Everything that happened at the CX Retail UK Exchange 2026

The 15th edition of the CX Retail UK Exchange answered some of the biggest questions facing CX right now, from how to safeguard loyalty when bots are doing the shopping, to whether or not AI really is going to take CX jobs. It also tackled new questions, like what do agentic commerce and accessibility have in common? 

Reflecting on the key themes from day one, this article explores where retailers are at when it comes to loyalty, agentic commerce, customer intelligence, shopping assistants, and accessibility. 

Agentic commerce is the hottest new channel in retail

AI-first customer journeys were a hot topic in the Think Tank sessions. Held behind closed doors, the first Think Tank session on day one, Redefining loyalty in an agentic commerce era, was hosted by Paul Sims, retail advisor at Equal Experts. It tackled one of the biggest questions facing retailers right now: what is loyalty for AI agents? 

Those present discussed how their business is responding to the rise in consumer AI use. Many don't see weekly grocery shops or high-end purchases being delegated to LLMs any time soon. Others highlighted that some current agentic commerce capabilities are nothing more than an upgraded autoreplenish feature. One delegate highlighted how the use of AI agents in travel would bring the entire online travel industry full circle. 

Still, the arrival of agentic commerce poses challenges and threats to retailers and brands in all industries. 

In response, some are looking at the micro-level details, such as how copy on packaging aligns with the wider brand voice. Others are focused on the macro, such as enhancing and maximizing the after sales journey. There was a broad understanding that product and experience quality are now essential to generating the positive customer reviews that LLMs scrape for their recommendations. 

The second Think Tank session, Harnessing agentic commerce to stay visible in an AI-driven shopping future, took the conversation a step further. Brands and retailers alike understand the power of customer reviews on platforms like Trustpilot, but Reddit is also playing an outsized role in discovery. 

Because Reddit is one of the largest – and therefore most valuable – archives of human conversation, AI companies use it to train their LLMs. When prompted by a human user, for example on which laptop to buy, LLMs also use Reddit as a source of information and recommendation. In response, brands are rushing to ensure they are part of organic conversations. While this presents a huge opportunity to play the system, it also raises the question of what will happen to the internet's last genuine community platform if thousands of brands include it in their discoverability strategy.

AEO and GEO were discussed at length. Some reported difficulty in measuring the benefits of their AEO programs, while others asked whether AEO and GEO are another "dark art". Much like SEO, both require multiple optimizations to be executed in tandem, meaning it's difficult to track which action is driving results and which is not. 

However, these conversations also exposed a gap in agentic commerce execution. While the basics of discoverability – strong content, good reviews, and cohesive information – are well understood and actively harnessed, the stronger performers are already taking this several steps further, looking at YouTube, influencer marketing and packaging claims. 

Superdrug: Customer intelligence without AI

The role of AI in physical retail may appear limited, but when it comes to customer intelligence, more and more organizations are using AI behind the scenes to analyze data and make decisions. Health and beauty retailer Superdrug is not one of them. 

On the main stage on day one, the chain's customer executive direction Matt Walburn explained how the retailer uses AI to analyze customer data – and why AI will never replace human decision making.

To obtain a single customer view, Superdrug unifies behavior signals pulled from direct feedback, service signals, and social listening among other sources. It does not, however, put AI in the middle to interpret the data. 

Walburn told delegates that AI is used for CRM automations, but not as a central system that guides strategy or makes decisions. "That's a human's job," he said. Instead, Superdrug's customer intelligence team is comprised of "talented people looking at data and making decisions". 

This is just one way Superdrug blends the old-fashioned charm retail was originally built on with the tech available today – without compromising brand values. For example, Superdrug allows customers to buy online but doesn't use a large warehouse network for fulfillment. And there are definitely no robots involved. Instead, orders are shipped from stores. 

Its loyalty program also breaks free from the "points for prizes" crowd. Built to identify VIP customers – those who spend more than £300 a year – it leverages gamification and exclusive experiences to prompt higher spending, averaging a 35 percent additional spend per member per year. 

