Commerce Isn’t Bad at Conversion; it’s Bad at Understanding Intent

Retailers have spent the last two decades focused on conversion: driving more traffic, reducing checkout friction, tuning site search, and personalizing the experience. Those things still matter, they just don't answer the question that now decides whether many shoppers buy: What is the customer actually trying to accomplish?

That's an intent problem.

For years, e-commerce search was built around keyword matching. Shoppers typed in a few words, the system returned products, and consumers sorted through the results themselves. That worked reasonably well when shoppers knew the product they wanted. It breaks down when they need help deciding.

Shoppers Are Asking Questions, Not Typing Keywords

Today’s shoppers often arrive with a problem to solve rather than a specific product in mind. They’re thinking about goals, budgets, preferences, or situations.

Someone might search for “the best mattress for a hot sleeper on a budget” or ask which hiking boot works on trails and around town.

These aren’t simple product searches. They’re decision-making questions.

At the same time, shoppers are getting used to more conversational experiences everywhere else online. A growing number now start their journeys on artificial intelligence platforms like ChatGPT or Google AI Overviews before they ever reach a retailer’s site.

That shift is happening quickly. In our latest Commerce Relevance Report, 67 percent of consumers said they expect online shopping experiences to evolve with conversational guidance, while 59 percent said they’re more likely to buy when guided to exactly what they’re looking for.

That changes the role of search. It still has to retrieve the right products, but it also has to know when the shopper needs answers, comparisons, or guidance.

Friction is Often a Symptom

Retailers often treat friction as the reason shoppers fail to convert. Usually, friction is the symptom. The deeper problem is missed intent.

Take a search like “waterproof watch.” One shopper might want a smartwatch for swimming, while another may need something durable for hiking or diving. Most e-commerce sites still handle all of those shoppers the same way: they return a large product grid and expect people to figure out the differences themselves.

Shoppers leave quickly when that happens. Our research found that 71 percent of shoppers abandon a site when search doesn’t deliver relevant results. The most common complaints aren’t technical problems. They’re relevance problems: too many choices, poor filtering, irrelevant recommendations, and difficulty narrowing options.

That is an intent problem showing up as a user experience problem..

Why Bolt-On AI Won’t Fix it

AI is making this problem more obvious.

Many retailers are rushing to add AI assistants or chatbots to their storefronts, but adding a chatbot to a weak discovery experience doesn’t automatically make it more helpful.

If the underlying search, catalog, content, and personalization systems don’t understand shopper intent, the AI layer won’t either. It may sound conversational, but the shopper is still doing the hard part of narrowing choices and deciding what's right.

Shoppers want more guidance during the buying process. They want help comparing products, narrowing choices, and understanding what’s actually right for their needs.

However, they’re not asking AI to take over the buying decision. They want AI to narrow the field while they remain in control of the experience.

Understanding Intent is the Next Conversion Lever

Retailers need discovery experiences that better understand what shoppers are trying to accomplish and respond accordingly, with the right products, comparisons, answers, or content to help them narrow choices more confidently.

That’s ultimately what shoppers are asking for: less effort, more clarity, and more confidence when they’re trying to decide what to buy.

Peter Curran is the chief product officer at Coveo, the leading platform for GenAI, AI agents, commerce, customer service, workplace and website experiences.