By We ran 324 queries across ChatGPT, Perplexity, and Google Gemini, asking each engine to recommend real estate agents in six Australian cities: Sydney, Melbourne, Perth, Brisbane, Adelaide, and the Gold Coast. Eighteen buyer-side prompts per city, three engines each. The result: more than 1,500 unique agencies named. Cross-engine agreement rate? Under 1 per cent in every city. Four of the six hit 0.0 per cent. Boutique and independent agencies accounted for 85 to 94 per cent of all recommendations. And the engines generated 912 new buyer prompts we never wrote. BrightLocal's 2026 survey puts AI usage for local service recommendations at 45 per cent. This is who those consumers find.
Data collected: March–April 2026.
- 324 queries across ChatGPT, Perplexity, and Gemini in 6 Australian cities
- Under 1% agreement — engines recommend almost entirely different agencies
- 85–94% boutique — independent agencies dominate across every city
- The question determines the answer — generic and need-specific prompts produce zero overlap
- 912 new prompts discovered through Gemini fan-out queries and Perplexity related questions
Four of six cities produced 0.0 per cent cross-engine agreement. Not low agreement. Zero. The engines read different sources, interpret buyer intent differently, and name entirely different agencies.
How Did We Test AI Real Estate Recommendations Across 6 Australian Cities?
We asked ChatGPT, Perplexity, and Google Gemini to recommend real estate agents across Sydney, Melbourne, Perth, Brisbane, Adelaide, and the Gold Coast. Eighteen buyer prompts per city, three engines each, 324 total queries. All run in clean-room conditions: no login, no personalisation, no conversation history.
The 18 prompts covered seven intent categories: generic recommendation, need-specific (first home buyers, investors, sellers, international buyers), comparative (franchise vs boutique, buyer's agent vs regular), suburb-specific, advisory, cost and fees, and city-specific niches. Every query was a fresh API call. No carry-over.
The engines did more than answer. Gemini decomposed our prompts into 372 internal fan-out queries across the six cities. Perplexity generated 540 related questions. Between them, 324 original queries produced 912 new buyer prompts we never wrote. Each fan-out is a content opportunity most agencies do not know exists.
| City | Prompts | Engines | Total Queries |
|---|---|---|---|
| Sydney | 18 | 3 | 54 |
| Melbourne | 18 | 3 | 54 |
| Perth | 18 | 3 | 54 |
| Brisbane | 18 | 3 | 54 |
| Adelaide | 18 | 3 | 54 |
| Gold Coast | 18 | 3 | 54 |
| Total | 108 | 3 | 324 |
Do ChatGPT, Perplexity, and Gemini Recommend the Same Real Estate Agents?
Almost never. Across all six cities, the cross-engine agreement rate was under 1 per cent. Four cities produced 0.0 per cent agreement: Sydney, Melbourne, Adelaide, and the Gold Coast. No single agency appeared on all three engines in those markets.
Perth came closest at 0.4 per cent, with 1 entity crossing all three engines. Brisbane recorded 0.6 per cent, but look at what actually crossed: 2 entities appeared on all three engines, and one of them was "local market knowledge." A concept. Not an agency.
Each engine reads different source material, interprets intent differently, produces a different list. Being visible on ChatGPT tells you nothing about Perplexity or Gemini.
| City | Agreement Rate | On All 3 Engines |
|---|---|---|
| Sydney | 0.0% | None |
| Melbourne | 0.0% | None |
| Perth | 0.4% | 1 entity |
| Brisbane | 0.6% | 2 entities |
| Adelaide | 0.0% | None |
| Gold Coast | 0.0% | None |
Interactive: Search for your agency
Data from our 324-query experiment across 6 Australian cities. See all free tools.
Does AI Recommend Franchise or Independent Real Estate Agencies?
