Somewhere in Australia tonight, a person will ask ChatGPT to recommend a business exactly like yours: a broker, a builder, a boutique, a firm. ChatGPT will answer confidently, name two or three options, and that person will probably contact one of them.
The question that should keep you up is not whether this happens, but whose names are in the answer.
There's no form to submit and no ad product to buy your way in. But recommendations aren't random either.
They're the output of signals you can influence, and most Australian businesses haven't started, which is precisely the opportunity.
How ChatGPT decides who to name
Two mechanisms matter. First, training knowledge: businesses written about widely and consistently over time become part of what the model simply knows.
Second, and more actionable, live retrieval: for current, local, commercial questions, ChatGPT searches the web and synthesises an answer from what it finds in that moment. Retrieval is where the near-term game is played, because it means today's web presence shapes tomorrow's answer.
What the model is doing, in effect, is looking for converging evidence: multiple independent sources agreeing that a specific business is legitimate, relevant and well-regarded. Your job is to make that evidence exist and agree.
Third-party proof beats self-praise
A language model trying not to give bad advice treats your website's claims the way a careful buyer would: as claims. What moves it is corroboration it didn't get from you: review volume and recency on Google, Australian surfaces like ProductReview.com.au, mentions in local media and industry publications, directory and association listings.
If the independent record about your business is thin, the safest thing an engine can do is recommend someone with a thicker one. Building that record (steadily asking for reviews, earning coverage, claiming the listings that matter in your industry) is slow, compounding work, and it's the heart of getting recommended.
Be the same business everywhere
Entity consistency sounds technical but it's simple: every mention of your business, everywhere, should agree. Same name (not three variants of it), same address and phone, same description of what you do and where you do it, across your site, Google Business Profile, directories, socials and third-party mentions.
When the records conflict, the model can't be sure the glowing reviews and the website belong to the same entity, and uncertainty gets resolved by leaving you out. An afternoon spent reconciling your listings is some of the highest-return GEO work available.
Make yourself easy to cite
Once ChatGPT can find and trust you, give it something quotable. State plainly who you serve, where, at what price point, and what makes you different: as facts, not slogans.
"Award-winning service" gives a model nothing; "a fixed-fee conveyancing service covering greater Adelaide, most settlements completed in under six weeks" gives it a sentence it can practically lift.
Answer the comparison questions buyers actually ask, including the honest ones about who you're not right for. Candour reads as trustworthiness to buyers and models alike.
A quick checklist of the facts worth stating in plain text on your core pages:
- Exactly what you do and who it's for, in one sentence a model could quote verbatim.
- Where you operate, named specifically: suburbs, regions or "Australia-wide", not "wherever you are".
- Prices, fees or honest ranges, with the factors that move them.
- Proof points that are checkable: years operating, volumes handled, accreditations, response times.
- When the page was last updated: stale pages get discounted, and quietly.
Turning a mention into a customer
A ChatGPT recommendation rarely arrives as a tracked referral click. It arrives sideways: the person Googles your name afterwards to verify what the AI said, calls directly, or walks in saying "ChatGPT suggested you."
If you're only watching website referral sources, most of this demand is invisible to you: it shows up as branded search, direct traffic and unexplained phone calls. Controlled research has started to quantify the first of those effects.
+4.3 pts
lift in branded search when AI recommends a brand to unengaged users
So instrument for it: watch branded search volume, ask every new enquiry how they found you, and make sure the pages an AI-referred visitor lands on confirm the recommendation and make contact effortless. And hold the honest line: a mention is visibility, not revenue.
Whether it converts depends on your offer and follow-up, which is why we measure AI visibility through to actual leads and enquiries rather than declaring victory at the citation. If you'd like to know whether ChatGPT mentions you today, run a visibility check with us.
No. There's no placement product for organic recommendations, and anyone selling guaranteed inclusion is selling something they don't control. The influenceable inputs are your content, your consistency and your third-party record.
Ask it: fresh sessions, the questions your customers would ask, by suburb and service. Ask directly what it knows about your business and note anything wrong or outdated. Repeat monthly, because answers shift as the underlying record changes.
Fix the sources rather than fighting the symptom. Wrong details usually trace to stale listings, old pages or inconsistent records. Correct them at the origin and answers tend to follow as they're re-retrieved.
The principles carry across (retrieval, corroboration, consistency), though each engine weights sources differently, and Google's surfaces lean harder on your Google Business Profile. Work that makes you citable generally travels well across engines.