AI SEO and answer engine optimisation
The technical work that makes a website eligible to be quoted by AI assistants. We do it as engineering, on our own builds and on sites we did not build.
Book a callThe 4-step system we use to get
Australian brands cited by AI
Map which brands AI assistants already recommend in your category, measure how often you appear, work out what the cited pages have in common, then track the same measures over time.
- Step 01
Map your AI competitors
The brands ChatGPT, Gemini and Perplexity already recommend in your category are rarely the same as your Google rivals. We benchmark your real competitive set inside AI answers.
- Step 02
Measure your AI mentions
Visibility Score, mention frequency and cited pages across ChatGPT, Gemini, Perplexity and Google AI Overviews. You vs your top two rivals, in one snapshot, tracked monthly.
- Step 03
Decode the “why”
AI doesn't cite brands at random. Each one gets paired with a driver: price, speed, trust, locality. We find the driver you can credibly own and engineer your site around it.
- Step 04
Build the benchmark loop
Schema, llms.txt, GPTBot allow, FAQ structure, citations. The top 10% of AU sites have all four. The median has none. We close the gap in days, then track it monthly.
Our research: State of AI Search Australia 2026 and AI visibility audit benchmarks.
What this involves
Crawler access
There are three separate permissions here and they are commonly confused. OpenAI documents OAI-SearchBot and GPTBot as independent controls: a site can allow OAI-SearchBot to appear in search results while disallowing GPTBot so its content is not used for model training. Perplexity asks site owners to allow PerplexityBot for a site to appear in its results. Anthropic's ClaudeBot is a training crawler that honours robots.txt directives.
Google's Google-Extended is not a crawler at all. It has no separate user agent string and exists only as a robots.txt token used in a control capacity. We audit which of these are currently allowed, then set them to what you intend. That is often not what the site is doing today.
Content in the HTML response
Google processes JavaScript in three phases and queues pages for both crawling and rendering, so rendering happens later rather than immediately. Most AI crawlers do not render pages at all. They read the HTML the server returns. A site that assembles its content in the browser serves those crawlers an effectively empty document, however good the content is once it appears. On a client-rendered application, moving that content into the served HTML is usually the largest single change.
Entity structure
Organization, LocalBusiness and Service markup, one consistent identifier across pages, and a business name, address and phone number that are identical everywhere they appear. This is what lets an engine decide that two mentions refer to the same business. Details that differ across a site, a directory listing and a Google Business Profile give it reason to treat them as three separate organisations.
Passages that survive quotation
A quotable passage names its subject and holds its meaning without the paragraph above it. Two sentences that could sit on the same page:
Not quotable
"We offer this in three tiers, depending on your requirements."
Quotable
"GetQuickSites builds web applications for Australian businesses from its Melbourne studio."
Measurement
We record what was implemented and re-check the site's eligibility signals. We do not claim to measure what an assistant will say, because that is not ours to control. What we report is what changed on the site and which signals it now meets.
When you need this, and when you do not
Worth doing
- You sell something people ask an assistant to recommend.
- You already rank reasonably on Google but do not appear in AI answers.
- You are commissioning a build anyway, where this costs very little on top of the build.
Not worth doing yet
Your site does not convert. Sending more people to a page that does not work multiplies an existing problem. Fix that first. We would tell you to spend the money on a website redesign instead.
You have no site, or a broken one. This is a layer on top of a working site, not a substitute for one.
Your buyers do not use assistants for this decision. Emergency trades, walk-in retail and tender-based work are three cases where local search and a Google Business Profile will do more for the same money.
You want a guarantee. Nobody can give you one. If that is the requirement, do not buy this from anyone.
How to evaluate a provider, including us
Six questions worth asking anyone selling this work, and what a good answer sounds like.
1. Show me the served HTML, not the rendered page.
Fetch the URL and read what comes back before any JavaScript runs. If the content only appears in a browser, no amount of strategy fixes it. Anyone selling this work should be able to do it in front of you.
2. Which crawlers are allowed, and was that decided on purpose?
Search access and training access are separate choices. A provider who cannot tell them apart has not read the documentation.
3. What will you guarantee?
The correct answer is nothing about assistant output. A provider guaranteeing citations is either misunderstanding these systems or misrepresenting them.
4. Is there a proprietary technique?
Google states there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimisations necessary. A provider selling a secret method is selling something that does not exist.
5. What did you actually change?
You should receive a list of files, markup and directives. A dashboard of metrics that cannot be traced back to a change is not a report.
6. What happens if it does not work?
Results vary by market and by the authority a site already has. Ask what the provider does next, and whether you keep the work if you leave.
What we can and cannot claim
AI systems choose their own sources and change without notice. No provider controls what an assistant says about your business.
What is controllable is eligibility: whether crawlers can reach you, whether the content is in the response, whether the business is identifiable as one entity, and whether passages can be quoted on their own. Google's own position is that appearing in its AI features requires a page to be indexed and eligible to be shown in Google Search with a snippet, and nothing further. Results vary by market and existing authority.
What the gap looks like in Australia
From our crawl of 1,200 Australian SMB websites, comparing the top 10% for AI visibility against the median. The method and the other five signals are in the AI visibility audit benchmarks.
| Signal | Top 10% | Median |
|---|---|---|
| Organization JSON-LD | 100% | 18% |
| FAQPage schema | 94% | 9% |
| llms.txt at root | 88% | 4% |
| GPTBot allowed in robots.txt | 100% | 39% |
| Server-rendered FAQ content | 96% | 44% |
Each of these signals is set by implementation rather than by spend. The gap between the top decile and the median records which sites have had the work done.
