AI can only sell what it can understand
Most webshops are used to optimising for people: great images, good UX and a purchase journey that feels easy. But shopping is increasingly influenced by AI. Not because customers stop visiting webshops, but because AI is playing a greater role in helping find relevant products, compare options and identify the best choices – and in some cases bringing the customer closer to checkout.
This changes one central thing: products are increasingly evaluated based on the data that exists about them – not just how they are presented on the website itself.
That is why the biggest hurdle for many webshops right now is data quality. AI can only work with the product information available in the webshop. If data is incomplete, inconsistent or buried in free text, it becomes difficult for AI to read and use products in a consistent way.
Preparing for the future of shopping therefore does not start with new tools, but with the existing setup. For many, this means increased focus on the e-commerce platform and potentially a PIM system where product data is created, structured and maintained.
From SEO and feeds to AI-readable products
If you work with SEO or Shopping feeds, you already know the logic: the better the structure, the better the opportunities for visibility. AI simply makes the requirement clearer.
A human can often figure things out along the way:
- Click through to find delivery information
- Find information in the FAQ
- Understand a variant from images
- Guess what is included
AI cannot guess in the same way. It typically chooses what is clearly stated.
If AI cannot find:
- A clear price
- Which variants are available
- Delivery time
- Return policy
…the risk is that the product gets rejected – before the customer ever sees it.
What does this mean in practice for webshops?
It does not mean everyone needs to build something new tomorrow. It means webshops should ensure their products are described so clearly that both people and machines can understand them.
Think of it as future-proofing: the better and more consistent your product data is, the stronger your position when shopping increasingly happens with AI as an intermediary.
Product data is more than what the customer sees
It is one thing for customers to be able to read information on the webshop. It is another whether that same information actually exists as structured data in the back end. This is typically where feeds, platforms and AI retrieve details about products, variants, pricing and availability – even when they do not read the webshop the same way a human does.
At the same time, there is a lot of product data that you would not necessarily show on the front end, but which should still be filled in on the back end. Examples include:
- Country of origin
- Care and safety information
- Certifications and standards
- Logistics details such as carton size, freight class or package dimensions
The more complete and consistent that data layer is, the easier it becomes for systems and AI to understand, compare and present products correctly in future shopping flows.
Google has already described what they expect
Google has already published documentation and guides describing the requirements they place on product data in AI-based shopping flows. This applies across Google Search, Merchant Center and new AI surfaces.
This is a clear signal that structure and data quality are not just a trend, but something being operationalised in practice. Webshops that work systematically with their product data will therefore be better positioned when AI plays a larger role in the purchase journey.
Checklist: what you should get in order in your product data
1) Clear product information
Start with the basics:
- Product names that explain what the product is (not just brand + model name)
- Descriptions that answer: what is it, what does it do, and who is it for
- Attributes such as size, colour, material and compatibility written clearly and consistently
The more unambiguous, the better.
2) Variants and stock that make sense
If you sell in variants, it should be easy to understand:
- What the difference between variants is
- Which are in stock
- Which are sold out
When variants are unclear, it is difficult to choose correctly – both for customers and for AI.
3) Prices and offers without hidden rules
AI and customers react the same way to unclear pricing: they lose trust.
Make sure you have:
- A clear base price
- A clear before/now price for offers
- Clear conditions for discounts that are not buried in lengthy text
4) Delivery and returns should be easy to find and understand
Delivery and returns are often what determines the purchase – especially when comparing options.
Make sure you have:
- A realistic and clearly stated delivery time
- Delivery cost or free shipping threshold that is easy to understand
- Return policy written in plain language
5) Checkout that is simple and predictable
The fewer surprises in checkout, the better.
Make sure you have:
- Familiar payment options
- A checkout flow that is easy to complete
- No unnecessary steps or requirements that make the process cumbersome
This benefits both conversion today and resilience in future shopping flows.
Platforms, infrastructure and differences between solutions
It is also worth noting that webshops on Shopify often have an advantage, because the platform is already built around standardised flows, consistent product data and close integration with Google's commerce ecosystem.
For webshops on other platforms, there may be more things that need to be structured manually. This does not mean they are worse off, but that the work on data quality and structure becomes even more important.
At the same time, Google is not the only player moving in this direction. Other AI platforms are also working on purchase-oriented flows, which underlines that this is not about one specific solution, but about ensuring product data can be used outside the webshop itself.
Why it is urgent – without causing panic
AI shopping is still evolving. But the pattern is already clear: those with the best data are easier to find and easier to choose.
Poor product data does not mean your webshop is poor. But it can mean you become harder to choose – especially if AI is helping customers filter options.
This is not something everyone needs to fix tomorrow. But it is something all webshops can benefit from starting to work on now.
