What Retailers Need To Know About AI Shopping
For years, retailers have adapted to new ways of selling online. First came search, then marketplaces, mobile commerce and social commerce.
Now, the next shift is beginning to take shape: AI-powered shopping.
At the centre of Google’s strategy for Gemini AI Search is the Universal Commerce Protocol (UCP), a framework designed to allow AI assistants such as Gemini to discover products, compare options, build baskets and help customers complete purchases.
While much of the conversation around AI shopping has focused on the consumer experience, retailers are starting to ask a different set of questions. What happens operationally? How will inventory be managed? What if product data isn’t accurate? And perhaps most importantly, what role will retailers play when AI agents become part of the customer journey?
What is Google’s Universal Commerce Protocol?
Google describes UCP as an open standard that allows AI systems to interact directly with retailer commerce platforms.
In simple terms, it acts as a bridge between retailer systems and AI shopping assistants. Rather than simply sending users to a website, AI tools can access product information, availability, pricing and checkout processes through a standardised framework.
The aim is to make AI-powered shopping more seamless while allowing retailers to remain in control of transactions and fulfilment.
Google launched UCP in early 2026 and is currently rolling it out through AI-powered shopping experiences. While availability remains limited today, it is clear that Google sees AI-assisted commerce as a major part of the future shopping experience.
Retailers Have Been Cautious, and Understandably So
The idea of AI agents interacting directly with retail systems isn’t something every retailer has immediately embraced.
As Dream Agility’s CEO Elizabeth Clark noted:
“As of last summer, at the Barclays Annual Retail Conference, larger high street retailers were unwilling to connect with AI agents as there are a number of potential problems and they had security concerns.”
It’s easy to understand why.
Retailers have spent years building customer experiences, managing promotions, optimising product merchandising and investing in loyalty programmes. Handing part of that journey to an AI assistant naturally raises concerns.
However, as Google’s plans become clearer, the conversation is shifting from whether retailers should engage with AI commerce to how they can prepare for it.
The Questions Retailers Are Already Asking
One of the most interesting aspects of UCP is not the technology itself, but the operational questions it raises.
For example, what happens if a customer adds products to a Universal Cart but never checks out? Does stock become reserved? How often is availability refreshed? What happens if a product sells out between recommendation and purchase?
Based on Google’s documentation, retailers remain the source of truth for stock availability and checkout validation. In theory, that means retailers maintain control.
The challenge is that theory and reality don’t always align.
Many retailers still struggle with inventory accuracy across ecommerce platforms, stores, marketplaces and fulfilment systems. If an AI assistant recommends a product that is technically available but cannot actually be fulfilled, the customer experience quickly falls apart.
This is where data quality becomes critical.
The Hidden Risk: Parent-Level SKUs
One area that deserves more attention is product structure.
Many retailers, particularly in fashion and homeware still manage products primarily at parent level.
For example, a product feed may show a particular trainer or dress as available, while stock accuracy actually exists at the child SKU level where sizes and colours are managed.
Traditionally, this hasn’t always been a major issue because customers browse product pages and make selections themselves.
AI shopping changes that.
If an AI assistant is recommending products on behalf of a customer, it needs confidence that the exact variant requested is available. Otherwise, retailers risk creating frustrating experiences where recommended products cannot ultimately be purchased.
At Dream Agility, we’ve spent years helping retailers improve product feed quality and data accuracy. Historically, that has been important for visibility within Google Shopping and other channels.
With AI commerce, it becomes even more important because product feeds are no longer just a marketing asset. They become part of the purchasing journey itself. With data feeds expanding to offer customers more information than would normally be contained on the web page, data just got more interesting and a lot bigger!
Why Data Quality Could Become the New SEO
For years, retailers focused on rankings, traffic and click-through rates.
Those metrics aren’t going away, but AI shopping introduces a new consideration.
AI systems rely on structured, accurate product data to make decisions.
That means retailers need to think carefully about:
- Inventory accuracy
- Product attributes
- Variant-level data
- Pricing consistency
- Feed quality
- Additional fields not previously required for PMax.
The retailers with the cleanest and most reliable data are likely to be the retailers AI systems trust and recommend.
This is an area where Dream Agility is already helping retailers prepare. Whether it’s feed optimisation, Merchant Center management or improving product data quality, the foundations that drive strong shopping performance today are likely to become even more valuable as AI commerce develops.
Is Your Product Data Ready for AI Shopping?
For many retailers, the biggest challenge may not be connecting to UCP.
The bigger challenge may be preparing the data behind it.
Retailers should be asking themselves:
- Do we have accurate inventory data?
- Are our product feeds complete?
- Can we trust our variant-level information?
- Are we maximising all the fields available to inform the AI?
- Are prices consistent across channels?
- Would an AI assistant understand our catalogue?
These questions aren’t new, but the consequences of getting them wrong are changing.
As AI-powered shopping becomes more sophisticated, weaknesses in product data will become increasingly visible.
Looking Ahead
Google’s Universal Commerce Protocol is still in its early stages, but it provides a fascinating glimpse into how shopping could evolve over the next few years.
Consumers may increasingly move from browsing websites to asking AI assistants to find products, compare options and complete purchases on their behalf.
Whether that future arrives quickly or gradually, one thing is becoming clear: retailers will need strong data foundations to succeed.
As Elizabeth Clark our CEO puts it:
“The retailers who succeed won’t necessarily be the first to connect to UCP. They’ll be the ones who have confidence in the quality of their product data, stock accuracy, and customer experience when AI agents begin shopping on behalf of consumers. Currently it’s only compatible with Shopify, Woo Commerce, Salesforce, and Big Commerce. If you’re on another platform you can apply for the pilot (if you’re in the U.S, Canada or Australia) you’ll have to make adaptations to your platform within 30 days of the start of permission to join to be able to participate in the results on Gemini Search”
The question for retailers is no longer whether AI commerce is coming.
The question is whether their product data, inventory management, and ecommerce infrastructure are ready for it. As this field rapidly evolves, we’re already helping retailers build those foundations and are happy to help with any questions you may have.
