Master Class

Search is becoming conversational

That behaviour, however, has become so familiar that we rarely think about how unnatural it actually is. If you were asking another person for advice on buying a television, you probably wouldn’t say “best 65 inch sports TV 2026” You would explain that you mostly watch football, that your living room is bright during the day, that you sit three metres from the screen and that you don’t want to spend more than € 1.500,00.

Until recently, search wasn’t particularly good at dealing with that kind of request. Now it increasingly is.

AI-powered search and conversational interfaces allow people to express not only what they are looking for, but the context surrounding it. Questions can be longer and more specific, while follow-up questions allow users to refine their thinking without beginning the search again. Instead of translating an intention into keywords, we can increasingly describe the intention itself.

Keyword search is certainly not disappearing. But alongside it, a different search behaviour is emerging, and it could have implications well beyond the search engine.

Finding information vs. expressing interest

The significance of conversational search is easy to underestimate if we see it simply as a new interface for an old behaviour. The change goes deeper than replacing a search box with a chatbot.

Traditional keyword search leaves much of the interpretive work to the user. We formulate a query, receive a selection of results and decide which ones are relevant. If the results aren’t right, we change the keywords and try again. Conversational search shifts more of that work to the system. A user can explain what they are trying to achieve, introduce constraints and preferences, and continue refining the request as new information emerges.

Adobe, for example, describes this as a move from short, fragmented queries with largely implicit intent towards longer, naturally phrased questions in which intent is considerably more explicit and context-rich. The expected result is changing too: from a list of potentially relevant links towards a direct, synthesised response.

Consider the difference between searching for “running shoes women” and asking: “I’ve recently started running three times a week, mostly on pavement. I sometimes get sore knees and I’m looking for a comfortable shoe for runs of around five to ten kilometres. What should I look for?”

The subject is the same, but the information available to interpret the request is entirely different. This is why the shift from keywords to conversation is ultimately a shift towards intent. The system doesn’t simply have more words to process; it has a richer picture of the problem the person is trying to solve.

Discovery becomes less linear

There is another consequence. A traditional search journey can often be understood as a succession of individual queries. Conversational search makes it possible for those queries to become part of the same developing thought process.

Someone researching a television might begin by asking what matters when choosing one, then narrow the question to watching sports, introduce the size of their living room, compare two technologies and eventually ask which models meet their requirements within a particular budget.

What began as exploration gradually becomes evaluation and, potentially, a purchase decision.

Adobe refers to this development as an intent spectrum, moving between exploration, validation, decision and synthesis. Importantly, those stages are not necessarily linear. People can move backwards and forwards as they learn more, reconsider what matters or introduce new information.

For brands, that changes the role of search. Being discoverable is still important, but discovery is no longer necessarily about appearing for one carefully selected query. Increasingly, it is about being relevant to a conversation that develops as the user gets closer to understanding what they need.

And this is where the implications become particularly interesting for retail.

What happens when this behaviour enters the store?

Physical retail has always involved a form of search. Customers enter stores looking for products, certainly, but they also arrive with questions that are considerably more complicated than the categories on the shelves.

Which running shoe is right for the way I train? Which foundation works with my skin and the finish I prefer? What should I cook for four people if one guest can’t eat gluten? Which television makes sense for my living room rather than simply having the best specifications?

A good sales associate can interpret those questions because a conversation reveals the context behind them. Digital experiences have traditionally had a harder time doing so. Many in-store interfaces still reflect the logic of conventional digital navigation. Customers choose a category, select filters, browse products or enter a search term. There is nothing inherently wrong with that approach, and for many interactions it remains efficient. But conversational behaviour introduces another possibility: allowing the customer to begin with their need rather than the retailer’s information architecture.

Imagine a customer standing in front of an interactive display in a beauty store. Instead of navigating through Face → Foundation → Liquid → Hydrating, she explains that she wants a lightweight foundation for dry skin, prefers a natural finish and would like to stay below €40.

The system could interpret those preferences against the retailer’s assortment and present a relevant selection. The customer could then ask which option offers the lightest coverage, whether her shade is available in that particular store, or where she can find the product.

In an electronics store, the conversation might begin with how the customer intends to use a product. In grocery, it could begin with a meal rather than an ingredient. In fashion, it might begin with an occasion rather than a product category.

The principle is the same: instead of asking customers to understand how the assortment is organised, the experience begins by understanding what the customer is trying to accomplish. That is a meaningful change in how we might think about digital touchpoints in physical environments.

A conversational interface is only as intelligent as the store behind it

There is, however, an important distinction between creating a conversational interface and creating a useful conversational retail experience.

Adding an AI assistant to a touchscreen is relatively easy. Giving that assistant enough reliable information to provide useful answers is considerably harder.

If a customer asks for a waterproof running shoe below € 150 that is available in size 39 in the store they are currently standing in, the quality of the answer depends on much more than the language model interpreting the question.

It depends on product information, inventory and pricing. Depending on the experience, it may also depend on store data, campaign rules, loyalty information, recommendations or other business systems. The conversational layer may make the interaction feel simple, but considerable complexity has to be resolved behind it.

That makes integration increasingly important.

The opportunity is not simply to place generative AI on top of existing digital experiences. It is to connect a more natural way of expressing intent with the systems that already know what products exist, where they are available, what they cost and how they relate to one another.

For retailers that have spent years connecting their digital and physical ecosystems, conversational interfaces could therefore become another way of making those investments useful to the customer.

Designing around the question behind the search

None of this means the search box is about to disappear.

Sometimes a customer knows exactly what they want, and the most useful experience is still the shortest one. Typing a product name, scanning a QR code or tapping a category can be considerably faster than having a conversation.

The change is that those interactions no longer need to be the only model available.

As consumers become accustomed to digital services that can interpret natural language, retain context and respond to increasingly specific requests, expectations are likely to travel with them. Adobe’s research already points to the scale of the behavioural shift, with one in four customers turning to AI-powered platforms as a primary source for information and purchase decisions.

What this will mean for physical retail is still being defined. Not every store needs conversational AI, and not every customer question requires it. But the underlying behavioural change is worth paying attention to.

For a long time, we designed digital search around the information we had available: products, categories, attributes and keywords. Conversational search gives us an opportunity to approach the interaction from the opposite direction and begin with the person standing in front of the interface.

Not simply: What are they searching for?

But: What are they actually trying to do?

For retailers and experience designers, that may prove to be the much more interesting question.