Research
Studies of how AI-assisted search systems name, recommend and cite sources when the prompt has a Perth or Western Australian context. Each study runs on a fixed prompt set, across named platforms, with repeated runs and a versioned methodology.
Published
We asked four AI systems 200 commercial questions about 20 Perth industries, three times over three days, and recorded 2,400 AI answers. Most of the businesses they recommended were not the ones ranking at the top of Google Maps, and most recommendations changed between runs.
Updated 2026-09-13
Published
We took every Google AI Overview from the Perth AI Search Study 2026, recorded all 2,845 sources it cited, and compared each one to the organic results Google returned on the same page for the same search. Most cited pages were already ranking. Nearly half of everything handed to a business website was the homepage.
Updated 2026-09-21
Listed so the programme's scope is public. A planned study has no findings and no page of its own. It becomes a report only when its observation file exists under the published methodology.
Planned · no observations yet
How ChatGPT answers Perth recommendation and comparison prompts: which businesses it names, in what order, what it cites, and how much the answer changes between runs and session profiles. Planned; no observations collected.
Results will be published under the methodology once observations exist.
Planned · no observations yet
How the Gemini app answers Perth recommendation prompts, which businesses and sources it names, and how that compares with Google AI Overviews on the same prompt set. Planned; no observations collected.
Results will be published under the methodology once observations exist.
Planned · no observations yet
Which domains AI-assisted search systems cite when answering Perth prompts, classified by source type: owned sites, directories, review platforms, publishers, platform-owned properties and user-generated content. Planned as a dedicated study. The Perth AI Search Study 2026 already publishes the raw cited-domain counts and the per-platform citation mix for 200 commercial questions; this study adds the source-class classification across a wider prompt set.
Results will be published under the methodology once observations exist.
Planned · no observations yet
Which sources Perplexity cites for Perth prompts, by source class, and how its citations compare with other platforms on the same prompt set. Planned; no observations collected.
Results will be published under the methodology once observations exist.
Planned · no observations yet
A repeated-round measurement of which Perth businesses AI-assisted search systems name and recommend across a fixed prompt set, by industry and platform. Planned as a repeated index. The Perth AI Search Study 2026 is its first round; the index begins when a second round on the same prompt set allows change to be reported.
Results will be published under the methodology once observations exist.
Planned · no observations yet
Which businesses AI-assisted search systems name and recommend when someone asks for an AI SEO agency or consultant in Perth, which domains those answers cite, and how much the answer changes between runs and platforms.
Results will be published under the methodology once observations exist.
Each prompt run on each platform produces one observation record. The record structure is fixed and published as a JSON Schema.
| Recorded | What it means |
|---|---|
| Brand mentions | Every business or brand named in an answer, whether or not it was recommended. |
| Recommendations | Brands the answer presented as a shortlist or recommendation, in the order given. |
| Cited domains | The registrable domains behind every citation, classified by source type. |
| Cited URLs | The exact pages cited, so recurring source pages can be identified. |
| Source overlap | Which sources recur across prompts, platforms and runs. |
| Platform differences | The same prompt set run across each platform in scope, compared side by side. |
| Query and prompt classes | Informational, commercial, comparison, recommendation, local, brand, competitor and trust-validation prompts, kept separate. |
| Change over time | Repeated rounds on a fixed prompt set, so movement can be measured rather than assumed. |