About one in four small and mid-sized Swedish online stores never came up when ChatGPT was asked where to buy what they sell, according to a study by KLOTA, a digitalmarketing agency in Borås that sells help with exactly this. Of 189 stores with SEK 3 million to 30 million in revenue, 28% were not named in any of ChatGPT’s three answers, and 29% were not named by Claude.
The study links visibility to the text on a store’s home page, not to size. 26% of stores with little text and no headings written as questions were never named, against 6% of stores with at least 500 words and question headings: by our arithmetic, about 4.3 times as often. KLOTA says the data shows the two are related, not that one causes the other.
Business Pill · Agentic Commerce
An evergreen Business Engineer explainer on agentic commerce: AI assistants and agents that research and buy on a shopper’s behalf. It explains the general idea only and does not discuss KLOTA or any store in this piece.
The key insight: In KLOTA’s test of 189 small and mid-sized Swedish online stores, 28% were never named by ChatGPT and 29% never by Claude when asked where to buy what they sell. 26% of stores with little text and no question headings were never named, against 6% of stores with at least 500 words and question headings: by our arithmetic, about 4.3 times as often.
Anders Karlsson, who runs KLOTA, ran the stores through the agency’s AI visibility analysis, version 2.3, on 4 November 2026. For each store a language model wrote three Swedish buying questions from the store’s home page, without the store’s name. They were put to Claude (Sonnet 5.5) and ChatGPT (chat-latest), both with web search: six answers per store and 1,123 answers in total, KLOTA says.
A store counted as mentioned when its name or domain appeared in the answer. KLOTA says it paid for the study itself, that no store paid to be included and that it names no individual store. Ehandel.se reported the findings on 6 November.
Share of the 181 small and mid-sized Swedish online stores with readable home pages that ChatGPT and Claude never named in six buying answers, by home-page type, from KLOTA’s 4 November test. The 4.3x is our arithmetic; KLOTA says the groups show a link, not cause.
Who Gets Named Instead
When the AI did not name the store, it recommended one of Sweden’s 100 largest online retailers in 4 cases out of 10, KLOTA found. The names that came up most often were Elgiganten, Bygghemma, Apotea, CDON, Kjell and Zalando.
The two assistants did not agree. Claude named the stores in 47% of its answers and ChatGPT in 39%, an 8-point gap by our arithmetic, and one store in five was named by only one of the two services, according to the study. 19% were named by neither.
What Went With Being Named
Stores with under 220 words on the home page were named in 2.4 of 6 answers on average; stores with over 970 words in 3.3 of 6, KLOTA reported. Stores with at least one heading written as a question were named in 3.4 of 6 answers, against 2.6 for those without.
A third of the stores had an llms.txt file, which is meant to help AI services read a site, but 16% of them were never named against 15% of stores without one. KLOTA found that 98% of stores let the AI services’ robots in through robots.txt, and that store revenue did not reliably change how often a store was named.
As we read the study, the stores that went unnamed were not shut out: KLOTA found 98% let the AI services’ robots in, so access was not what separated them.
As we read KLOTA’s figures, the two assistants are not one channel: Claude named the stores in 47% of answers and ChatGPT in 39%, and one store in five was named by only one of them.
As we read the study, a small store that is not named is often replaced by a large one: when a store was left out, the assistants pointed to one of Sweden’s 100 largest retailers 4 times in 10, KLOTA found.
Anders Karlsson, KLOTA, via Ehandel.se (our translation)
“Many e-retailers are spending time on technical AI tricks right now. Our data suggests that what actually makes a difference is the plain old thing: text that answers customers’ questions.”
Three Implications
FOR SMALL ONLINE STORES KLOTA found stores with more home-page text and question headings were named more often, and says the link does not show cause.
FOR THE LARGEST RETAILERS When a store was not named, the assistants recommended one of Sweden’s 100 largest online retailers in 4 cases out of 10, KLOTA reported.
FOR LLMS.TXT A third of the stores had the file, but 16% of them were never named against 15% without it, KLOTA found.
The framework argues, describing the medium cycle of industry restructuring: “Search gives way to assistants.”
As we read it, the KLOTA study is a small measurement of that shift from the merchant’s side: who an assistant names when a shopper asks where to buy. The framework predates this study and does not discuss Swedish retail.
What Is Not Established
KLOTA sells AI visibility services, so it has an interest in the result, and it says so. The study is a single measurement on 4 November; KLOTA notes that AI answers vary between runs and that answers through the API may differ from the apps. It covers Swedish stores with SEK 3 million to 30 million in revenue that show Swedish text and prices in kronor, and the links it reports between text and visibility do not show cause. The figures here are taken from KLOTA’s Swedish-language page in our translation; we did not contact KLOTA.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
In KLOTA’s test of 189 small and mid-sized Swedish online stores, 28% were never named by ChatGPT and 29% never by Claude when asked where to buy what they sell. When a store was left out, the assistants pointed to one of Sweden’s 100 largest retailers 4 times in 10. Stores with more home-page text and question headings were named more often; by our arithmetic, stores with little text and no question headings went unnamed about 4.3 times as often.
A note on sourcing. The figures are from KLOTA’s Swedish-language study page of 7 November 2026, read in full, in our translation, and from Ehandel.se’s 6 November report. KLOTA sells AI visibility services and paid for the study itself. The 4.3x ratio and the 8-point gap are our arithmetic. We did not contact KLOTA.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
