At Coveo’s Relevance 360 Commerce event, retail and B2B leaders came together around a question that is hard to avoid: what does it actually take to turn AI ambition into measurable business impact?
What happens when AI-search, conversational discovery, and agentic experiences move beyond demos and into the realities of large catalogs, complex buyer journeys, operational constraints, and revenue accountability?
What connected the stories shared by Commerce leaders from Freedom, ADI, and ABB Robotics was not industry or scale. It was posture. These teams did not wait for a perfect roadmap. They leaned in, tested, optimized, and treated search and discovery as something far more important than a website feature. They treated it as a growth layer.
These organizations making real progress are building systems that help buyers find the right product faster, make better decisions, and move forward with more confidence.
Performance Problems Start Earlier Than Most Teams Think
One of the clearest themes from the event was that commerce friction does not start at checkout. It starts much earlier, often in the first few seconds of discovery.
In commerce, weak discovery does not just create a bad experience. It quietly reduces conversion, wastes merchandising effort, and makes growth harder to sustain.
This point showed up in three different ways throughout the event.
- For Freedom, the issue was shopper confidence in a high-consideration retail journey.
- For ADI, it was the operational drag of managing search at scale across a complex B2B distribution environment.
- For ABB Robotics, it was precision in a buying journey where compatibility, technical content, and service context all matter.
Different businesses. Same underlying lesson: if discovery breaks, performance breaks with it.
In Retail, Discovery Is a Confidence Strategy
Paula Mitchell, Digital General Manager at Freedom, described a buying journey where product discovery is inseparable from buyer confidence.
That is especially true in furniture. Customers are not just searching. They are researching. They are comparing finishes, validating style choices, and picturing how a product will fit into their home. As Mitchell put it, search was not just functional; it was a confidence-building tool.
That became more urgent as Freedom’s catalog expanded from 10,000 products to 45,000. The company had outgrown its existing setup. The search results were so poor as to cause internal embarrassment. Mitchell’s example captured it perfectly: search for a black sofa, get a white one.
What changed is instructive. Freedom did not just replace one search tool with another. The team reframed the problem around product discovery, speed-to-value, and scale. Mitchell described launching within three months, then moving from a rule-heavy environment to AI-led optimization across search, listing pages, recommendations, dynamic facets, and merchandising. Fourteen months later, the business had grown from roughly 45,000 products to around 75,000, without creating the same operational burden on the team.
That is the deeper lesson in Freedom’s story. Better discovery does not just improve the front-end experience. It changes the operating model behind it. Teams spend less time on repetitive manual work and more time shaping the experience in ways that actually influence performance.
Just as important, Freedom’s next move says a lot about where commerce is headed. Mitchell spoke about conversational product discovery as a way to bring more of the in-store experience online: understanding style, preferences, and intent well enough to guide shoppers naturally instead of forcing them down rigid paths. That is a much more ambitious idea than “better site search.” It is a move toward guided decision-making.
Scaling Search Means Moving Beyond Manual Rules
ADI’s story brought a different kind of clarity.
Hunter Brady, ADI’s Global Digital Product Manager, described a B2B distribution environment spanning multiple verticals, global sites, mobile apps, and very different user behaviors. Before the change, one of the company’s biggest challenges was trying to fix search issue by issue, rule by rule. Brady’s description was memorable: it felt like trying to plug a hole in a ship with a thousand different holes.
That image gets at a problem many commerce teams know well. Manual search management can feel manageable for a while. Then the catalog grows, user intent diversifies, edge cases multiply, and the rules meant to fix one issue start creating problems somewhere else. At that point, what looked like control turns into drag.
ADI’s results show what happens when that burden starts to lift. Brady shared that null search rates dropped from the double digits to below 3%. Customer satisfaction with search improved. Internally, digital merchants saved multiple hours per week through the merchandising interface and workflow improvements.
But the most useful part of ADI’s story may have been the advice. Brady urged commerce leaders not to over-engineer the system too early or come in with too many assumptions about how search should behave.
Start slower. Let the machine learning do more of the work. Then get much more rigorous about understanding the patterns in your data. Not just the top search terms, but the types of searches users are making: part numbers, categories, brands, attributes, and where performance breaks by class, not just by keyword.
That is a smart correction for any commerce team trying to scale. The goal is not to react faster to every symptom. It is to understand intent well enough to solve the underlying pattern.
In Complex B2B, Precision Becomes the Experience
ABB Robotics added another layer to the story: what discovery looks like when the buying journey is technical, multi-persona, and operationally high stakes.
Franco, ABB Robotics’ Global E-Commerce Manager, described a business serving buyers, plant managers, maintenance teams, and partners all needing different things. Some are searching for products. Others are looking for compatibility details, software, CAD files, specifications, spare parts, or service information. In that environment, precision and reliability are not nice-to-haves.
In B2B, discovery is about tightly connecting products and content so users can make accurate decisions without jumping between systems or second-guessing what they see.
Franco explained that ABB is using Coveo across both products and content, including a product base of roughly 35,000 items spanning robots, software, accessories, and spare parts. The aim is to make sure users get the exact answer to the exact question in a grounded, authenticated context. And when there is no safe answer, the system should not pretend otherwise. In ABB’s world, getting it wrong can have real downstream consequences.
The early numbers are strong. ABB reported an average click rank of less than two, meaning users are typically finding what they need in the first or second result. It also saw 26% higher engagement from customers using generative answering.
That performance matters, but the strategic direction matters even more. ABB’s next phase is a stronger, authenticated experience shaped around the user’s role and specific needs. That is where B2B commerce is heading: toward experiences that are not just searchable, but context-aware.
The Next Shift Is Not More AI. It is Better Context
Peter Curran’s argument in his session was that commerce is sitting in a liminal space between existing digital experiences and emerging AI surfaces, and that companies need to build for both without losing control of the business. His point was architectural, AI is only as dependable as the content and context beneath it.
Not just catalog data. Context. Inventory. Compatibility. Account data. Entitlements. Service records. Content. Fitment. Availability. All of it brought together so a system can answer more complex questions with more precision. That is a much bigger vision than adding a conversational layer on top of a storefront. It is about making discovery intelligent enough to support actual decisions.
Curran also made an important cautionary point. Many commerce experiences still fail on the basics. One poor query interpretation can send a buyer down the wrong path or off the site entirely. In that sense, the race toward agentic commerce does not remove the need to fix search fundamentals. It raises the stakes.
The Bigger Picture
What came through most clearly at Relevance 360 Commerce was not enthusiasm for AI in the abstract. It was a more disciplined view of what commerce leaders should actually build.
Freedom showed that better discovery creates buyer confidence and gives teams room to operate at scale. ADI showed that modern search cannot be managed effectively through endless manual rule creation, and that understanding intent at the pattern level is what makes progress sustainable. ABB showed that in complex B2B environments, precision is not just part of the experience. It is the experience.
Together, those stories point to the same conclusion.
The next phase of commerce is building experiences that can understand what buyers are trying to accomplish, connect products and content in the right context, and guide people from exploration to decision with less friction and more confidence.
Watch the full session and find an upcoming Relevance 360 event near you.