The Maze: Tesco and Asda both appeared among the top 100 brands named by four AI services in every month of Semrush's UK study. Their websites were far apart as sources: in July, Tesco ranked 25th in ChatGPT's UK citation list and Asda 606th. A mention names a retailer. A citation lists a website as a source for the answer. Semrush, Adobe's search and brand visibility software company, added its first UK benchmark on October 13. The lesson is practical: being talked about and supplying the answer are separate jobs.
Source: Semrush. The official study panel lists findings that held in all seven monthly comparisons, January–July 2026.
The same retailers look different on another AI service. Tesco and Asda ranked 29th and 60th among Gemini's UK sources in July, a much narrower separation than ChatGPT's. Both brands remained in all four platforms' top-100 mention lists throughout January–July. These are recorded citation positions, not judgements about supermarket quality. The result does not tell us that ChatGPT recommended Tesco over Asda, or that Tesco gained more shoppers. It also does not identify why one retailer's pages appeared more often.
The question pools matter as much as the headline ranks. ChatGPT, Gemini and Google's conversational search service AI Mode each received the same monthly pool of 2.5 million UK questions and 25 million US questions, built from Semrush search data. Google's AI Overviews, the summaries shown with search results, used a separate, variable collection of search keywords. Do not add these collections together as one count of unique people or questions. The study examines answers generated for those inputs; it does not represent every shopping conversation. The ranking method separates market, month, industry and platform. Its 22 categories classify websites, rather than every question someone asks.
UK answers agreed less across platforms. The average pair of services shared 63–70 of its top 100 named brands in the UK, versus 70–78 in the US. Sharing a list entry is different from assigning it the same rank. Across the study's categories and platforms, fewer than half the top-mentioned brands also appeared in the matching top-100 source lists in every month, in both markets. An international retailer should therefore check its own country and AI service separately. A strong US result supplies a comparison to investigate, not a UK guarantee.
Seven months of repetition are not seven months of improvement. The June edition covered US data from January–April; the new study adds Britain and extends the window to July. Its monthly ranges show how often a result held, rather than proving growth. The study's stated limits also rule out attributing a commercial result to a visibility change. More source appearances could matter to discovery, but a citation itself records neither a click nor an order. The figures cannot diagnose Asda's website or prove that a particular content change would close its gap.
Why it matters: Retail teams need two separate checks: whether an AI answer names their brand and whether it uses their pages as sources. Run both against the questions, country and services that matter to your customers. If a gap appears, investigate its cause before buying a fix. For readers outside the UK and US, that is a useful audit method; this report supplies no local benchmark. Measure actual visits and purchases separately. An answer that says your name has not yet paid the bill.
Sources: Semrush October release · Methodology · June release · Selected study coverage
Images: 1 Semrush via Semrush AI Visibility Index · Cover AI-generated
