Google has now rolled out AI Overviews and AI Mode across most European markets – France since 22 July 2026, after the UK and the rest of the EU. The engine synthesises multiple sources, offers an initial answer, then lets the user refine their search through a conversational exchange.
For brands and retailers, the stakes go beyond organic traffic. Google now takes part in product discovery, comparison and shortlisting, before a product page is ever opened. Does it have information clear, reliable and current enough to understand those products and present them correctly?
Google builds an answer from purchase intent
AI Overviews condense the landscape into a single answer; AI Mode carries the search further without losing context. It relies in particular on query fan-out: Google breaks a request down into several sub-topics and runs the corresponding searches. For a sunscreen, it might explore protection level, finish across skin tones, price and availability separately, then cross-reference the results.
In retail, keyword search is therefore giving way to multi-criteria intent. A request such as “SPF 50 sunscreen for dark skin, no white cast, under €25” combines a category, a profile, a benefit and a price, and can then continue with further questions.
SEO remains essential: a page has to be accessible and indexable. But ranking well on a generic query is no guarantee that a product will be selected once the request becomes more specific. AI interprets the criteria, compares the offers and rephrases whatever information is available.
What Google needs to understand before it presents a product
A generative engine can only work with information that is attached to the right product, relevant to the query and still up to date. Visibility therefore depends first and foremost on product data quality.
Brand, SKU, GTIN, variants and compatibilities make each item identifiable. Uses, composition, certifications and verifiable benefits explain why it meets a need. In our sunscreen example, “high protection with a pleasant finish” carries far less substance than “SPF 50+, fragrance-free, invisible finish, tested on dark skin tones”.
The information also has to still be true at the moment Google uses it. The engine can cross-check what it reads on the page, in the structured data and in the Merchant Center feed. An attribute, for example a product format or a flavour, that is correct in the PIM but out of date on a marketplace, or different again at a retailer, is unreliable for Google and for the shopper alike. The risk compounds for brands selling across several European markets, where the same reference is described in different languages, by different retail partners, under different assortment and labelling rules.
Google is also evolving Merchant Center, the platform through which brands and retailers submit their catalogues to the engine. New fields now accommodate Q&As, documents and links between products. They help the AI understand and compare items, without guaranteeing they will appear in its answers.
How to prepare your product data for these new journeys
There is no specific markup and no recipe that guarantees a presence in Google’s generative experiences. Three priorities nonetheless stand out:
- Start from real-world requests. Which uses, profiles or constraints do consumers actually mention, and how do those vary from one market to the next? Check that the corresponding criteria exist among the catalogue’s usable attributes. The goal is to enrich product records wherever information is missing, vague or hard to compare.
- Consolidate and synchronise the data. Your website, Merchant Center, marketplaces and retail partners all use different formats, yet they have to describe the same product, in every language and every country you sell in. Centralised management and tailored syndication limit discrepancies in marketing copy, imagery and assortment.
- Monitor the generated answers. Do your products appear on strategic requests? Which attributes are they associated with? Is the information accurate? Performance is now also measured by how generative systems discover, describe and compare products. Google does plan its own reports in Merchant Center (share of voice, search terms, attribute completeness score), but these are currently live only in certain English-speaking markets outside Europe. For European retailers and brands, measurement therefore still relies on third-party solutions.
Product data: the raw material of the answer
AI Overviews make neither product pages nor SEO obsolete, they make their weaknesses more visible. Every new question can surface a criterion that the catalogue either documents precisely or leaves in the dark. And a gap that stays hidden in one market tends to reappear in the next.
Centralising, enriching and syndicating data is no longer enough: you also need to control how it is rendered online. When Google can answer before the product page, being present isn’t enough. You still have to be understood, and correctly presented.
Is your product data ready for these new journeys? Equadis helps you enrich it, make its syndication reliable, and control how it is rendered online, across your markets.
Talk to our experts to review where you stand.
