ChatGPT, Perplexity, Claude and Google’s AI Overviews answer by citing a few pages, which they first find through a search: Google’s index for its overviews, Perplexity’s own, OpenAI’s crawler and partner engines for ChatGPT. A store that ranks well starts with an advantage, and the work described above remains the base.
We then check three points specific to assistants. Access: the store’s robots.txt, editable on Shopify, lets the search crawlers of OpenAI, Perplexity and Anthropic through, and the text of each page is in the HTML, because these crawlers do not run JavaScript. Passages: every collection and every guide opens with a clear answer that names the brand and the product and makes sense out of context. The entity: the brand’s name, address and profiles are the same on the store, in the structured data and on third-party listings.
The limits are real. Nobody can guarantee that an assistant will cite a store, or in what position: answers change from one question and one day to the next. Assistants mostly cite third-party sources, comparisons, press, forums and reviews, which are worked on outside the site. Neither the llms.txt file nor structured data has a proven effect on these citations: we set them up for Google and for brand consistency, not as a lever.
Measurement uses a panel of questions put to the main assistants every month, noting whether the brand is mentioned and whether it is cited with a link, and the AI citations report in Bing Webmaster Tools.