Friday 15 Podcast

Is the Search Box Now the Intent Box in B2B Ecom?

Dr. Andrea Polonioli of Coveo joins Andy Hoar and Brian Beck to unpack why B2B search still breaks on basic queries, how AI is closing the gap with buyer expectations, and what agentic commerce means for the search box.

Friday 15 Podcast · Guest: Dr. Andrea Polonioli, Product Marketing, Coveo

Key takeaways

  • Roughly 50% of B2B website users rely on site search to find products, climbing to 60-75% for retailers with very large catalogs.
  • Search-driven visits convert at roughly two and a half times the rate of browsing-only visits, according to industry consensus.
  • Semantic and vector search alone do not solve B2B search's basic precision problems, such as returning non-alcoholic wine for a non-alcoholic beer query.
  • The paradox of choice means returning more results is not automatically better. Buyers can struggle to decide when faced with too many options in large B2B catalogs.
  • More than half of all internet traffic is now agentic or bot-based, according to Cloudflare data, raising new questions about how search and retrieval work when the buyer is a bot rather than a person.

Home Depot’s Pro momentum keeps building as SRS gains share

The breaking news segment returned to a familiar theme: B2C companies moving deeper into B2B. Home Depot’s second quarter results, published just days before the episode, showed comparable sales in its Pro segment, contractors typically served by distributors, outperforming its DIY segment. The company reported strength across pro-heavy categories including portable power, decking, dimensional lumber, pipe and fittings, and fasteners. SRS Distribution, the B2B distributor Home Depot acquired roughly two and a half years ago, posted comparable sales above the company average with positive comps across all its divisions.

Andy said the pattern keeps repeating because the economics favor it. “B2B is an untapped opportunity and it’s more profitable, it’s more reliable,” he said, adding that Home Depot and Lowe’s have effectively become B2B companies without the organizational hesitation that slows down traditional distributors and manufacturers. Brian pointed out that Home Depot’s retail background gives it an edge most B2B incumbents lack: comfort with direct-to-end-buyer selling and digital channels as core to the business, rather than something layered on top of a sales-rep-driven model. Both hosts agreed the higher service expectations of B2B customers, who depend on deliveries and functioning equipment in a way B2C customers generally do not, create real operational risk alongside the profit opportunity.

Is the search box becoming the intent box?

The main topic set up a session Master B2B will run at the B2B Exchange at Shoptalk Fall in Nashville: whether the search box in B2B ecommerce has effectively become an intent box. Andy opened with the scale of the problem. Roughly 50 percent of B2B website users rely on site search to find products, a figure that climbs toward 60 to 75 percent for retailers with very large catalogs, citing Cardinal Health as an example near the high end. Because professional B2B buyers often already know exact part numbers and specifications, they frequently bypass browsing entirely and go straight to the search box. Andy called this a doubly important dynamic in B2B, since buyers behave more like hunters than farmers. Industry consensus puts conversion rates roughly two and a half times higher when a visit involves search rather than browsing alone, a difference large enough to directly affect revenue for companies that get search right.

Andrea Polonioli: search is still broken at the foundation

Joining the conversation was Andrea Polonioli, a product marketer for Coveo’s commerce line of business, whose academic background is a PhD in philosophy and cognitive science from the University of Edinburgh. Coveo provides search and product discovery technology for large manufacturers and distributors across categories including pharma, medical, and food and beverage.

Polonioli was direct about the current state of B2B search. “Search is still broken at the foundation,” he said, describing how a missing dash or dropped prefix can still break basic retrieval on many B2B sites. He pushed back on the idea that semantic and vector search alone solve the problem, citing an example from a food and beverage distributor: a shopper searching for non-alcoholic beer should not be served non-alcoholic wine or alcoholic beer, a precision failure vector search does not automatically fix. He connected this to a widening expectation gap. Buyers now use tools like ChatGPT in their personal lives and expect that experience at work, even as many B2B sites still require near-perfect query syntax to return the right result.

Humans make mistakes, they have incomplete thoughts. You can use AI to fill in these gaps and determine what they really meant to say, instead of what they actually typed.

