Elastic Path is the API-first, AI-ready commerce platform for brands with complex requirements: large catalogs, multi-tier pricing, and multiple brands and markets. Manufacturers, distributors, and retailers run account hierarchies, quoting, and negotiated pricing natively, not through bolt-on apps and workarounds.

Jonathan Taylor became CEO of Zoovu in January, and he came into our Manhattan studio for this week’s episode, which was made in partnership with Zoovu. The company sells search, discovery and conversational tools to retailers and manufacturers. Taylor spent a surprising amount of the conversation on the ways chatbots fail.

Taylor says Zoovu’s data shows shoppers who choose a conversational experience convert at a higher rate. Most shoppers don’t want one, though, and he says they leave when a site forces them into it. So a retailer that puts a chat window in front of every visitor can see chatbot conversion go up while losing more sales from people who wanted the search bar and category pages.

Taylor blames some of the inflated expectations on Google. A lot of traffic now arrives from AI summaries, where shoppers have no choice about the format, and brands assume those shoppers want a conversation once they land. On the retailer’s own site, the shopper gets to decide.

He didn’t give me a number for that conversion lift, and I’d want one before taking it to a board. His fix is sound anyway. Split incoming traffic into cohorts and measure total conversion across the site. Chat counts and engagement rates on their own remind me of the impression reports we all sent around in the early 2000s.

According to Taylor, a team can get a conversational experience about 70% of the way there in an afternoon now, and that version demos well to an executive. The remaining work is protection against prompt injection and keeping customers from typing personal data into the chat, which he says creates compliance exposure in Europe and in the US. He also found that shoppers in the US want different answer lengths and tone from shoppers in Europe or APAC.

He puts the real cost of a self-built bot in years two and three. Retailers, in his view, are rushing the build when the slow part is learning how customers use it. Zoovu sells the alternative to building, so he has a stake in that argument, and I think he’s right about where the time goes regardless.

Taylor says half of Zoovu’s business is B2B, with customers selling products like fire suppression valves and breathing apparatus for firefighters. One of those customers gave a conversational tool to its own sales and support reps first. Those reps used to search PDF manuals for answers and call the customer back once they’d found one. After the tool proved accurate for staff, the company opened it to buyers.

Staff make tough testers because many of them have spent decades in the industry and will spot a wrong answer fast. Taylor’s argument is that a retailer should find the flaws there before customers do, and B2B buyers, who tend to know precisely what they’re asking, would find them quickly.

Early in the conversation Taylor said he doesn’t think most people want an agent doing their shopping. He thinks they want discovery to be easier and still want to make the choice. Plenty of retailers are building agentic strategies right now, and I don’t know how many of them have asked their customers which of those they want.