How customers discover products and information is constantly changing.
Search engines, social media and marketplaces like Amazon and Ebay, previously shaped how consumers discovered brands. But now AI-driven Large Language Models (LLMs) form another layer of that ecosystem, with more than 50% of online adults having used generative AI to find answers to questions
Thanks to LLMs, people can go beyond traditional search to ask questions and compare options through natural conversation, often before they ever visit a website.
What do businesses need to know to adapt in the face of that influence?
Discovery is becoming conversational
For what felt like forever, keywords shaped digital discovery. Customers would type a short query into a search engine and receive a list of links. But LLMs have changed that dynamic, enabling people to solve their needs in a more detailed and natural way.
Instead of a simple search for “best running shoes” consumers may now ask what products they need for marathon training, whether certain shoes suit a specific foot type, how they compare with alternatives, and where to buy them.
This changes the role of the search interface. Rather than sifting through a page of results themselves, users are increasingly comfortable letting AI narrow the field for them before they decide where to go next.
Consumers are already using AI tools for practical research and purchase-related decisions. We found informational (39%) and transactional (37%) prompts account for more than three-quarters of LLM usage.
And AI is not just acting as a neutral directory of options. 70% of LLM responses position a single brand as the primary recommendation.
Increasingly, people are being handed a single, curated recommendation rather than a page of results to compare themselves.
Intent starts to form before consumers reach the open web
LLMs are not just impacting discovery, but changing where research and purchase intent develop.
Historically, organizations have relied heavily solely on search queries, clicks and browsing behavior to understand user searches. But conversational AI introduces another signal, in the form of the questions people ask even before they know what to search for.
Consider a shopper comparing skincare ingredients, a traveler planning a family itinerary, or a business buyer evaluating software vendors. All reveal more context through their conversations than through simple keyword queries.
According to Gartner, consumers are using AI to research and compare products, but only 11% of consumers said they would be willing to let AI make purchase decisions on their behalf.
So while the final transaction may still happen on a retailer’s website, app or a physical store, the research and decision-making process is beginning much earlier, inside an AI conversation.
Meanwhile, AI-influenced journeys often reach their highest conversion point after five to six prompts, with 75–85% of those journeys converting within two weeks. Consumers rely on AI to narrow choices, test assumptions and build confidence, before moving to the open web.
That has implications for how organizations understand demand, but also how they understand and speak to the consumers driving it.
First, it may mean less traffic arriving directly on their website – but the users who do land there after an AI conversation are likely to arrive with a clearer understanding of the product, and a higher likelihood to purchase.
Second, it changes the nature of the signals organizations can learn from.
Traditional search data tends to be brief and transactional, while AI conversations are longer, more exploratory, more emotive, and reveal more about what people are trying to understand before they buy. Looking at both the questions consumers ask and the responses they receive offers a richer picture of how purchase decisions take shape.
Different models mean different pictures of the consumer
It’s not just the type of data that’s different, but its consistency.
Different LLMs draw on different sources, weighting information and signals in different ways. A company that appears frequently in one may barely register in another, in fact brand recommendations can vary by as much as 27 percentage points across ChatGPT, Gemini, Claude and Perplexity.
But a lot of what’s marketed as AI consumer insight isn’t based on real consumer behavior.
Often AI insight is built by taking top search terms, feeding them into a model, and treating the output as a proxy for what consumers think or want. This AEO or GEO-style focus understands the model but not the consumer.
A more useful approach starts with real consumer behavior; looking at what people actually ask, how they phrase it, and what they’re trying to work out. All observed directly and not inferred from a model’s output.
Defining success in an AI landscape
As AI becomes another discovery layer, organizations will need to broaden how they measure digital performance. Search rankings, website traffic and conversion rates will remain important. But they won’t tell the full story.
The way consumers research and decide has become more layered, playing out across search, social, marketplaces and now AI conversations, often before a brand ever sees a website visit.
Understanding that shift means going beyond how a model behaves to understand how people actually think, ask and choose. For businesses to succeed in this new era, the key is not simply having the biggest dataset, but having the one that offers the most complete picture of consumer intent, and understanding what that reveals about the opportunity ahead.
We’ve featured the best AI chatbot for business.
This article was produced as part ofTechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.
The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here:https://www.techradar.com/pro/perspectives-how-to-submit
