Six Reads from Shoptalk Europe 2026

Three days at Fira Gran Via last week made one shift impossible to ignore: the moment of discovery has moved off the retailer’s own shelf and into AI assistants, social feeds and communities no brand controls. That changes the marketer’s job from buying attention to being findable, credible and worth recommending, and from reporting spend to proving it. 

Here are the six threads worth taking back to your desk.

1. The funnel collapsed onto one screen.

The cleanest articulation came from Bianca Sing, GM and Global Head of E-commerce Strategic Partnerships at TikTok Shop. “Gone are the days that you want the linear and predictable way of shopping,” she said. “You want interest, conversion and consideration all happening close to one another.” Her proof point was P.Louise, the Manchester cosmetics brand that turned a single live-shopping session into more than £2 million in roughly four hours, a record she said held not just in the UK but globally.

You heard the same collapse described from the prestige end of the market. Nadine Graf, President UK&I at The Estée Lauder Companies, said the fundamental change is that the consumer now “discovers, experiences and buys, sometimes in the very same moment.” When discovery, consideration and purchase happen in one scroll, planning in neat upper-, mid- and lower-funnel stages stops describing reality.

For marketers, the signal is to stop treating the funnel as a sequence and start treating it as a single compressed event, with content that can carry attention and conversion in the same breath.

2. Search has left the search box.

In Google’s session on agentic commerce, the company’s definition was deliberately plain: agentic commerce is “AI computing a shopping-related task on behalf of a user.” The behavioural detail mattered more than the definition. Queries in AI mode now run around three times longer than traditional search queries, because people are talking to the tool the way they would talk to a knowledgeable shop assistant, not feeding it keywords. The same session named the gating factor for the whole shift in one word: trust. Without it, the executive said, “there won’t be adoption.”

Asked whether retail is genuinely entering a generative-search era, Amish Mehta, Director of AI and Agentic Commerce at Boots answered, “I think we’re already there.” His reasoning was a description of friction every marketer knows: “How many people today go on Google, click on a website, then click on the sidebar filters and search the internal site? People do not want this cognitive load.” His prescription was unglamorous and correct. With 50,000-plus products, Boots is treating structured data, product, inventory, pricing and context, as the thing large language models actually need in order to surface a brand. He called focusing on data “the path of no regrets.”

That view was echoed on the retail media panel, where P&G’s Rob argued that keyword-led search on retailer sites “is gone,” with discovery shifting to social and to LLMs. The practical takeaway is uncomfortable for teams that have spent a decade on on-site merchandising: the new shelf is your data and your authority signals, and they are now demand-generation infrastructure, not back-office hygiene.

3. AI hands you information. Humans still hand you trust.

If AI is where shoppers now ask, communities are where they check the answer. David Trencher, MD for UK, EMEA and Australia at Reddit, made a quotable case of the week. “A lot of individuals are taking what they’ve got from AI and gone to Reddit to validate that information,” he said, “because what AI doesn’t have is that human experience.” Across 100,000-plus communities, his argument was that the brand is no longer the authority on itself: “The community itself are the influencers.”

His advice for brands wanting in was a party analogy that landed because it is true. Do not be the person who walks in, pops a bottle of champagne and shouts. “Don’t force your way into the conversation.” He pointed to Dove turning unfiltered review copy into out-of-home creative, and to Philadelphia hijacking a long-running chive-chopping thread in real time, as examples of brands using community as a source rather than a billboard.

The practical move is to treat the communities that validate AI answers as research and reputation infrastructure, not as another media buy, and to earn standing before asking for anything.

4. The most credible creator is your shop floor.

The creator conversation has matured past reach. At TikTok Shop, Sing described MAC turning its in-store beauty advisors into livestreamers, on the logic that “everyone has creativity in them, and everyone can be a creator.” The people who sell the product all day are more convincing than a borrowed celebrity.