Andrea Polonioli, Coveo

Polonioli also raised the paradox of choice from cognitive psychology: returning more results is not automatically better, since buyers can struggle to decide when faced with too many options, a problem he said is common in B2B given typically large catalogs. He argued that companies with weaker search relevance today have a leapfrog opportunity, since they no longer need to rebuild twenty years of search evolution step by step to close the gap with a partner rather than in-house development.

Will bots buying from bots make search irrelevant?

Hoar raised agentic commerce directly, citing recent Cloudflare data showing more than half of all internet traffic is now agentic or bot-based, up sharply from a few years ago. Polonioli, referencing Andy’s own earlier book on the changing nature of B2B sales, described himself as a pluralist on where this goes. He expects multiple channels, storefronts, dealer portals, and apps, to keep coexisting rather than being replaced outright, and expects bot-to-bot commerce to take hold first for routine, lower-value transactions rather than complex ones. “We are still going to have many different kinds of touch points and channels coexisting,” he said.

He confirmed generative, ChatGPT-like search experiences are already live at scale, including with Fortune 500 customers running Coveo’s generative answers, though not yet fully multi-turn. He flagged a genuine open problem in measurement: when a conversational search flow provides a direct recommendation, buyers are less likely to click through a separate product carousel, creating a cannibalization effect that changes how performance needs to be measured, a challenge many companies are still working through.

What bot traffic means for search going forward

Brian and Andy closed by connecting the search conversation back to agentic buying more broadly. Andy described a scenario where an AI agent, similar to tools like Claude Cowork, could independently identify what a procurement manager needs to buy, check prices and availability, and place items on hold, all without a human directly issuing a query in a search box. He argued this still involves a query, just issued by a bot rather than a person, meaning the underlying search and retrieval problem does not disappear, it simply operates differently. Both hosts agreed that data quality underlying these systems, a topic they plan to cover in more depth at the B2B Exchange and next year’s Mindshare Summit, will determine how well this actually works in practice.

Frequently asked questions

How important is site search in B2B ecommerce?

Very important. Roughly 50% of B2B website users rely on site search to find products, and that share can climb to 60% to 75% for retailers with very large catalogs. B2B buyers often already know exact part numbers and specifications, so they frequently bypass browsing and go straight to search.

Does using search increase conversion rates in B2B ecommerce?

Yes. Industry consensus puts conversion rates roughly two and a half times higher when a website visit involves search rather than browsing alone, making search quality a direct driver of ecommerce revenue.

Why does B2B search still fail on basic queries even with AI?

Many B2B search experiences still struggle with basic retrieval, such as a missing dash or dropped prefix in a query. Semantic and vector search improve relevance but do not automatically fix precision problems, like returning the wrong product type for a specific query, which require additional work to close.

What is the paradox of choice in ecommerce search?

The paradox of choice is a cognitive science finding that returning more results is not automatically better for the buyer. When too many options are available, people can struggle to make a decision, which is a common risk in B2B given the large catalogs many manufacturers and distributors carry.

Will agentic AI commerce make B2B search obsolete?

Not according to Coveo's Dr. Andrea Polonioli, who expects multiple buying channels and interaction types to coexist rather than one replacing the others. He expects bot-to-bot commerce to take hold first for routine, lower-value transactions, with more complex purchases continuing to involve human buyers and traditional search for longer.

Sources & methodology

  1. The Home Depot, Second Quarter Fiscal 2026 earnings results, August 18, 2026
  2. Cloudflare, internet traffic composition data on agentic/bot-based traffic
  3. Coveo, B2B search and product discovery research
  4. Friday 15 Podcast, Master B2B
Andy Hoar Andy Hoar
Co-Founder, Master B2B

Andy is a Co-Founder of Master B2B, founder of Paradigm B2B and author of the book Bot2Bot: The New Future of B2B Commerce. Andy is one of the leading global authorities on B2B commerce strategy.

Brian Beck Brian Beck
Co-Founder, Master B2B

Brian is a co-founder of Master B2B, Managing Partner of Amazon agency Enceiba, and author of the book "Billion Dollar B2B Ecommerce." Brian has also been C-level digital commerce executive with two decades of experience.