As AI services have become a real search and recommendation channel for Swedish consumers during this year's Black Week and Christmas shopping season, it's clear that not all online stores are included in the responses when consumers ask AI for purchasing advice.
In early November, the marketing agency Klota conducted a survey testing 189 Swedish e-commerce businesses with a turnover between 3 and 30 million Swedish krona. By asking niche-specific purchase questions about exactly what the stores sell to ChatGPT and Claude, and analyzing a total of 1,123 AI responses, the aim was to see how well smaller and medium-sized players are asserting themselves in this new channel.
The Giants Take the Place
The results show that approximately one in four stores – 28 percent in ChatGPT and 29 percent in Claude – were not mentioned even once in the AI responses. Instead, it is the very largest players that dominate the recommendations. In 4 out of 10 cases where the smaller, niche store was omitted, the AI chose to recommend one of Sweden's 100 largest e-commerce companies, such as Elgiganten, Bygghemma, Apotea, CDON, Kjell & Company and Zalando.
How Swedish online stores perform in the new AI tools is a question that engages the entire industry right now. Ehandel.se recently highlighted the development through a mapping from Topdog which showed how 45 of 50 Swedish e-commerce giants suddenly lost visibility in Google's own AI responses, AI Overviews. SEO expert Jorge Castro has also mapped in a previous analysis how the largest e-commerce companies own Google's new AI surface. While previous studies have focused on the industry's giants and Google's search results, Klota's new review sheds light on the conditions for smaller and medium-sized e-commerce businesses within the dedicated AI chats.
Technical Shortcuts Have No Effect
A common perception in the industry is that the lack of visibility in AI tools is due to technical obstacles or that the store is too small to be noticed by the language models. However, Klota's review shows that this is not the case. Turnover size within the range of 3 to 30 million krona had no bearing on how often the stores were recommended, as stores with a turnover of 12 to 30 million krona were not seen more often than those with a turnover of 3 to 6 million krona.
Nor were technical blockages behind the invisibility, as 98 percent of the reviewed sites allowed the AI robots in via their robots.txt file and no server blocked the traffic. The investigation also debunks the myth of the talked-about llms.txt file, a technical shortcut that is said to help AI models read a site. One-third of the stores in the study had added the file, but they were neither mentioned more or less often than others, which is fully in line with Google's previous statement that no AI systems actually use the file.
The Text on the Homepage
During Stora Ehandelsdagen (The Big E-commerce Day) earlier this year, the concept of GEO, Generative Engine Optimization, was discussed and how important it has become to optimize your content to gain visibility in AI's responses. Klota's investigation confirms that it is precisely the editorial work and the basic structure of the site that determines the outcome.
The clearest success factor in the study was the amount of text and how the headlines were designed on the homepage. Among e-commerce businesses that had at least 500 words on their homepage, combined with headlines formulated as direct customer questions – for example, how to choose the right size – almost all were mentioned in the AI responses. In this group, only 3 out of 50 stores were never seen. Among the stores that, on the other hand, had little text and lacked question headlines on the homepage, one in four received no mentions at all.
Anders Karlsson at Klota, who is behind the investigation, believes that e-commerce businesses should focus less on technical shortcuts and instead prioritize their basic content.
Many e-commerce businesses are spending time on technical AI tricks right now. Our data suggests that what actually makes a difference is the old-fashioned way: text that answers customers' questions. Those who write more and better texts probably also work more on their visibility in general, so the text is probably not the whole explanation. But that's where most can start.
Simple Measures
In addition to coherent texts and clear headlines, Klota found another concrete shortcoming at many of the investigated sites. As much as 40 percent of the e-commerce businesses lacked clear information about shipping, delivery times or returns in the readable text on the homepage.
When a consumer asks ChatGPT or Claude for tips on where to buy a specific product ahead of Christmas shopping, the AI model looks for security factors and concrete purchase conditions in order to provide a good recommendation. For the e-commerce business that wants to increase its chances of being seen in AI responses ahead of this year's most important sales weeks, the advice is therefore uncomplicated: make sure the homepage has a well-written text that explains what the store sells, use headlines that answer customers' most common questions and write out the terms for shipping and returns in plain language.