Independents dominate. Boutique and independent agencies accounted for 85 to 94 per cent of all AI recommendations across six cities. Franchise names surfaced mainly in response to generic "best agent" prompts. The moment a buyer specified what they actually needed, franchise names dropped off.
| City | Franchise % | Boutique/Independent % |
|---|---|---|
| Sydney | 7.2% | 92.8% |
| Melbourne | 13.7% | 86.3% |
| Perth | 6.2% | 93.8% |
| Brisbane | 10.4% | 89.6% |
| Adelaide | 5.6% | 94.4% |
| Gold Coast | 15.3% | 84.7% |
In terms of franchise shares, Adelaide had the lowest at 5.6 per cent and Gold Coast had the highest at 15.3 per cent, driven by the presence of Ray White Alliance. Melbourne was slightly elevated at 13.7 per cent, which could reflect the auction culture. Jellis Craig kept showing up for auction-related prompts.
Throughout our six-city table, boutique and independent agencies dominated the recommendations with 84.7 to 94.4 per cent.
What Sources Do AI Engines Use to Recommend Real Estate Agents?
Each engine weighs sources differently, and those differences drive the recommendation gaps. ChatGPT leans on Google Maps and Google Business Profiles alongside agency websites. Perplexity gives significant weight to review and directory platforms as a secondary source behind agency sites. Gemini draws from the broadest source mix: agency sites, portals, directories, government sites, industry publications, forums.
Google was the most-cited domain in every city. Star ratings, review counts, street addresses pulled directly from Google Business Profiles. Narrowest source pool of any engine. If your Google Business Profile is incomplete, you are significantly less likely to appear in ChatGPT's recommendations.
Both review and directory platforms as well as non-agency sites such as RateMyAgent, WhichRealEstateAgent and Top10RealEstateAgent appeared consistently with citations in all six cities. In both the Brisbane and Gold Coast data, YouTube appeared.
We saw the most diversified mix of sources on Gemini. Portals such as realestate.com.au and everything from directories to government and education sites all influenced the citation picture. Gemini decomposes queries into 372 fan-outs across the six cities from 108 prompts, which makes for a large citation surface to cover for agencies who want to get cited.
Total unique domains cited across all cities and engines: over 200.
Does the Question You Ask Change Which Agent AI Recommends?
Completely. In every city, the agencies recommended for generic "best agent" prompts had zero overlap with those recommended for need-specific prompts like "first home buyer agent" or "investment property specialist." The question rewrites the answer.
We tested seven intent categories: generic, need-specific, comparative, suburb-specific, advisory, cost and fees, and city-specific niches. Generic prompts returned franchise names alongside high-review generalists. Need-specific prompts returned specialists. Suburb-specific prompts returned local boutiques with suburb-level content on their websites.
Then the city-specific niches. Flood-prone specialists in Brisbane. Heritage home and wine country agents in Adelaide. Waterfront and holiday rental specialists on the Gold Coast. Mining town agents in Perth. Auction specialists in Melbourne. Off-plan corridors in Sydney. Each niche produced an entirely separate pool of recommended agencies, invisible to every other prompt type we ran.
The generic prompt and the specific prompt produce zero shared results. An agency optimised for "best agent in [city]" is invisible to buyers asking about specific needs. This held across all 324 queries.
An agency that thinks it has AI visibility because it appears for "best real estate agent Brisbane" may be completely absent when a first home buyer asks about flood-prone areas in Rocklea. Different recommendation pools. Different source material. Different agencies entirely. This is a content strategy question, not a brand awareness question.
What Do AI-Visible Real Estate Agencies Have in Common?
Across all six cities, the agencies visible on multiple engines share clear traits. Review platform presence, specialisation clarity, third-party mentions, Google Business Profile completeness, suburb-level content, and individual agent profiles on directory platforms.
Across the board, engines preferred agencies that clearly stated what clientele and location they serve, and consistently appeared for niche-specific prompts. Generalist "full-service" framing produced zero citations or mentions.
RateMyAgent, WhichRealEstateAgent, LocalAgentFinder and Top10RealEstateAgent were some of the recurring mentions on Perplexity and Gemini, even though these are not even real estate agencies.
On Gemini, third-party mentions carried a lot of weight. Forum discussions, awards and news articles mentioning real estate agencies promoted citation frequency.
