There is an uncomfortable finding making the rounds in the search world this week, and it matters for any Australian small business hoping to be found in the age of AI answers. The SEO and AI search analyst Lily Ray, whose read on Google updates the whole industry watches, published analysis in mid June 2026 showing that AI search will cheerfully cite a brand's self-promotional content, then recommend a competitor instead. In plain terms, your own page can end up working for the other side.
The pattern she describes goes like this. A business publishes a glossy article declaring itself the best plumber, the best accountant, the best cafe in town, often with a tidy list of rivals underneath to look balanced. When AI search builds an answer, it treats that page as a useful source and quotes it. But when it decides who to actually recommend, it frequently reaches past the author and picks one of the competitors named on the very same page. By her reading, in a large share of cases, roughly two in three, the brand that wrote the article is left out of the recommendation entirely.
That distinction, between being mentioned and being recommended, is the whole game now. For years the goal was simple: rank on the first page and earn the click. AI search has quietly split that goal in two, and most small businesses have not noticed the seam.
Being cited is not the same as being chosen
When someone asks an assistant for the best option in your field, two different things happen under the hood. The AI gathers sources it considers relevant, and it forms a recommendation. These are not the same list. Google's own description of AI Overviews, the AI-generated summaries now sitting at the top of a growing share of searches, makes clear they pull key information from across the web and surface links to dig deeper. The footnote that cites you is a long way from the sentence that tells the customer where to go.
GEO, or generative engine optimisation, is the new discipline of getting recommended by these AI answers rather than just appearing in the old blue links. And the lesson from this week's analysis is blunt. The signals that once helped you, a confident sales page, a self-ranking list, the word best repeated often enough, can now actively count against you. The machine reads your list of competitors as a genuine list of options, and you have handed it the alternatives.
Why telling everyone you are the best now backfires
The deeper shift is about who the AI trusts. A claim you make about yourself is, to an assistant weighing options, the least reliable evidence in the room. What it leans on instead is what the rest of the web says about you: the mentions, reviews, comparisons and references that you did not write. If those point at a competitor, that is who gets recommended, no matter how polished your own page is.
This is genuinely good news for honest operators, and bad news for anyone hoping to talk their way to the top. You cannot game your way into an AI recommendation with adjectives. You earn it by being the business that other people, and other trusted sources, already point to. That is harder than writing a better headline, and it is exactly why it is worth doing well.
Telling an AI you are the best is the weakest evidence in the room. It decides who to recommend on what everyone else says about you.NextAura
What actually earns an AI recommendation
You do not need to chase every new AI search feature, and you should be wary of anyone selling a magic checklist. What matters is understanding what good looks like, so the assistant fielding a customer's question puts your name forward with confidence. The same thinking applies whether a customer is searching on Google or asking an AI assistant to do the choosing for them.
- Your reputation lives in places you do not own: reviews, mentions, and references on sources the AI already trusts, not just your own about page.
- Your site states plainly what you do, who you serve, and where, so an assistant can match you to a real customer question without guessing.
- Self-praise gives way to evidence: specifics, results and third-party validation an AI can take at face value rather than discount.
- You are the clear, consistent answer to the questions your customers actually ask, across every place an AI might look for one.
- You measure the right thing, whether AI answers recommend you, not just whether your page got cited somewhere in the footnotes.
The honest first step is not rewriting your homepage with more superlatives. It is working out what AI search currently says when a customer asks for the best in your field, and whether that answer sends them to you or quietly to someone else. Most owners have never checked, and the answer is often a surprise.
This is exactly the work we do at NextAura. We help Australian small businesses get recommended by AI search, not just mentioned in passing, by building the reputation and AI search visibility that makes an assistant confident to put your name forward. If you would rather have people who track these shifts daily handle it, get in touch and we will find out who AI is recommending in your field, and make sure it is you.