Website builds for Australian businesses
A complete digital presence designed to convert visitors into customers, delivered in 48 hours.
Custom-engineered builds, delivered in 48 hours.
Read the data, or check your own site
AI visibility audit benchmarks
Ten signals measured across 1,200 Australian SMB websites, with the method used to collect them.
Read the benchmarksCheck your own site
A short check of the eligibility signals described on this page, run against your site.
Run the checkTerms used on this page
- Answer engine optimisation (AEO)
- Making a page eligible to be quoted inside an AI-generated answer, rather than only ranked in a list of results.
- Citation
- A named, linked reference to your site inside an assistant's answer.
- Entity
- The machine-readable identity of a business, held together by consistent structured data and consistent contact details across every page.
- llms.txt
- A proposed plain-text file at the domain root that gives AI systems a factual summary of a site. It is a convention, not a standard. llmstxt.org
- Server-rendered
- Content present in the HTML a server returns, before any JavaScript runs in the browser.
- Structured data
- Machine-readable markup that describes what a page is about, using the schema.org vocabulary. Google's introduction to structured data
Frequently asked
What is AI SEO and how is it different from traditional SEO?
AI SEO, also called answer engine optimisation or generative engine optimisation, is the work that makes a site legible and citable to AI assistants such as ChatGPT, Gemini and Perplexity. Traditional SEO targets a position in a ranked list. AI SEO targets being quoted and attributed inside a generated answer, where there is no second page and only a handful of sources are named. The technical foundation is shared, so the two are not alternatives. Google states there are no special optimisations for its AI features beyond ordinary search eligibility.
How do AI answer engines decide what to cite?
They need to reach the content, parse it, and attribute it confidently. In practice that means four things: crawler access for AI agents, which is controlled separately from Googlebot; content present in the served HTML rather than only after JavaScript runs, since most AI crawlers do not render pages; structured data that identifies the business as a specific entity; and self-contained passages that still make sense when quoted without the paragraph above them. A page can rank well on Google and fail every one of these.
Can you guarantee my business will be recommended by ChatGPT?
No, and treat any provider who does as a warning sign. AI systems choose their own sources, change without notice, and give no vendor control over their output. What can be controlled is eligibility: whether crawlers can reach the site, whether the content is in the HTML response, whether the business is identifiable as one entity, and whether passages can be quoted on their own. We implement those, then report exactly what changed. Results vary by market and by the authority a site already has.
Does AI SEO work alongside conventional SEO?
Yes, and they share most of their foundation. Crawlability, page speed, clean semantic structure and valid structured data serve conventional search and AI answer engines equally, which is why the work is rarely wasted. Google's own guidance is that appearing in its AI features requires a page to be indexed and eligible for a snippet, with no additional technical requirements. AI SEO adds three things on top: AI crawler directives, entity consolidation across the site, and answer-shaped passages built for extraction.
How much does AI SEO cost?
AI SEO engagements are custom-quoted after a free consultation. Scope, integrations and timeline determine the figure. Site size, current technical condition and how much structure and content work is needed all change the scope, so we audit first and quote against defined work rather than publishing a package price. An audit tells you which of the four eligibility signals your site already meets, and most sites meet at least one. Where a site is already well built, the work is small. Where content is rendered entirely in the browser, it is not.
Do I need an llms.txt file?
llms.txt is a proposed convention for a plain-text file at your domain root that gives AI systems a factual summary of the site. It is not a standard, and no major AI engine has publicly committed to reading it. Publishing one is cheap and it forces a useful exercise: writing down what the business does in plain language. Treat it as a low-cost addition rather than the thing that earns citations. Crawler access, server-rendered content and entity markup matter more.
Will blocking GPTBot stop ChatGPT from recommending my business?
Not necessarily, because they are separate controls. OpenAI documents GPTBot and OAI-SearchBot as independent robots.txt tokens: GPTBot governs whether your content may be used to train models, and OAI-SearchBot governs whether your site can appear in ChatGPT search results. You can allow one and disallow the other. Many sites block both without realising the second decision was made for them. To appear in ChatGPT search while keeping content out of training, allow OAI-SearchBot and disallow GPTBot.
Does AI SEO work if my content is rendered by JavaScript?
Usually not well. Google documents that it processes JavaScript in three phases and queues pages for both crawling and rendering, so rendering happens later rather than immediately. Most AI crawlers do not render JavaScript at all; they read the HTML the server returns. If the content is assembled in the browser, those crawlers receive an effectively empty document. The fix is to serve the content in the HTML response. On a client-rendered application this is usually the largest single change.
How is this different from what an SEO agency already does?
Most of it overlaps. Crawlability, page speed, clean semantic structure and valid structured data serve conventional search and AI answer engines equally. Three things are specific: explicit robots.txt directives for AI crawlers, which are separate from Googlebot; entity consolidation, so an engine can tell that scattered mentions refer to one organisation; and passages written to hold their meaning when quoted without the surrounding page. If your current provider does those three, a separate engagement is not needed.
Ready to look at your own site?
Melbourne studio, 48-hour delivery
GetQuickSites, Melbourne software & web development studio.