Susie Donaldson, VP of Marketing EMEA at SharkNinja, took it further into product design. “Virality isn’t by accident,” she said. “It’s absolutely by design.” SharkNinja brings creators into development before launch, choosing people “who understand the space, already have real use cases and are great storytellers, not just influencers with reach,” so that by the time anyone posts, they have actually lived with the product. Her example was reactive rather than planned: when Aryna Sabalenka was filmed using SharkNinja’s Chill Pill fan at Roland Garros, the team turned the organic moment into earned and paid amplification within 72 hours, lifting product mentions 44% and selling out in the UK.

The takeaway is to recruit creators with genuine use cases ahead of launch, and to build the operational speed to convert an unplanned moment into momentum before it cools.

5. Speed and culture became the operating system.

The boardroom sessions reframed all of the above as an organisational problem rather than a marketing one. Bastian Sievers, CEO of Flaconi, was the standout. He inherited a loss-making German beauty retailer and, on day one, found the real issue was not the technology. “I opened the door and I found four people, including me. So it was a company without a team.” He tore up his own 100-day onboarding plan inside a day and a half and rebuilt around culture and pace. “I don’t see the trade-off,” he said of moving fast versus protecting current trading. “It’s not about having the right strategy. It’s really a question of how good and how excellent you are on execution, and also in terms of speed.” The scoreboard he cited: a Net Promoter Score of 84, with 95% of revenue now on mobile.

Graf put the same idea in five words that several other speakers could have used: “Relevance has a short shelf life.” Estée Lauder’s answer is to compress innovation timelines with AI “from six months to 72 hours,” so the company can show up while a trend is still live. Notably, 72 hours surfaced twice across the floor, from a prestige conglomerate and a kitchen-appliance challenger, as the new unit of competitive speed.

For marketers, the implication is that agility is an org-design question, not a tooling one. Decision rights, team structure and the willingness to stop doing things matter more than the next platform.

6. The measurement reckoning is here, and the maths is harder.

If discovery is fragmenting, attribution is the bill that comes due. The “Brands Unfiltered” retail media panel was the most candid session of the event. Coca-Cola’s Anna described retail media as a “trifecta” of on-site, in-store and off-site, fed by first-party purchase signals rather than interest-based proxies. RMS Beauty’s Alejandra was honest about the cost of getting measurement right, describing a two-year build of a triangulation method across attribution, incrementality and the “halo” effect of spend across Amazon, TikTok Shop and physical stores. “It humbles you,” she said. “What you think is working is not.” P&G’s Rob pushed for total ROI over isolated ROAS: “Break the silos, make one business plan.”

The hard numbers on retail media’s growth came from David Trencher at Reddit, citing TransUnion research: retail media up eightfold from 2022 to 2025, with return on ad spend for retail advertisers up sevenfold over the same period.

The takeaway: retire attribution theatre in favour of incrementality, halo and total ROI, and prove it with at least one closed-loop test per priority retailer rather than a deck.

The counter-current: the experiences a model can’t replicate

For all the AI on the agenda, the most emotional sessions were about physical retail, and they were not nostalgic. Mark Blundell, Chief Retail Officer at Harrods, set a single bar for every activation: “It has to be one of one. No one can do it that way anywhere else in the world.” That principle produced the Burberry takeover that dressed the store’s century-old doormen and drew 15.9 million social mentions, with 18% of buyers new to the brand, and the Jellycat Airlines pop-up that pulled an unexpected luxury crossover. His warning to brands was about people, not product. Do not default to agency staff for a brand moment; “show up with your best of the best people.”

Sarah Clark, GM EMEA and Global Franchise at Lululemon, made the most disciplined version of the argument. Before a new store exists, the first hires are community leads and strategic sales partners, on the ground six months early to build local relationships. The brand met 55,000 people in real life across EMEA in the last year and posts a NPS of 80 at its Regent Street flagship. “In a digital world,” she said, “we’re still standing for meeting physically with our guests.”

That two NPS scores in the 80s, from Flaconi and Lululemon, anchored two of the floor’s most confident sessions is not a coincidence. The brands betting on experience and culture had the loyalty numbers to back it.