Agencies with suburb-specific pages appeared for suburb-specific prompts. Agencies without suburb content were invisible to those same prompts. No exceptions across 324 queries in six cities.
| If You... | AI Sees You As... | What to Do |
|---|---|---|
| Have no review platform profiles | Harder for Perplexity to find | Claim and verify profiles on RateMyAgent, WhichRealEstateAgent, LocalAgentFinder |
| Use generic "full-service" website copy | Invisible to need-specific prompts | State your specialisation clearly on your homepage |
| Have no suburb-level content | Invisible to location queries | Create pages for each suburb you operate in |
| Have few Google reviews | Harder for ChatGPT to find | Build your Google Business Profile review count and completeness |
| Have no third-party mentions | Fewer signals for Gemini to find | Get mentioned in industry publications, forums, and award lists |
Calculator: What is AI invisibility costing you?
Estimate based on market data and AI adoption rates. See all free tools.
What Does This Mean for Australian Real Estate Agents?
AI recommendations are happening now. BrightLocal's 2026 survey found 45 per cent of consumers use AI for local service recommendations. If your agency is invisible across ChatGPT, Perplexity, and Gemini, you are invisible to nearly half the market looking for an agent.
Throughout our research we noticed three recurring mistakes:
Agencies that clearly optimised for one engine, such as Google via SEO, might get cited on one engine but are completely invisible on two others. This confirms that each engine has its own retrieval and weighting system.
Those who optimise for generic prompts and ignore need-specific queries get completely left out of informational queries.
Very few agencies actually took the time to check what AI engines say about their business.
We built a free tool to answer that question. Check your agency's AI visibility now (30 seconds). Or request your free AI Visibility Report for the full picture across all three engines.
Each city produced different AI visibility patterns. The full data lives in the deep dives:
- Sydney — 0.0% cross-engine agreement. Off-plan and international buyer prompts surfaced agencies invisible to every other query type.
- Melbourne — Auction-related prompts created an entirely separate recommendation pool. Jellis Craig was the only agency to appear across five prompt categories.
- Perth — Smart Realty was cited 15 times by Gemini alone. REIWA appeared as a trust signal on all three engines, something no other state industry body matched.
- Brisbane (deep dive coming soon) — Asking about flood-prone areas in Rocklea produced completely different agents than asking for "best agent Brisbane." Place Estate Agents led Perplexity with 3 mentions.
- Adelaide (deep dive coming soon) — 94.4% boutique, the highest independent share of any city. Heritage home and Adelaide Hills wine country queries opened niche-specific recommendation pools.
- Gold Coast (deep dive coming soon) — 15.3% franchise share, the highest in the experiment. Lucy Cole Prestige dominated Perplexity with 6 mentions. Waterfront and Airbnb prompts split the market again.
Frequently Asked Questions
Do ChatGPT, Perplexity, and Gemini recommend the same real estate agents in Australia?
Very rarely. Across the 324 queries in six cities we found sub 1% agreement between engines.
Does AI recommend franchise or independent real estate agencies in Australia?
Independent and boutique agencies dominate 85%-94% of the recommendations across all cities in our dataset. Franchise agencies mainly landed citations for generic "best agent" prompts, never on informational queries.
How many queries did you test?
18 buyer prompts across the six biggest cities in Australia on 3 engines. 324 total.
Which Australian real estate agency has the best AI visibility?
Each city and query type produced a different citation landscape with different visibility leaders. Jellis Craig in Melbourne and Smart Realty in Perth are honourable mentions for their respective markets. Read our city-specific reports for Sydney, Melbourne, and Perth for the full insight, with Brisbane, Adelaide, and Gold Coast coming soon.
How can I check my real estate agency's AI visibility?
Head to wearecited.com/check and run an instant check for free.
Sources
- Cited Research, Australian AI recommendation experiment, March–April 2026. 324 queries across ChatGPT, Perplexity, and Gemini in Sydney, Melbourne, Perth, Brisbane, Adelaide, and the Gold Coast. Raw data collected via API in clean-room conditions.
- BrightLocal, 2026 Local Consumer Review Survey. Referenced for the 45 per cent AI usage statistic.
- RateMyAgent, WhichRealEstateAgent, Top10RealEstateAgent, LocalAgentFinder — Australian real estate agent review and comparison platforms cited by AI engines across all six cities.