Shopping journeys are becoming more conversational
Customers no longer just type "black shoes fast delivery" into a traditional search engine. They ask longer questions, compare options, request recommendations and sometimes go through assistants that summarise the available offers.
For a merchant, this changes the priority: a beautiful product page is no longer enough. Product information must be accurate, structured, accessible and consistent everywhere: shop, merchant feed, marketplaces, search engine, AI chat, emails and customer service.
What actual adoption shows, with France as an example
Figures from 2026 on the French market, where we are based, confirm that this is no longer a marginal phenomenon, while revealing adoption that is still uneven across generations and usage that does not always lead to a completed purchase:
French online shoppers using generative AI in their purchasing journey: 31% (around 12.8 million people).
Young people aged 15 to 24 using AI for their purchases: 49%.
Users who have never completed a purchase through a conversational tool: 52%.
That last figure deserves attention: despite strong adoption for searching and comparing, more than one user in two has never yet completed a purchase directly within a conversational tool, often for fear of unsuitable advice or commercial bias. The role of these tools today is therefore mainly that of a research and comparison assistant, which then sends people on to the merchant's site to complete the purchase. That makes catalogue quality all the more decisive in turning the visit into a sale.
Product data becomes the key battleground
An assistant or an AI-enhanced search engine cannot guess your margins, stock levels, lead times or returns conditions. If your data is vague, contradictory or incomplete, the buying experience becomes fragile, and a customer sent to your site may abandon when faced with information that does not match what the assistant told them.
The points to make reliable first:
clear product titles, without piling up pointless keywords;
useful descriptions, with benefits, dimensions, materials, compatibility and limitations;
clean variants: size, colour, pack size, model, availability;
consistent, lightweight images, properly named and attached to the right product;
prices, promotions, delivery charges and lead times that are genuinely up to date;
a readable returns policy;
Product, Offer and Review structured data where relevant;
a correctly maintained Google Merchant Center feed or equivalent.
What this changes for Shopify, WooCommerce and marketplaces
On Shopify, WooCommerce or a marketplace, product fields should no longer be seen as an administrative formality. They are commercial data. A poorly filled-in variant can lead to a bad recommendation, an undisplayed stock-out can block a sale, and a returns policy that cannot be found can cost the customer's trust, in a context where AI amplifies the impact of these errors by repeating them in every recommendation.
Shopify's announcements around AI and more agentic storefronts point in this direction: shops need to expose their information better so that it can be used in more automated journeys. This ties in with our analysis of Shopify Editions 2026.
Why so many users search without buying directly
The figure of 52% of users who have never completed a purchase through a conversational tool is largely explained by two documented fears among French users: 46% worry about advice unsuited to their real needs, and 28% suspect commercial bias in the recommendations the assistant offers. These reservations will not disappear overnight, but they reinforce a strategic point for merchants: the AI assistant currently acts as an adviser and comparison tool, while the final decision and trust are still very largely won on the merchant's own site, at the moment the customer checks with their own eyes what was recommended to them.
The merchant checklist
Clean up product titles and descriptions.
Standardise variants and attributes.
Check stock levels and delivery lead times.
Publish a simple, visible returns policy.
Add product FAQs where the same questions keep coming up.
Implement the relevant structured data.
Monitor merchant feed errors.
Link product pages to guides, comparisons or advice pages.
Not content for robots, but clean data
Should you produce special content for AI assistants, a "ChatGPT version" of your catalogue? The answer is no, and that is good news. It is not about adding an extra layer, but about making what already exists reliable. Clear titles, complete descriptions, up-to-date prices and stock, clean variants and properly implemented structured data serve three audiences at once: traditional search engines, AI assistants and, above all, your direct customers.
That is the whole point of the exercise: the work that prepares a shop for assisted journeys is exactly the work that improves the experience of a human visitor. An accurate, honest product page converts better whatever the entry channel. Optimising for AI is therefore not one more constraint; it is a useful excuse to finally clean up a catalogue you kept putting off.
Consistency, the first condition of trust
One point deserves particular vigilance: the consistency of information from one channel to another. The same product showing one price in the shop, another in the merchant feed and a third on a marketplace sends a disastrous signal, to AI and customers alike. When an assistant announces availability or a lead time that does not match what the visitor finds on arriving at the site, trust breaks in the first second, and the sale with it.
Checking this consistency is the best test of whether a catalogue is ready for these new journeys. The same product must show the same price, the same availability and the same lead times everywhere, without exception. It is unspectacular data hygiene, but it is precisely what makes the difference when AI repeats your information at scale: an isolated error becomes an error reproduced in every recommendation.
Structured data, the common language of machines
Among the tasks to carry out, one deserves to be spelled out, because it seems technical when it has become essential: structured data. This is discreet markup, invisible to visitors, that describes a product in a language machines understand unambiguously: here is the price, here is the availability, here is the average rating, here is the brand. Search engines and AI assistants alike rely on it to display and recommend an offer correctly.
Long regarded as a refinement reserved for large sites, this data is becoming a baseline, even for a small shop. As discovery increasingly goes through automated systems, a properly marked-up catalogue is better understood, better presented and less exposed to misinterpretation than one that leaves machines to guess. It is little work for a benefit that grows every month.
AI rewards clean catalogues
The challenge is not to "sell to a robot", but to prevent poorly structured information from stopping a customer from buying, whether that customer arrives directly on your site or through an AI recommendation. The more assisted shopping journeys become, the more strategic catalogue quality becomes. A merchant who masters their product data gains visibility, reduces errors and makes conversion easier, whatever channel the visitor comes through.
We can audit your catalogue, product pages and feeds as part of our e-commerce maintenance and support service or an online shop creation project. Contact us to get your shop ready.
Sources: Campioni, 31% of French shoppers use AI to buy (in French) · Siècle Digital, shopping in ChatGPT is not taking off (in French) · Google Search Central, product structured data