Search has quietly stopped being a list of blue links and become a single answer. For Australian agents, that shift decides who gets the appraisal call — and the data says most agencies aren’t in the conversation. The good news: the asset that wins this game is one you already own.
A seller in 2026 doesn’t always open Google, scroll three agency sites and fill in an appraisal form. Increasingly they open ChatGPT, Gemini or Perplexity and type something close to: “Who are the best agents to sell my house in [suburb]?” The tool doesn’t return ten links. It returns a short, confident list of names. If yours isn’t on it, you were never in the running — and you’ll never know the call didn’t come.
Tom Panos has been hammering this point all year: the business is shifting from a cold-calling operation to a brand-led one, and owners are now asking AI which two or three agents to call. The agents who surface are the ones who’ve stayed visible and consistent. Everyone else is optimising for a version of search fewer of their clients still use.
This isn’t a forecast. It’s measurable in the Australian market right now, and the numbers are blunt.
The 1.2% problem
In March 2026, research firm Cited ran a controlled experiment on exactly this. It put 162 queries through ChatGPT, Perplexity and Gemini across Sydney, Melbourne and Perth, asking each engine to recommend agents the way a real seller would. Every query returned named agencies — the engines aren’t hedging, they’re recommending.
The catch is they barely agree with each other. In a focused Sydney run, only three of 254 surfaced agencies appeared on all three engines — a cross-engine agreement rate of roughly 1.2 per cent. Put another way: about 92 per cent of agencies that showed up at all had partial visibility at best, strong on one engine and invisible on the others. The overwhelming majority weren’t named by any engine, for any prompt.
There’s a second finding that should unsettle the big brands. Of the agencies visible across all three engines nationally, none were franchises. Decades of brand recognition built on signage and portal spend did not translate into AI visibility. The engines were rewarding something else.
For why this matters now and not next financial year: drawing on BrightLocal’s 2026 consumer survey, around 45 per cent of consumers already use AI for local service recommendations. This channel is already steering close to half the market.
Why your SEO playbook doesn’t transfer
The instinct is to treat this as “SEO, but for AI.” It isn’t, and assuming so is the most common mistake agencies are making.
Two things are different. First, there’s no single algorithm to satisfy. Each engine draws from a different primary source, and those sources barely overlap. ChatGPT leans on Google Maps and Google Business Profiles. Perplexity favours review platforms and directories. Gemini pulls from the broadest mix. Each has a gatekeeping source — and missing it shuts the door completely. Thin Google reviews make you invisible to ChatGPT. No credible third-party mentions make you invisible to Gemini. You’re not optimising one front door; you’re optimising three, and they want different things.
Second, the question changes the answer. Ask for the “best agent in Sydney” and you get one pool of names. Ask for a “first home buyer agent in Sydney” and the list rewrites itself entirely — specialists take over, with almost no overlap between the two sets. An agency engineered to win the generic prompt simply isn’t in the pool when a seller asks the specific one. Clear, corroborated specialisation is now a visibility strategy, not just a positioning one.
“Information gain” and the end of recycled content
The field has two names, used interchangeably: Answer Engine Optimisation (AEO) and Generative Engine Optimisation (GEO). Both describe the same goal — getting AI systems to cite and recommend you when someone asks a relevant question.
The part that’s shifted most, according to Elite Agent’s Samantha McLean, who’s been running agents through this in the publication’s AI Sprint, is what the engines reward: trustworthiness, and specifically information gain. The engines want sources that tell them something they don’t already know. Recycled market commentary, the same suburb profile every other agency publishes, auto-generated filler — these add nothing the model can’t produce itself, so they earn no visibility.
A March 2026 Google update sharpened this, targeting AI-generated content built to game rankings. The irony is pointed: agents pumping out automated content to “win at AI” are producing exactly the low-information-gain material the systems now discount. The agents winning are publishing things only they could know — real local sales insight, genuine campaign data, specific neighbourhood knowledge — and getting credible sources to reference them.
That last point is where most agents quietly give up. Only they could know. Where is that supposed to come from?
The asset hiding in plain sight: your database
Here’s the reframe that matters for our industry.
