ShelfSeen Tests Whether AI Search Notices Small Shopify Stores

Field notes

ShelfSeen assigns each participating store a treatment product and a comparable control product, using a coin flip to determine which receives a new question-specific page. Queries are written before authority-site coverage is checked, then run five times each through Gemini with web grounding and ChatGPT with search, using fresh sessions and no agent tools. The stated schedule includes two weeks of baseline measurement and four weeks after intervention, with the target product’s top-three appearance and clickable merchant link serving as the deciding outcome.