Retail Economics, in partnership with Metapack, published its Ecommerce Delivery Benchmark 2026 report at the beginning of February, highlighting how increased AI adoption is shaping e-commerce and delivery trends.
Parcel and Postal Technology International’s editor, Hazel King, sat down with Richard Lim, the CEO of Retail Economics, at The Delivery Conference in London on February 3, to discuss the report in more detail.
Can you give us a high-level overview of what this year’s Ecommerce Delivery Benchmark Report revealed?
The customer journey has become so much more complex over the last few years, and Gen AI has added more complexity to that. The industry is now trying to figure out the touchpoints of how AI is changing how consumers are researching products, discovering products and how those products are served through different AI platforms. In the report, we talk about different phases of AI. The first phase, where we are now, is very much around AI-assisted commerce, where consumers are using platforms like Claude and ChatGPT to help in that research phase. There are also on-site AI agents such as Rufus on Amazon, which provide a different prism. And then there are agentic AI browsers such as Comet that add another layer of complexity.
In relation to how that feeds through to what retailers and brands need to do to react to this shift, a lot of it is just about discoverability. How do retailers make sure that when consumers are searching on an AI platform, for example, ‘high-resolution 55-inch TV that costs £1,000 and can be delivered next day, and what is the returns policy?’, that their product is discoverable in this new platform of interaction. Within the research we look at that. We look at how it differs between different types of consumers and different product categories to identify which products are likely to be affected first by the use of AI in online shopping research.
Are shopping habits changing, and how is this affecting e-commerce fulfillment?
One hundred percent. Within the research, we found that 28% of consumers have specifically used AI chatbots and platforms to assist their shopping habits – that’s a huge amount. That figure goes up considerably when you look at younger generations –almost 50% of UK shoppers under 45 are using LLM platforms for specific retail queries.
When it comes to fulfillment and delivery expectations, one of the critical components for consumers when they’re selecting a product to purchase is around delivery options, and that is one of the attributes that consumers use when using these AI platforms – they are more likely to choose a brand based on delivery rather than on the actual brand they’re buying. Obviously, there are lots of other factors that impact purchase decision, but delivery has been shown to be a real differentiator. Therefore, retail brands need to make sure they’re surfacing that delivery information when consumers are doing their product research.
How is AI influencing reverse logistics?
From a consumer perspective, returns plays a really important role in terms of making sure the policy is clear, easy and frictionless – making all of that information easily accessible on those AI platforms is key for brands. From a retailer angle, they are looking at how they can use AI technology to help them reduce return rates; for example, better sizing in clothing or utilizing and leveraging consumer data more effectively.
We’ve seen a shift over the last 12 months toward increased profitability. It is no longer a case of ‘growth at all costs’; retailers are charging for returns now and also segmenting their customer base using more sophisticated AI technology to identify different consumers, and having different returns policies depending on how each consumer shops and how profitable they are.
What are the barriers to AI adoption?
From a consumer point of view, trust is the biggest barrier. Generally, younger consumers are more trusting of social media and technology, so they are more likely to adopt AI. There are other barriers in terms of retailers not implementing technology at the same rate – there is an uneven landscape around how businesses are adopting AI and how that is impacting their operating modes.
There is also a technical barrier around payments. There is currently no standard protocol that will allow AI agents to talk to other AI agents to enable payments across different structures.
Have you got any predictions for the next 12 months for further AI developments in retail and delivery?
We have outlined different stages of agentic commerce. The first is AI-assisted commerce, which is what has happened over the last 12 months, and we’ll see further development of semi-autonomous agents in 2026. Then we have fully agentic AI – this is where, depending on our objectives and the criteria we set, AI agents will be going and purchasing things without our intervention. Things such as renewing car or home insurance, for example. And the final one will agent-to-agent commerce – this is where my AI agent will be talking to a retail AI agent and will be able to negotiate the best price or quality, and will be able to go and speak to different agents to find that information. There is already a platform called Moltbook, which is a social media platform for AI agents, where they are independently having conversations with each other. For example, ‘my human has asked me to do this’ and another agent can respond with advice or information on how to approach the problem. We have predicted agent-to-agent commerce by 2028 or 2030, but I think perhaps those predictions are too generous in terms of time because these things are already happening.
Read more about the report here