AI Shopping Recommendations
That’s encouraging news for smaller brands, many of which assume AI assistants will simply default to the biggest names. Around a third of the merchants surveyed believed exactly that. The shopper data suggests otherwise, at least in some categories. Unfamiliar-brand purchases were highest in categories like beauty, at 63%, and lowest in electronics, at 43%. Even there, though, more than two in five AI-influenced purchases went to brands the shopper hadn’t heard of. Where the products are comparable and the reasons to choose one are clear, the shortlist is more open than many assume.
It also matters when AI enters the journey. The most common reason shoppers gave for their last AI-assisted purchase was comparing products they were already considering, at 39%. Only about one in five were looking for brand-new ideas. AI is mostly being used at the evaluation stage, the moment where a shopper weighs specifications, prices, reviews and suitability. That’s the work product pages and comparison content used to do, and it’s now happening inside an AI answer before the shopper reaches anyone’s website.
Australians to spend more this BFCM as AI shopping grows is what earns a shopper’s trust once they see a recommendation. The single biggest signal, chosen by just over half of AI shoppers, was the AI explaining why it picked that product. Next came recommendations that reflected the shopper’s specific needs, at 37%, followed by answers that cited sources or links they could check, at 33%, and those pointing to reviews or ratings, at 27%. Only around 3% said they trusted a recommendation without checking it at all. Shoppers aren’t taking AI on faith. They act when the recommendation comes with reasons they can verify.
Brands, meanwhile, are mostly working on something slightly different. The most common area of effort among the merchants surveyed was technical work, such as structured data, product feeds and schema markup, cited by 56%. Owned content followed at 39%, with reviews, community forums and third-party coverage further down the list. Asked to choose a single priority, only 4% of brands named third-party reviews. The technical work isn’t wasted. It’s what helps AI tools find and read product information in the first place. But it’s only half the job. Structured data gets a product onto the shelf. What gets it chosen is whether the AI has something specific and credible to say about it.
This is where product information becomes a commercial question rather than a technical one. An AI assistant can only explain why a product suits a particular need if the details are there to explain it: dimensions, materials, compatibility, use cases, care instructions, warranty terms, delivery times and how it compares with alternatives. Those details need to be accurate and consistent everywhere they appear, because an AI drawing on a website, a shopping feed and a marketplace listing that disagree has no good reason to trust any of them. And claims need support. A product that says it’s waterproof is less convincing than one that states its water-resistance rating and has reviews from customers who’ve used it in the rain.
The report also highlights a confidence gap. Some 61% of merchants worry about not showing up in AI answers or losing ground to a competitor. Yet 67% rate themselves as even with or ahead of their competitors. Both can be true, but few businesses actually test it. Measurement adds to the problem. Around a quarter of merchants couldn’t say how AI-referred traffic converts, and a similar share couldn’t say what it does to order value. As we covered in our recent post on measuring AI search, much of AI’s influence happens before a click and never appears in standard analytics, so the businesses that check directly will have a clearer picture than those relying on referral numbers alone.
There are a few practical steps any eCommerce business can take. Start by asking a few of the main AI assistants the questions your customers ask when comparing products in your category, and note whether your products appear, what the assistant says about them and whether the details are right. Then look at your best-selling product pages through a shopper’s eyes and ask whether they give an AI enough to explain why that product is the right choice for someone specific. Check that specifications, prices and stock agree across your website, feeds and marketplaces. And give reviews and third-party coverage a clear owner, because they’re among the signals shoppers say they trust, and one of the areas brands are least focused on.
The invisible shelf isn’t a reason for alarm, and it isn’t reserved for the biggest brands. It rewards the same things good retail always has: products that are clearly described, accurately presented and backed by evidence. What’s changed is that the first reader of that information is increasingly an AI, and it will only recommend what it can explain.