The average search query handled by a large language model runs roughly 24 words, four times the six-word norm of traditional Google searches. That figure, cited by Lucas Tieleman, CEO and co-founder of Optiversal, on the latest episode of Conversations with MarTech, signals a structural shift in how consumers find products online — and in what retailers must do to stay discoverable.

From keyword queries to rambling questions

Shoppers no longer feel the need to "dumb down" their search queries, Tieleman explained. Instead of rigid keyword strings, they now input detailed, conversational questions packed with context, adjectives, personal brand preferences, specific use cases and even family dynamics. They expect the search platform to understand all of it.

"The same product can be different things to different people," Tieleman said during the episode. That reality breaks the old model of static catalog and product pages built around a fixed set of keywords.

Websites as living organisms

AI has collapsed the incremental cost of content creation. Tieleman argued that retailers can now treat their websites as living organisms, building and managing custom landing pages at scale to match shifting social trends and natural-language queries — something prohibitively expensive under the old rules of catalog construction.

Bots, not humans, are driving traffic

Autonomous AI agents are not yet completing purchases on consumers' behalf, Tieleman noted. But they are heavily driving research traffic. Bot traffic is soaring as LLMs crawl retail sites to answer user prompts, which forces marketers to rethink the relationship between impressions and sales. A visit no longer reliably means a human saw the page.

Feeds matter more than ever

To remain visible in AI-mediated search, retailers must feed highly structured data, conversational attributes and detailed FAQs directly into AI and advertising platforms, Tieleman said. Product feeds optimized for keyword matching will not carry over cleanly into a world of 24-word conversational queries.

"The biggest lie in marketing"

Asked what the marketing industry tells itself that simply is not true, Tieleman pointed to the obsession with finding "one right answer." Modern technology lets marketers test multiple paths simultaneously and fail cheaply. "With the technology that you have now, you can be wrong a lot more efficiently," he said.

He closed with a provocation on design orthodoxy: "Apple is the proof that UX and design are important; Amazon is the proof that it isn't."

The full episode, available on Buzzsprout, YouTube, Apple Podcasts, Amazon Music and Spotify, walks through how shopping behavior changes ripple through retail catalogs, product pages and media strategy. For ecommerce teams, the actionable takeaway is concrete: restructure data feeds for conversational queries, plan for bot-heavy analytics, and abandon the search for a single right answer in favor of efficient, parallel testing.