockets
- affordable scrubs
- the most comfortable scrubs for 12-hour shifts
- the best scrubs for nursing students
What happened
-
- Named in 8 of 10 answers: all 5 on ChatGPT and 3 of 5 Google AI Overviews. In October it was named in 9 of 9.
- Top pick in 0 of 10. In October it was #1 in 4. This time FIGS took 5 of the 10 #1 spots, Mandala 2, and Dagacci, Healing Hands and Scrubzone one each. The brand usually came 3rd to 5th, as the value or budget choice.
- 1 of 5 results pages showed a shopping ad from its own store, and that ad had no star rating. On the scrub-pants page, the only ad for one of its pants came from a reseller, also without stars, and it appeared twice. The brand’s own unpaid listings on that page showed 4.0 to 4.4 stars
The brand still has what most brands want: AI knows it and trusts it for value. What it lost in a month is the top spot, to brands whose answers came with specific facts.
Why: the winners came with specific facts
Website: strong. In October, Google’s Rich Results Test passed both product pages we tested: names, prices and product identifiers were in the page, and no AI crawler was blocked. One fix since then: the web address of a top we flagged in October as “2-pocket” now says “3-pocket”, matching its product page.
Feed: its own offer barely shows up. Across five results pages, the brand ran one shopping ad from its own store, with no stars. Where a reseller advertised its pants instead, that ad had no stars either and was listed twice. Its own ratings exist; they reach the unpaid listings, not the ads.
Catalog: the facts that win aren’t product fields. The #1 picks came with specifics: an 11-pocket pant, a named performance fabric, a “best for softness” label. For this brand, AI said “affordable”, “practical pockets” and “durable”. On the pockets question, ChatGPT listed its cargo pant last, at 7 pockets, after two pants with 8. In October we found pocket count, stretch and use (nursing school, 12-hour shifts) stored in page copy and collection names, not as product fields the feed and the assistant can compare.
So.. Who fixes what?
| Gap | Owner | First fix |
|---|---|---|
| Ads with no stars, codes in titles, duplicates | Ecommerce / paid search team | Connect ratings to the feed; one listing per product; titles that say what it's for |
| "Pockets," "stretch," "for students" only in copy | Merchandising | Add pocket count, fabric stretch and use (nursing, students, 12-hour shifts) as catalog attributes |
| Generic answers: "affordable, practical pockets" | Merchandising + ecommerce | State pocket count, fabric and use the way the #1 brands do, the same on the page, in the feed and at resellers |
The lesson for every brand that sells through resellers
Being named isn’t being chosen. AI can know your brand and still put someone else first, and a month is enough for the order to change. The fix isn’t more content. It’s making your own product data the most specific and consistent version of your product, everywhere it appears.
Is AI sending your shoppers to someone else?
Email one product URL to tzobrist@zobristinc.com and within a week I’ll send your free AI Answer Check.
