An article out of New Zealand this week made me think back to my last post about intention over panic. It’s the kind of story that can push a business owner straight into anxiety-mode.
RNZ reported this week that AI is changing how consumers shop, and businesses that haven’t prepared their content and pricing for AI agents risk being left out of the recommendation entirely.
But there are two things happening at once, and conflating them is where the anxiety creeps in.
What’s actually happening
Nearly half of consumers are already turning to AI tools somewhere in their buying journey, and research puts that figure at 45 percent globally. McKinsey has estimated that agentic commerce (where an AI assistant does some or all of the shopping on a person’s behalf), could orchestrate $1 trillion in US retail revenue alone by 2030, and $3 to 5 trillion globally. Woolworths has announced a partnership with Google to bring agentic AI into its ‘Olive’ chatbot here in Australia later this year, so it can plan meals and build shopping baskets on a customer’s behalf, and similar shifts are underway across the Tasman.
So agents are already instrumental in research, comparison and shortlisting. So they’re doing the finding, but research shows that humans are still doing the deciding, and the buying becuase consumers are sceptical about letting an agent actually complete the purchase. The gap between AI-assisted research and AI-completed purchase is still wide.
The survey out of the UK, found just 19 percent trust AI to make everyday purchasing decisions, compared with 55 percent who trust a human expert or adviser. A separate global study found only 17 percent of consumers trust AI enough to complete a purchase outright. Most major surveys through mid-2026 show trust in single figures to low twenties for UK and global samples.
Why this matters for your website
Whether an agent is shortlisting you for a person to choose from, or a person is doing the final research themselves after an agent narrowed the field, the mechanics matter.
Agents work by reading structured, accurate data, not by browsing a homepage the way a person does. That means the old-school SEO fundamentals are back in the spotlight, proper heading hierarchy so a machine can tell what’s actually important on the page, descriptive alt tags on product images, schema markup on pricing and availability, and product or service descriptions detailed enough that an agent doesn’t have to guess. If your information is thin, outdated, or doesn’t match what you actually charge, the agent moves on to a competitor, and the human never even sees you on the shortlist.
Why this isn’t only a checkout problem
For organisations not running ecommerce, the mismatch trap is still relevant.
Procurement teams are increasingly using AI tools to shortlist suppliers before a human conversation ever happens, even while the final decision and sign-off stays firmly human. If your site’s capability statements, policies, service scope, or case studies are vague, poorly structured, or inconsistent with what you actually deliver, you could be filtered out before anyone picks up the phone.
Getting the fundamentals right
The real trap isn’t falling behind on tools, it’s not knowing what problem you’re solving. In this case, the problem is specific and solvable:
- Is your website the single source of truth for what you offer, structured properly, and accurate, current and detailed
- Does your pricing or scope match what a customer actually experiences once they engage you
- Do you have real social proof, case studies, testimonials, evidence of service quality, sitting where an agent or a person can find it
Still the same message
Agents are shaping the shortlist, but people are still making the decisions. Whoever wins the AI recommendation still needs somewhere trustworthy to send the enquiry, and for now, that’s still your own website. Now is the time to get your own house in order, not to chase the next tool. Owning your narrative, on digital infrastructure you control, matters more than ever.