The chain also uses trigger campaigns on email to engage customers, using customer data on open rates for different campaigns to tailor which campaigns they will receive in future. That means that if a customer only engages with campaigns about fragrance, cosmetics campaigns will be suppressed. This is not, however, a hard and fast tule. There are some campaigns that all customers need to know, and over personalizing communications can mute a brand's voice. 

As Walburn said, customer intelligence and good retail still need humans. AI will never be more than a tool. 

ASOS and Sierra: An agent that can read the room 

As part of a standard CCaaS upgrade, ASOS assessed switching to a single platform that would meet its CCaaS needs and support the future roll out of AI agents. However, after assessing the options, it decided to go with separate vendors that were best-in-class for each capability.

Enter Sierra, the AI agent specialist that has worked with Ocado, Marshmallow, SiriusXM, CLEAR and others. It was recruited by ASOS to co-create new agent capabilities with a forward deployed engineer (FDE) seconded to the fashion retailer. 

The aim was to resolve, rather than deflect contacts with a conversational experience that would feel fully human. The new ASOS agent has been trained to reciprocate the emotions expressed by the customer and effectively read the room before responding. That means that when it feels it's appropriate, the agent will use colloquialisms such as "bestie" or include emojis in its messages. If a customer is shorter in tone the bot will keep it formal and to the point.

The approach responds to trends observed over the years by head of customer care experience and transformation, Kiran Uppal. During the session, she shared how people had become accustomed to curt interactions with rules-based bots, but when LLMs arrived, customer behavior changed. Conversations started to be more interactive and dynamic and the more human customers could be with the LLM – especially LLMs with memory – the more satisfied they were with the experience. 

The ASOS agent project went from design to full roll out – including all intents, journeys and scenarios – in six weeks. 

Testing took an A/B approach in the retailer's biggest market, the UK. Fifty percent of traffic had the old bot and the other 50 percent had access to the new agent, with satisfaction tracked across both. Once tested, the agent capabilities quickly scaled to all markets. 

Today, the agent isn't just handling conversations. New capabilities such as automated resolution and refunds were layered on as needed and the agent is also constantly analyzing performance and optimizing. It can identify and fix journey breaks, identify and report on points of friction and even flag reasons for complaints. 

ASOS is no stranger to conversational AI innovations. As early as 2017, the fashion retailer launched a "gift-guiding" chatbot to help customers navigate its extensive catalog. This was followed in 2018 by Enki, a proactive fashion assistant that personalized browsing and shopping recommendations based on a user's past interactions and purchases. Powered by Google Assistant, Enki also allowed ASOS to extend its reach into voice-assisted shopping, making it one of the first UK fashion retailers to adopt this cutting-edge technology.

Regulation may not improve digital accessibility but LLMs will

From electric toothbrushes to closed captions, there are hundreds of everyday innovations that benefit everybody, but started as accessibility features for a smaller group of users. 

The third Think Tank session of day one was called Beyond compliance: Creating accessible retail experiences across web, mobile and store. Hosted by Lee Lobb, retail partnerships manager at Level Access, part of the conversation focused on the progress that has been made since the European Accessibility Act came into force in June 2025. 

Attendees at this session heard how the Act covers pure physical journeys, pure digital journeys and those that mix the two, for example, self-checkouts – and that many journeys still fall between the cracks. 

Those in the room were eager to address accessibility better at their own organizations, but the main drivers of change were company values, rather than the financial, moral, or regulatory obligation to ensure all customers can access all journeys. 

CX Network's annual research into the state of CX found that despite new regulations, fewer than half of respondent organizations in 2026 were actively working to improve accessibility online. When respondents were asked how they adapted to new regulatory changes in the EU and US markets, 47 percent said they have taken steps to make digital customer journeys more accessible in line with new regulations, while 22 percent said they have not, and 33 percent were unsure. 

However, change may be on the horizon. Online, many of the site optimizations organizations are making for LLMs will also benefit accessibility. This means that market forces may still bring about change, even if it is for bots rather than humans. 

Quick links 

• How accessible is CX in 2026?
  
• All you need to know about All Access: AI + Data 2026
• The evolution of in-store CX: Strategies for enhancing engagement and loyalty