Shopping through AI assistants is growing, but it is still a minority habit: only one in ten consumers starts a product search by asking an AI. The decision worth making now is not to launch a shopping agent but to get your product data in order. When adoption speeds up, assistants will choose between well-described catalogues, and the ones that are not will simply not show up.
What happened
A recent survey of more than 9,000 US consumers shows that only 10% begin a product search with an AI tool. Among younger shoppers (Gen Z and millennials) the figure rises to 14%. The real gap is income: almost 70% of people earning $150,000 or more a year say being able to shop with AI matters to them, compared with 53% of those earning under $45,000.
The study’s conclusion is clear: most people are not yet handing purchases to an agent, but adoption will grow, starting with replenishment items and easily comparable products. Its main recommendation to retailers is to invest in product data quality, because an AI assistant reads attributes, availability and price, not the creativity of your storefront.
Why it matters to your business
For a leadership team, there are three layers to this.
Sales: no rush for the agent, real urgency for the catalogue. At 10% adoption, your own shopping assistant is unlikely to move the P&L this year. An incomplete catalogue, however, is already costing you today: in on-site search, in Google Shopping, on marketplaces and in the answers of the assistants your customers do use. The same investment pays off across all of those channels at once.
Margin: the market splits in two. The study expects the premium end to keep depending on brand and relationship, while the lower end will be decided on specs and price. If you sell comparable products, an agent that compares attributes and price will squeeze your margin. If your attributes are poorly defined, you will not even make the comparison. Mid-market brands have to handle both forces at the same time.
Risk: losing the customer relationship. If the assistant decides what to see and what to buy, a brand can be reduced to one supplier on a list. That is why several large US retailers are already launching their own assistants: they want the conversation to happen on their turf.
One more point for those selling to higher-income customers: they are exactly the ones who value shopping with AI most, often as a styling or curation adviser. In mid-to-premium fashion, beauty or home decor, the window may open sooner than the average suggests.
Decisions to make
- Audit the catalogue through a machine’s eyes. Take the 100 or 200 best-selling SKUs and check whether they carry every attribute a buyer would use to compare: dimensions, materials, compatibility, sizes, real availability, up-to-date price. It is weeks of work, not months, and the return shows first in on-site search and Shopping campaigns.
- Appoint an owner for product data. In many mid-sized companies the product page belongs to everyone and no one: purchasing, marketing and eCommerce all edit it separately. Without an owner and a single source of truth, the catalogue degrades every season.
- Decide which part of the range competes on price and which on brand. The strategy for replenishment items is not the same as for signature products. That choice drives where you invest in content, pricing and channels.
A useful question for your next leadership meeting: if an AI assistant had to recommend our ten best-selling products, would it have enough information to do it well?
What we would do at TRILOGI
Before talking about shopping agents, we would sort out the data. We would start with a diagnosis of the highest-revenue part of the catalogue, connect ERP and eCommerce through a PIM as the single source for attributes, prices and stock, and leave the catalogue ready for search engines, marketplaces and AI assistants. That is what our PIM integration projects do: describe each product well once and deliver it consistently to every channel.
With that foundation in place, deciding whether you need your own assistant stops being a bet and becomes a matter of numbers.
Every Tuesday, the 5 news stories that matter: https://trilogi.com/blog/#newsletter