There are two games being played at once. The first is being named by the machine — the AEO/GEO long game above. It’s real, it compounds, and you should be in it. But it’s partly outside your control and it rewards patience over months.
The second game is faster, more controllable, and almost entirely yours: the intent already sitting in your database. While you fight to be the answer to a stranger’s AI query, there are sellers in your own list signalling they’re ready to move right now — and most agencies never see them, because their database is a static CRM export, not a live read on who’s engaged.
This is also the answer to the information-gain problem. The single richest source of insight only you could publish is your own database: your days-on-market by street, your auction results with context, the genuine read on a micro-market that a model can’t synthesise from public data. Activate that, and you solve two things at once — you produce the content the engines reward, and you surface the sellers you can call today.
That’s the entire premise behind how iRealty works. Rather than firing the same newsletter at a static list, a behaviour-based intelligence layer reads how each contact actually engages — which properties they click, which suburbs they keep returning to — and builds a live profile of intent that’s independent of whatever your CRM recorded four years ago. The output agents act on is a weekly seller-signal report: a shortlist of contacts showing active buying or selling behaviour, ready for a call. The machine deciding whether to name you is a long game. The seller in your database who clicked three appraisal-adjacent emails this week is a conversation you can have this afternoon.
Staying top-of-mind used to mean being remembered when a seller decides to sell. Now it means two things — being remembered by the seller and by the machine. Both reward the same discipline: consistent, relevant, branded contact, powered by data only you hold.
The playbook: how to become the answer
The inputs are concrete, and most are within your control. Here’s where to focus.
1. Own your Google Business Profile and your reviews. The highest-leverage move, because it’s ChatGPT’s primary source and feeds the others. Complete every field, keep it current, and build a steady flow of recent, genuine reviews. Low or no reviews is the fastest way to be invisible on the most-used engine.
2. Get named off your own site. Gemini and Perplexity weight directories, review sites and independent mentions. Your website calling you the suburb’s best agent is worth little; a recognised directory, a local news mention or an independent review platform saying it is worth a great deal. Audit where you are — and aren’t — listed.
3. Specialise out loud. Decide the specific prompts you want to win — first home buyers in a corridor, prestige in a pocket, investment stock in a postcode — and make that specialisation explicit and consistent everywhere your name appears. The generic “best agent” pool is crowded and franchise-skewed; the specific pools are winnable.
4. Publish for information gain, using data only you have. One piece a month built on your own numbers — days-on-market by street, auction results with context, an honest read on a micro-market — beats a content calendar of generic explainers. If an AI could have written it without you, it won’t help you. Your database is the well this comes from.
5. Audit your visibility like a seller would — then work the intent you already own. Ask ChatGPT, Perplexity and Gemini the exact questions your prospects ask, from a clean session with no login. The names that come back are your real competitive set in this channel. Then turn inward: the seller signals already in your database are revenue you can act on this week, regardless of what the engines say today.
The stakes
Traditional search isn’t disappearing tomorrow, and the agencies winning at AI visibility tend to be doing the fundamentals well anyway. But the direction is unambiguous. Consumers are already routing recommendation decisions through AI, the engines are already returning specific names, and the agencies in those answers are compounding an advantage that’s harder to overtake the longer it runs.
The uncomfortable truth in the data is that this lead is being built right now — mostly by agencies that decided to be early, not by the biggest brands or the biggest portal budgets. For an industry that has always run on being top-of-mind when a seller decides to sell, the question has simply changed venue. It used to be whether they remembered your name. Now it’s whether the machine does too — and whether you’re acting on the sellers who are already in your database, signalling, today.
Stop guessing who in your database is ready to sell. iRealty turns the contacts you already own into listing conversations — behaviour-based, branded, and automated, so you stay top-of-mind without the manual chase.
Book a Strategy Call → · +61 468 004 358
Send Smarter. Sell Listings Faster.
Sources and further reading: Cited’s March 2026 Australian AI-recommendation studies (wearecited.com); Elite Agent’s AI Sprint series with Samantha McLean; commentary from Tom Panos; BrightLocal 2026 consumer survey. Figures are current as at the dates reported by each source.