North Sole turns a three-SKU catalog into a one-question sale

North Sole sells against its own category. Every mainstream competitor sells arch support and heel lift; North Sole's product deliberately has neither, and its bundle page says so outright, most insoles "'fix' this by adding arch support and heel lift, which forces your foot out of its natural position and defeats the whole point of going barefoot."

That makes the purchase explanation-dependent in a way a product grid cannot solve. The chat data shows it precisely: the two dominant themes across the window were size certainty and fit inside shoes, and choosing between Legacy, LegacyThree and the bundle. Nearly every chat is one of two questions.

There is a third pattern the brand did not expect. Shoppers arrive asking about plantar fasciitis and foot pain (the exact framing North Sole positions away from) and about wide toe-box and zero-drop compatibility with specific shoes they already own. The agent has to answer honestly inside someone else's mental model without either overclaiming a medical benefit or losing the sale.

The site made this harder than necessary. The sizing chart is published as a single image, so its content is unreadable to search engines, to answer engines and to screen readers, and unavailable to a shopper right at the moment they need it.

The Website Agent went live on 24 July 2026 as North Sole's only channel. No WhatsApp, no Instagram, no email flows. One surface, one job.

AI Sales Chat was trained on the full catalog, the sizing data that had been locked inside a PNG, the trim-to-fit instructions, the transition-period guidance from the FAQ, and (critically) the brand's argument against arch support. It reads which page the shopper is on: someone on the LegacyThree page asking about standing on concrete all day gets pointed to the 6mm Legacy, not a list of both.

Smart Nudges cover the shoppers who would have left without asking. On a catalog this small, the failure mode is not a wrong answer; it is a shopper who never asks which thickness they need and closes the tab.

Conversion is the whole story here. North Sole did not add traffic; it converted more of what it already had.

Let the agent make the thickness call. Product Info was the single largest intent at 281 chats, followed by Product Browsing at 147 and Product Discovery at 69, 88% of all classified intent was pre-sale, not support. The Legacy + LegacyThree bundle was the most-discussed product on all 46 days of the window, ahead of either single item. The agent is not deflecting tickets here; it is doing the selling.

Answer inside the shopper's mental model. Shoppers ask about plantar fasciitis and about whether the insole will fit a specific pair of Merrell zero-drop shoes. The agent answers the question that was actually asked, explains what the product does and does not do, and lets the shopper decide. That is the difference between a 32.52% conversion rate and an argument.

Free the sizing data from the image. The most repeated shopper question in the window was a version of "Will this fit my shoes?" The agent now answers it, but the underlying chart still lives only as a picture on the product page. Publishing the same table as real HTML would earn organic and answer-engine traffic for a query cluster the brand currently forfeits entirely.