Holiday Season Wins Will Come Down to the Human Touch

Every retailer heading into Christmas has access to the same review data, the same on-site search logs, and roughly the same handful of CRM and AI tools. The gap between brands is no longer who has more data. It is who reads what they already have, and who resists the pull toward doing everything the same automated way.

That was the throughline running underneath “Preparing for Peak Shopping Seasons: How to Win Christmas” at eCommerce Expo London. The panel, moderated by Chris Hutchins of Audiense, brought together Ellie Bradford of Clearer.io, Naomi Boonin, VP Commerce at WPP Media, and James White, Senior Director of EMEA Partnerships at Klaviyo. Between them, they moved from search bars to retail media budgets to AI shopping agents, and landed on the same conclusion from different directions.

The data retailers ignore is sitting in their own search bar

Bradford opened with a direct point. Most of what retailers need for peak prep already exists on their own site. She pointed to three sources worth mining before Christmas. Top search terms show what shoppers wanted last peak season and what they might want again. Zero-result searches expose catalog and tagging gaps that quietly turn shoppers away. Review data, especially complaints about delivery, signals where proactive communication about delays would prevent bad reviews from landing in the first place. Her fix was practical.

Bradford extended that logic to gifting. A shopper buying for a coffee-loving relative will not search for a “58-millimeter portafilter.” They search “gift” and “coffee lover.” A retailer that only optimizes for expert search terms misses the customer who has money in hand and just needs the right product surfaced quickly. Get that right, she argued, and a one-time gift buyer becomes a repeat customer.

Channel count matters less than channel fit

White made a related point from the CRM side. Most merchants are running five or more different platforms, and the data sits fragmented across them. His recommendation was to point an AI assistant at that fragmented ecosystem and ask whether those data points are being used to segment customers and personalize campaigns. One concrete output of that exercise is channel preference. A customer with email, SMS, and WhatsApp consent will not respond equally well to all three, and matching the channel to the customer beats blasting every channel at once.

Boonin pushed the same instinct further, from an agency seat. As retail media networks multiply, the temptation is to add more placements rather than curate fewer, better ones. She described a campaign where Tesco and Nestle broke down separate brand and trade budgets to plan together rather than compete for the same shopper, a collaboration that went on to win multiple industry awards. For marketers, the signal is that consolidating spend around a clear outcome beats spreading it across every available network.

Automation should carry the tickets while people carry the relationship

The panel’s most concrete example of freeing up human time came from White, describing how IKEA moved roughly 8,300 support staff off routine ticket resolution and into a concierge-style design service. Instead of measuring agents on ticket volume, IKEA started measuring them on whether they could help a customer plan a room, and that shift turned single-item purchases into multi-item baskets. White traced that back to an earlier point about door-to-door retail sales, where the deals that closed came from curiosity about a customer’s motivations rather than a rehearsed pitch. Automation, he argued, should take the tactical load off that conversation so people have more room to have it well.

Bradford’s closing point connected this to how shoppers now search. As more customers use AI agents and chatbots to shop, they attach specific constraints, such as a price ceiling, a star rating, or a size requirement. Product data that only states a name and price cannot answer those constraints. Pairing structured product data with detailed review data, the kind an AI assistant can parse, is what determines whether a retailer’s products get recommended at all.

What this means for senior marketers

White’s closing point addressed why brands are starting to look alike. As more brands adopt the same AI tools trained on the same data, output starts to converge, and brands that lean too far into that convenience risk becoming interchangeable. His example was a company called Organic Protein Company, which sends a handwritten note with every order. “It’s the thing I remember most about the brand,” he said, and it costs the company almost nothing in time.

That is the peak season lesson from this panel. The data is available to nearly everyone. The tools are available to nearly everyone. What still separates one retailer from another is where they choose to put a person back into the moment that matters.