Why Retailers Must Get Their Data in Order Before They Can Embrace Agentic AI

In our last post, we asked how close we are to living in a world where AI can shop for us, with as little or as much of our input as we choose.

The consensus is that the emergence of autonomous shopping will likely be driven by consumer demand. The fundamental technology required is already with us, as so-called agentic AI rapidly evolves so AI tools can turn their phenomenal data processing capabilities into reliable, independent decision-making.

Autonomous shopping ultimately relies on AI getting good enough at knowing what we want to be able to make our purchase decisions for us. Use of AI to search for products and make recommendations is rising rapidly. Consumers already appreciate the potential benefits of having AI agents constantly scouring markets on their behalf and proactively surfacing great deals and timely suggestions. It’s only a small step to then allowing the same AI tools to make certain purchases without needing human approval.

For consumers, it’s all about the two magic C’s – convenience and choice. Sometimes it’s fun to shop for things yourself. Sometimes it’s a chore. Wouldn’t it be great to have a personal AI shopping assistant that could take care of those times when it is a chore?

Technology will make this possible within a few short years. And in the meantime, agentic AI will continue to play more and more of a supportive role, from comparing reviews to scouring the best deals, making the shopping experience ever more efficient and satisfying.

As with all emerging consumer-led trends, retailers who respond early will gain a significant competitive advantage. But there are significant barriers to overcome before retailers can take advantage of this opportunity. And top of this list is getting to grips with data.

Data is the point of difference

Data is the fuel that powers AI. Getting ready for a future where agentic AI plays more and more of a role in shopping journeys therefore carries significant data demands for retailers.

Picture these examples. A shopper tells the AI shopping assistant they have installed on their phone that they want a new pair of shoes. The AI already knows their size, plus style and colour preferences, and typical budget. Its job is to scour all available options online and come up with a shortlist of recommendations. But some shoe retail sites have better data than others. Some are hard for the AI to read, have incomplete, conflicting or out-of-date information. The AI agent simply overlooks these, and makes its choices from the sites where product data is well-organised and complete. Retailers become visible to the customer on the strength of their data.

Or imagine a scenario where a customer is already in a shoe shop, but decides to use their AI assistant to check what other styles the store has in stock, or what user reviews have to say about a certain product, or to look up fulfilment options etc. But the store hasn’t yet got its house in order data wise, and the AI can’t find what it’s looking for. The customer leaves unsatisfied.

The data gap facing retailers

The challenge for retailers is that dealing with AI agents means dealing with an entirely new type of audience – one that communicates in the language of data. And yet data is proving a sticking point in retailers realising their AI ambitions. This recent survey highlighted what it calls a ‘confidence gap’ between what retailers want to achieve with AI and their own appraisal of their current capabilities. Much of the lack of confidence stems around data.

For example, retailers recognise that AI’s biggest opportunity is its potential to personalize shopping experiences in real time. And yet three of the top five barriers cited as blocking real-time personalization relate to data – delays in making data available and actionable (56%), limitations to personalizing at scale (63%) brought on by shortages in first-person data, and, spelled out clearly, “a lack of agent-aware AI decisioning” (58%).

Agent-assisted shopping, however long it takes to transition into semi- or fully-autonomous AI purchasing, is in many ways next-level personalization. Rather than collecting sufficient data to target the right message at the right customer at the right time, it’s about making enough data available in the right way so that shoppers, through their AI intermediaries, can in effect take full control of the shopping experience.

But that feels a long way off if businesses are struggling with outbound personalization. Retailers need to learn to walk before they can run when it comes to agentic AI.

AI to the rescue

The good news? AI can be retailers’ friend in laying the foundations for autonomous shopping. Rather than getting carried away by the hype and dreaming of ‘full agentic roll out’, retailers would be better served considering how AI can help with the immediate challenges they face around data. AI can, for example, be applied to the task of making data assets agent-ready, from pulling complex, messy, unstructured data sources into some kind of structured (and therefore bot-friendly) order, to using its pattern recognition capabilities to filling in gaps in data to improve quality.

In summary, then, 2026 might not be the year when AI agents take over retail. But it can be the year when retailers take a step closer to an agent-led future by utilizing AI to get their data in order.