Is traditional search engine optimization, SEO, dead now that consumers are increasingly turning to AI services for product recommendations? That was the million-dollar question when moderator Ola Andersson welcomed Jorge Castro from Growth Marketing Sweden, Bård Farstad from Crystallize and Jimmy Bergstedt from Etals onto the stage to discuss the new keyword battle.
The panel’s answer was a resounding no. Instead of dying out, the discipline has evolved into what is now called GEO, Generative Engine Optimization, where traditional craftsmanship forms the very foundation for even being discovered by the new language models, so-called LLMs.
“No, not at all. In fact, it’s becoming more important than ever, because what we’re seeing is that good SEO is what drives good GEO,” states Jorge Castro, founder and AI SEO specialist at Growth Marketing Sweden, who recently presented a mapping on Ehandel.se of the e-retailers that dominate Google’s AI surface.”
Two Paths into Language Models
To understand how e-retailers can influence AI assistants like ChatGPT to recommend a certain brand, it is important to distinguish how the technology works behind the scenes. Jorge Castro explains that there are two phases to consider: the fixed training data and the so-called retrieval phase, often abbreviated RAG (Retrieval-Augmented Generation).
While the training data can only be influenced in the long term before the language models’ next learning period, the retrieval phase takes place in real time. This is when the AI does not have the answer in its existing database but goes out on the web to retrieve and index fresh information that meets the user’s specific query.
By actively working with mentions, so-called mentions, and digital PR on external sites and in the right forums, you can achieve quick results in this phase. Jorge Castro highlights e-retailer Motion Fitness, which has a turnover of just under 50 million Swedish krona, as a concrete example. The company had a brand that did not appear when customers asked AI services for tips on building a home gym. By working purposefully with digital PR and mentions in the right context, they were able to see how the AI responses changed to the company’s advantage within just one week.
The same pattern is seen in significantly larger players such as Däckline, with a turnover of over 100 million Swedish krona, where external mentions have been the single most important factor for a year. A study from the analysis company Ahrefs, which reviewed 75,000 web pages, confirms the picture: the time when you could fill your own pages with long, keyword-rich texts to rank on narrow search phrases is over in the AI era. Now it’s about building authority and credibility across the web.
“The more you look into the optimization phase itself, you start to realize: wait a minute, this is just classic SEO, only it’s more modern SEO. Because now it’s not about Search Engine Optimization, it’s about Search Everywhere Optimization. The LLMs retrieve data from Youtube, from forums, from someone who tweeted – it comes from everywhere,” says Jorge Castro.
Fewer Clicks – But Extremely Warm Leads
The change in consumer search behavior is clearly noticeable in e-retailers’ traffic flows, where traditional organic traffic from Google is pointing downwards for many. But Bård Farstad, CEO of the e-commerce platform Crystallize, believes that the coin has a very lucrative flip side. The traffic that actually clicks through from an AI assistant to the online store has a completely different purchase intention.
According to studies from both Ahrefs and Adobe, visitors coming via language models convert up to 42 times better than traditional search traffic.
“Think of it like having a salesperson out there. The customer who searches chats and eventually leads to a site, and then they are quite warm. The lead is quite warm compared to the traditional search where we do a search, ten blue links, look around,” explains Jorge Castro.
This, however, places completely new demands on the e-commerce experience when the customer lands in the store. According to Bård Farstad, we are now in a shift where the entire value chain is becoming agentic – both how you build and run the site, but also how products are discovered.
“Without getting too nerdy, it’s about having control over product data. It must be super structured and semantic so that robots can read it. But then it’s very easy to focus only on robots or agents that crawl. You also have to think about us humans who land there, who get a brand experience and product storytelling. You have to optimize for both,” says Bård Farstad.
AI as an Engine for Structured Product Data
Getting your product data in order is easier said than done, and it is often a hugely time-consuming bottleneck internally at many retailers. Here, Jimmy Bergstedt, CEO and founder of Etals, believes that AI itself is the solution to the problem. By using AI tools, e-retailers can both conduct in-depth research into what questions customers are actually asking about a particular product category, and automate the heavy data work in the background.
“You usually have supplier data that is unstructured, which you then have to get structured into your own systems. And AI is, I think, a fantastic tool that can help companies with these processes,” says Jimmy Bergstedt, pointing out how the technology can retrieve raw data from suppliers and automatically structure it into the business systems while creating content that genuinely helps the customer.
Once you have structured your data, the door also opens to building your own smart shopping assistants on your own site. Bård Farstad highlights the garden chain Plantagen as a success example. To drive omnichannel sales and get customers into the physical stores, Plantagen has structured all its product data, guides and care instructions. This not only makes them visible in external AI searches, but also drives their own AI agents directly on the site, where customers can receive customized advice for their specific garden and growing zone before the weekend planting.
Unique Internal Data Beats AI Spam
During the debate, a critical audience question was also raised regarding credibility in the new AI world: how can mentions and mentions continue to be reliable as a ranking factor when anyone can build AI agents that spit out mentions on the fly?
According to Jorge Castro, the winners in the long run will not be those who spam the most, but those who work smartest with genuine PR and unique content. The key is to dig where you stand and use the unique data you already have internally. At Däckline, for example, they chose to analyze data from their own customer service to see what questions and technical problems customers most often called about. Based on that, they created in-depth guides and made PR efforts around the topic – which both relieved customer service and built strong authority in the eyes of language models.
“It’s who becomes best at PR, who becomes best at being visible. That’s the one who will be able to feed the LLMs best. The one who does it best is, for example, if you share internal data and do studies. That’s what provides content and knowledge, but also sets an entity around your brand,” says Jorge Castro.
When asked how e-retailers can organizationally succeed in navigating a technological development that is moving so quickly, the panel agrees that long-term, set-in-stone strategies must be replaced by constant experimentation.
“I think it’s important to stay updated on what’s happening, but also to keep testing yourself. Start measuring, see how you’re doing today, keep an eye on the news around this and test yourself continuously,” advises Jimmy Bergstedt.
Bård Farstad concludes by sending a clear message to e-commerce leaders about the company culture required in 2026 and beyond:
“The one who is most willing to change will win. What is true now is not true in three months, it is not true in six months. So having a change culture, I think, is important.”