Category: Sanitizing & Hygiene Wipes · July 30, 2026 · published in full, failures included: this is our own brand
When shoppers ask an AI assistant what to buy in your category, the answer is a short list of brand names. We asked the 12 questions your buyers actually type, live in Claude and Gemini, and extracted every brand named in every answer.
answers named MagiCare. The board below shows who the engines recommend instead.
| Wet Ones | 18 | |
| Purell | 14 | |
| Babyganics | 8 | |
| WaterWipes | 8 | |
| Care Touch | 7 | |
| Clorox | 7 | |
| Amazon | 6 | |
| Lysol | 5 | |
| Pampers | 5 | |
| Honest Company | 5 | |
| MagiCare | 0 |
Mentions across 24 live AI answers. Same questions, same wording on every future scan, raw answers included, so progress is measurable and nobody moves the goalposts.
Full raw answers available on request.
Getting named is half the game. When the shopper's agent arrives, it has to read your store, quote your policies and complete a purchase. We tested that too.
agent-commerce readiness checks passed.
| ✓ | AI crawlers can read the store | GPTBot-class crawlers get a normal page, not a block |
| ✓ | UCP discovery profile is published | agents can find the store's commerce endpoints at /.well-known/ucp |
| ✓ | Agent shopping endpoint (MCP) is live | catalog search, product details and cart are callable by agents |
| ✓ | Current-generation UCP endpoint responds | the endpoint replacing legacy MCP (deprecated Aug 31, 2026) is reachable |
| ✓ | Agent knowledge base has real answers | returns and shipping answered with real, specific policies |
| ✓ | Agent instructions files exist | llms.txt tells visiting agents how to interact with the store |
| ✓ | Product schema on product pages | product schema with offers found on 4 of 4 products checked |
| ✓ | Review ratings visible to machines | review scores are in the product schema on 3 of 4 products checked, so engines can cite them |
Correction, August 13, 2026: this section originally read 7 of 8, with "Review ratings visible to machines" marked failed. That was our measurement error, not the store’s — our check sampled a single product (a newly launched item with no reviews yet) and reported it as a store-wide verdict. It now samples five products and reports the ratio. Re-measured August 13, 2026: 8 of 8. The shelf-score section above is unchanged and was not affected.
AI referred traffic to US retail grew 393% year over year (Adobe, Q1 2026) and converts 54% better than other channels (Adobe, Q2 2026), because the shopper arrives already holding a shortlist. Being absent from the shortlist means you were never considered.
The fix is real work the engines actually read: a machine readable store, reviews as structured data, comparison content on the questions nobody defends. That is what our 21-Day AI Shelf Sprint ships. Positions in AI answers are decided by the engines; nobody can buy a ranking, including us.