At Candid Commerce London, hosted by Kyn and The FMCG Guys, we heard how the digital shelf is entering a new era, as AI-powered discovery and conversational commerce reshape how shoppers find, engage and buy from brands.

John Lewis’ Tom Langley, MARS’ Swagat Choudhury and CommerceIQ’s Vanessa Dodd joined The Magnum Ice Cream Company’s Melda Hamarat to explore how the rise of AI assistants, Large Language Models (LLMs) and agentic commerce is forcing retailers to rethink discoverability.

Conversational AI is driving dialogue-led discovery

With consumers turning to LLMs to start their product searches or to seek inspiration for purchases, “AI discovery is quickly becoming the new entry point for shopper journeys,” said Melda Haramat, Magnum’s Global Head of Digital Commerce Strategy.

One of the clearest shifts for brands is consumers moving away from traditional keyword searches towards conversational, intent-led prompts, as they transition from search engine to answer engine discovery behaviours.

“In a keyword world, people used to search three words,” explained Swagat Choudhury, Global Director of Digital Commerce at MARS. “Now, three has become between 15 to 30” as shoppers increasingly layer in context, intent and occasion within searches.

Conversational AI is turning product discovery into an ongoing dialogue rather than a single search moment, Choudhury suggests. And this is fundamentally changing how brands need to think about discoverability, content optimisation and product information management (PIM).

The digital shelf beyond the PDP

A recent survey by Akeneo showed that now over two thirds (67%) of consumers rate AI assistants for accurate and detailed product information, highlighting the rapid adoption of AI-first search.

With this change, the panel agreed that discoverability spans a much broader spectrum than retailers’ product description pages (PDP) alone, as AI platforms and agents simultaneously pull information from multiple sources – from Reddit to reviews and editorial content to social channels.

Retailers, therefore, need to think more holistically about where – and how – they appear online as their brand truths are being increasingly defined by sources outside of their controlled and curated owned-channels and platforms.

“This isn’t just an evolution of a playbook. It’s a fundamentally different way consumers are interacting with products to achieve a shopping mission.”

Vanessa Dodd, Director of Solutions Consulting, CommerceIQ

The rise of AI discovery is also elevating the role of earned visibility and authority signals, meaning digital commerce no longer operates in isolation from brand, PR and media teams, added Choudhury.  

“Citation is a big conversation in the world of LLMs,” he noted, highlighting how backlinks, mentions and trusted third-party references are becoming increasingly influential in how AI models interpret and recommend brands.

The “brilliant basics” still matter

AI may be changing how products are surfaced, but the fundaments – or “brilliant basics” – are still critical, the panel concurred. Poor product data, out-of-stocks, lack of delivery options and questions about brand trust can still undermine conversion decisions in AI shopping journeys.

Recent data from delivery platform, Ingrid, for example, showed that accurate and flexible delivery choices in AI buying journeys still ‘make or break’ conversions for 61% of shoppers, while half (49%) of shoppers polled by invent.ai believe retailers should use automation to better avoid stock-outs to improve their spending propensity.

“It’s very easy and tempting to see what’s new,” said Choudhury. “But be very clear what’s not changed; your pricing, supply, availability – all those things have not changed at all.” 

Dodd agreed adding “an AI isn’t going to recommend a product that’s out of stock, so you still need to get those availability metrics and pricing in check.”

“There’s no way to hide anymore with pricing because instead of physically walking around stores or going onto multiple blue links, LLMs are providing that service for consumers free of charge. You need to have a very robust and holistic pricing strategy.”

Brand trust still remains a defining factor, especially in big-ticket or high consideration purchases, as Tom Langley, Head of Personalisation & Retail Media at John Lewis & Partners, explained.

“People might not quite be ready to trust an AI recommendation fully through the purchase of a £1,000 television, but they will trust John Lewis. Our job as a retailer is to build a really trusted, curated range, so customers know the products being recommended are genuinely right for them.”

“The key two things to invest in at this stage are your data – so make sure your shop-level data is defined. And the second thing is invest in your brand. Because never has it been more important that your brand is one of the first things the consumer thinks of.”

“The one thing you can control is your data,” said Langley, adding that structured product data and well-organised content will ultimately determine how effectively brands and retailers adapt to emerging AI shopping environments.

Dodd agreed, advocating a single view of data would help retailers pivot to new AI-led demands.

“Having one pane of glass to see all of that data – and to be able to then spend less time analysing and more time executing – is really pivotal for brands. You can’t do that if you don’t have a good digital shelf tool in place.”

Vanessa Dodd, CommerceIQ

AI will reshape pricing, relevance & retail media

The panel also explored how AI is accelerating consumer transparency, particularly around pricing and promotions.

“AI search is making consumers smarter,” said Choudhury, noting how AI assistants are simplifying comparison shopping while exposing pricing and promotion history more openly. As a result, retailers face growing pressure to rethink promotional strategies, pricing consistency and long-term value positioning.

At the same time, the emergence of sponsored prompts and AI advertising models is expected to reshape retail media economics.

“I have no doubt that ChatGPT will become a major advertising platform,” predicted Langley. “Retail media will have to become much more of a blend between organic and sponsored, rather than simply whoever bids the highest winning visibility.”

“We should stop thinking sponsored products and retail media version 1.0. We have to change the technology and the algorithms of retail media to be more of a blend of organic and sponsored, becoming better at being relevant as well as auction-based.”

Tom Langley, John Lewis

Dodd built upon this, suggesting that organic, conversational AI will create new opportunities for brands to target broader shopper missions and occasions, rather than relying solely on transactional keyword targeting.

The next generation of retail media will also need to balance paid visibility with relevance more carefully than traditional sponsored listings, creating what Choudhury described as “personalisation on steroids.”

As AI-powered discovery continues to evolve, the panel agreed the digital shelf is rapidly becoming more conversational, contextual and fragmented, forcing brands and retailers to rethink not just where products appear, but how they are surfaced, validated and recommended across decentralised shopping journeys.

John Lewis, MARS, CommerceIQ and The Magnum Ice Cream Company were speaking at Candid Commerce London, which took place on 30 Apr 2026.

To find out more about future Candid Commerce events and the Kyn Collective, a curated community for senior retailers & brand leaders, visit: Kyn Agency.