A New Ranking, and Primark Is Losing It
Marketing agency Marketing Signals analyzed how 59 UK fashion and apparel retailers show up across ChatGPT, Google AI Overviews, and Gemini when shoppers ask for buying recommendations, then scored each retailer's AI visibility. Primark, one of the UK's largest clothing retailers by footfall and one of its highest-traffic websites at 6.65 million monthly visits, ranked 50th out of 59 with a visibility score of just 3.39 out of 100. That is a striking gap between physical and digital scale on one hand and AI discoverability on the other, and it lands just as Primark has been investing heavily in its wider digital operation.
The retailers that did well were not necessarily the ones with the most traffic. Harrods topped the index at 36.3, and Whistles ranked second at 24.8 despite drawing only 217,000 monthly visits, a fraction of Primark's audience. Next finished last at 1.6, and M&S ranked 54th at 5.0, putting several household names near the bottom alongside Primark. Marketing Signals managing director Gareth Hoyle summarized the mechanism plainly: "AI tools don't reward the retailers with the biggest stores or highest website traffic. They reward the ones that get written about and cited across the web."
Editorial Presence, Not Footfall, Is the New Currency
The gap is especially notable given Primark's operational investment elsewhere. The retailer has put 90 million pounds into an automated fulfillment center in Sheffield to support home delivery, a substantial bet on physical and logistical capability. Yet none of that investment moves the needle on whether an AI shopping assistant recommends Primark when a customer asks where to buy a specific item, because AI models draw their answers from press coverage, review sites, and third-party editorial content, sources that have little to do with a retailer's own operational scale or transaction volume.
This creates a genuinely new marketing discipline, distinct from both traditional search engine optimization and social media strategy, focused specifically on earning citations and mentions across the sources large language models actually draw from when generating shopping recommendations. Retailers that have historically underinvested in press relations, product review partnerships, and third-party editorial coverage in favor of paid search and social advertising are now discovering that underinvestment shows up as near-invisibility in a fast-growing discovery channel, one that younger shoppers already report trusting for product recommendations.
Someone Is Already Building the Measurement Layer
This gap is exactly the opportunity agentic commerce optimisation startup Azoma is chasing. The company has landed new investment from dunnhumby ventures, the venture arm of the customer data specialist whose clients include L'Oreal, Unilever, and Mars, to accelerate its roadmap for a platform that measures and improves how AI shopping agents discover, interpret, and recommend products for retailers and CPG brands. Azoma CEO Max Sinclair said the backing matters because "dunnhumby understands this, and brands and retailers' current data blindness better than anyone," framing the core problem as most companies having little visibility into why an AI does or does not recommend them.
Azoma's own Q2 2026 research adds useful texture to the Primark finding: earned and social media accounts for 86.5% of citations behind Alexa for Shopping recommendations, 76% of Walmart Sparky's recommendations trace back to owned retailer accounts, and ChatGPT draws 37.1% of its product citations from retailer-owned sources. Each AI platform weights source types differently, which means an optimisation strategy tuned for one assistant will not automatically transfer to another. Dunnhumby ventures head Leo Nagdas, who is joining Azoma's board, called this backing "how we bring emerging technology to our clients as consumer behaviours evolve."
The Generational Pressure Behind the Numbers
Separate research from Mastercard, covering 26,000 parents and teens, shows why this visibility gap will only compound over time. Among teenagers surveyed, 27% said they would likely use a fully AI-run shopping assistant, compared to just 16% of parents, and 62% of teens already use AI at least monthly for product research versus 49% of parents. Mastercard executive vice president Brice van de Walle put the trajectory plainly: "AI increasingly helps people decide what to buy; tomorrow it will help do the purchasing for them too."
That generational gap means the retailers with weak AI visibility today, Primark and M&S among them in the Marketing Signals index, are least discoverable to precisely the shopper cohort growing most comfortable delegating purchase decisions to AI. A retailer can treat a 3.39 visibility score as a rounding error today because most transactions still start elsewhere, but the Mastercard data suggests that assumption has a shrinking shelf life as today's teenagers become tomorrow's primary household spenders with AI-first shopping habits already formed.
Implication for the Roadmap
Retail and CPG marketing leaders should treat AI citation share as a metric worth tracking independently from search rankings and web traffic, since Primark's case proves the three can diverge sharply. That means auditing where and how a brand or retailer is currently discussed across review sites, press coverage, and third-party content, the raw material AI models pull from, alongside optimizing owned website content and paid placements. The Primark and Harrods contrast should be a wake-up call for any retail marketing team still measuring success purely through traditional web analytics.
The practical next step is deciding whether to build this measurement capability internally or buy it from an emerging vendor category that Azoma and its peers are racing to define. Given how early this space is and how differently each AI platform sources its recommendations, buying a specialized tool with existing research across platforms is likely faster and more accurate than building internal tracking from scratch this year. Budget owners should expect this line item to sit alongside SEO and social spend within twelve months as a standard part of the marketing stack.
