For two decades, the open secret of selling online in Australia was that you had to please Google. Businesses poured money into keywords, backlinks and page speed because a top spot in the search results was the closest thing retail had to a licence to print money. That logic is now being quietly rewritten, because a growing share of shoppers no longer type a query and scan ten blue links. They ask an AI assistant what to buy, and they act on whatever it tells them.
That shift is the backdrop to new advice circulating among marketers this week, which argues that the way most businesses build their websites is actively working against them in the age of generative search. According to analysis published by Dynamic Business, roughly a third of the average retail homepage is effectively invisible to AI tools, which means a large slice of the content a brand has carefully produced is never seen by the systems increasingly deciding what gets recommended.
Why AI can’t see your best work
The problem is less about bad writing than about how modern sites are assembled. A lot of the slick, interactive content that looks impressive to a human visitor, including product carousels, pop-up panels, content that only loads when you scroll and information tucked inside images, can be opaque to the crawlers and language models that feed AI answers. A shopper might see a beautifully designed page, while the AI reading it sees a fraction of the words.
The fixes being recommended are not glamorous, which is rather the point. The advice centres on making information explicit and machine-readable: writing plain descriptions rather than relying on graphics, using structured data so a price, a review score or a stock status is labelled clearly in the code, answering the obvious customer questions directly on the page, and making sure the important text is actually present in the page’s underlying HTML rather than summoned by scripts after the fact. In other words, the techniques that once felt like housekeeping for Google now determine whether a business exists at all in an AI’s mental map of the market.
This is the discipline the industry has started calling GEO, or generative engine optimisation, a deliberate echo of the SEO playbook that preceded it. FluentSea readers will recognise the theme from our earlier look at the emerging tactics Australian brands are using to surface in AI overviews and ChatGPT results. The novelty here is the blunt diagnostic: not that brands should optimise for AI one day, but that a measurable chunk of their current content is already being left on the cutting-room floor.
Two ways to read the warning
There are two reasonable responses to all this, and they are both worth holding at once. The optimistic reading is that GEO is simply good practice dressed in new clothes. Clean code, honest product information and clearly answered questions have always helped customers, and if they also help an AI assistant understand a catalogue, then the work pays off twice. On this view, the businesses that treated SEO as an ongoing craft rather than a one-off trick are well placed, because the underlying habits transfer.
The more sceptical reading is that brands are being asked to chase a moving target they cannot measure. Unlike a Google ranking, there is no public leaderboard showing where a company sits inside ChatGPT, Gemini or an AI-generated shopping summary. The models are opaque, their training and retrieval methods change without notice, and a tactic that works this quarter may be ignored the next. For a small retailer with no in-house developer, rebuilding a site so that every price and review is wrapped in structured data is real money spent on a channel whose returns are still hard to prove. There is also a fairness question lurking underneath, because the brands best equipped to feed the machines cleanly are often the largest, which risks AI search concentrating attention on incumbents rather than widening the field.
What it means for Australia
For Australian businesses the stakes are sharper than the global framing suggests. The local retail sector is dominated by small and medium operators who have spent years building a presence on Google and social platforms, and many simply do not have the engineering muscle to re-plumb a website on demand. If AI assistants become a meaningful front door to shopping, and the early signs locally point that way, then the gap between brands that publish in a machine-readable way and those that do not could translate directly into lost sales.
The timing matters too. We have already reported on Australian-first experiments such as Event Cinemas wiring session search and booking into ChatGPT, and on the way AI overviews are reshaping how local brands think about discovery. Those are not distant trends. They are the leading edge of a change that will reach suburban homewares stores and regional wineries soon enough. A business that is invisible to AI today is, in practical terms, invisible to a slice of customers who have already stopped using a conventional search engine as their first move.
There is a consumer angle as well. If a third of a homepage cannot be read by the tools customers increasingly trust, then the recommendations those customers receive are being made on incomplete information. That should concern anyone who cares about whether AI shopping assistants give fair, accurate guidance, because a model that cannot see half the market cannot genuinely compare it. For Australian shoppers in particular, there is a risk that AI answers skew toward large international brands with polished technical setups, at the expense of local operators offering a better product that the machine simply never registered.
What’s next
The practical takeaway is unusually clear for an AI story. Businesses can audit how their most important pages appear to a crawler, strip the critical information out of images and scripts, and label their products in structured data, all without betting the company on any single platform. None of it requires predicting which chatbot wins. It just requires publishing in a way the machines can actually read.
The larger question, which the industry has not answered, is whether GEO settles into a stable craft the way SEO eventually did, or whether it stays a guessing game dictated by a handful of opaque models. Australian regulators and industry bodies have so far said little about transparency in AI-driven commerce, and that silence is likely to be tested as more spending flows through assistants. For now, the message to local brands is pragmatic rather than panicked: the shopfront has a new audience, and that audience reads differently.
Sources: Dynamic Business.
