The shift from search to answers
When someone Googles "best accountant in Parramatta," they get ten blue links. When they ask ChatGPT the same question, they get one answer — maybe with a few names cited.
There is no page two. There is no "scroll down." The AI either mentions your business or it does not. This is the new reality of search, and it is already affecting how Australian businesses get found.
Generative Engine Optimization (GEO) is the emerging discipline of making your business visible to AI systems. It is not a replacement for SEO. It is the next layer on top of it.
May 2026: Google redesigns Search around AI
At Google I/O 2026, Google announced the biggest change to Search in 25 years. The new AI Mode replaces traditional results with AI-generated summaries. Even more significant: information agents now monitor topics in the background for users — reading, filtering, and summarising the web continuously, not just when someone types a query.
For businesses, this means the shift from “rank well once” to “be citable all the time” is accelerating. Even if your content helps shape an AI answer, the user may experience it as Google’s recommendation — without ever visiting your website. The GEO strategies below are how you stay visible in this new reality.
What about llms.txt and llms-full.txt?
You may have heard of llms.txt — a proposed standard (created by Jeremy Howard in September 2024) where websites publish a machine-readable Markdown file at their root to help AI systems understand their content. There are actually two companion files:
- llms.txt — a curated index with links and one-line descriptions. Think of it as a table of contents for AI agents, telling them what pages exist and where to find them.
- llms-full.txt — the entire site’s content concatenated into a single Markdown file. Agents can ingest everything in one request instead of crawling page by page.
Who actually uses it?
Adoption reached roughly 10% of websites by mid-2026 (based on a SE Ranking study of 300,000 domains). Major tech companies like Stripe, Cloudflare, Anthropic, Cursor, Vercel, and Supabase all publish one. But where things get interesting is who is reading these files.
The real consumers: AI coding agents, not search
Out of 515 million LLM bot traffic events monitored over 90 days, only 408 targeted llms.txt directly. The major AI search crawlers (GPTBot, ClaudeBot, PerplexityBot) almost never fetch it.
The tools that do actively consume llms.txt and llms-full.txt are IDE agents: Cursor, Claude Code, GitHub Copilot, Windsurf, Cline, and Aider. When a developer points one of these tools at a documentation site, the agent looks for /llms.txt first. Companies serving Markdown instead of HTML report up to 10x token reductions — meaning faster, cheaper, more accurate agent behaviour.
Does llms.txt improve AI search citations?
No. A SE Ranking study found that an XGBoost model trained on AI citation data actually performed better when the llms.txt variable was removed — the file added noise rather than predictive signal. Having the file does not measurably improve your odds of being cited by ChatGPT, Claude, Gemini, or Perplexity.
Google’s John Mueller compared llms.txt to the deprecated keywords meta tag — vendor-controlled metadata that crawlers can already get from the page itself. Google has explicitly stated it does not support llms.txt and has no plans to.
That said, Anthropic and Perplexity have publicly confirmed some support for llms.txt in their retrieval workflows. OpenAI and Mistral show observable response patterns without explicit public commitment. The picture is mixed, but the evidence is clear: llms.txt is not a search ranking factor.
So should your business create one?
The practical recommendation
Developer-facing companies (SaaS, APIs, open-source projects, documentation-heavy products): yes, publish both llms.txt and llms-full.txt. Your buyers and their AI coding agents are already looking for them. The cost is a half-day of work and the upside is real.
Local businesses, service providers, retailers: your time is far better spent on the GEO fundamentals below — structured data, entity clarity, and topical authority. These are the techniques with proven, measurable impact on AI citations. You can always add llms.txt later if AI providers begin to require it.
Think of llms.txt as Business-to-Agent (B2A) infrastructure rather than a search optimisation tactic. It helps AI agents that are already looking for you work more efficiently — it does not help you get discovered in the first place.
SEO vs GEO: what changes?
The skills overlap, but the optimisation targets are fundamentally different:
| Traditional SEO | GEO | |
|---|---|---|
| Goal | Rank on search result pages | Get cited in AI-generated answers |
| Key signal | Backlinks and keyword relevance | Entity clarity and structured data |
| Content | Keyword-optimised pages | Authoritative, factual, citable statements |
| Competition | 10 organic spots per page | Often 1–3 mentions per answer |
| Measurement | Rankings, impressions, CTR | Citation audits across AI platforms |
| Timeline | Weeks to months | Training data lag (months) + real-time retrieval |
7 GEO strategies that work today
These are practical, proven techniques — not speculation. Each one improves the chances that AI models will find, understand, and cite your business.
Get your structured data right
Schema.org markup (LocalBusiness, Organization, Product, Course, FAQ) gives AI models machine-readable facts about your business. Name, location, services, pricing, opening hours — all in a format AI can parse directly. This is the single highest-impact GEO action for most businesses.
Build entity clarity
AI models resolve businesses as "entities" — a consistent name, description, and set of attributes across the web. Use the exact same business name, address, and description on your website, Google Business Profile, LinkedIn, industry directories, and social profiles. Inconsistency confuses AI models just like it confuses search engines.
Write citable statements
AI models look for clear, factual, self-contained sentences they can quote. "We run small-group AI workshops in Sydney for business teams of up to 12 people" is citable. "We provide world-class innovative solutions leveraging cutting-edge AI" is not. Be specific. Be concrete. State facts, not claims.
Develop topical authority
Publish depth, not volume. A business that publishes five genuinely useful articles about AI training for Australian businesses will outperform one that publishes fifty shallow posts on random topics. AI models weight topical consistency and expertise signals heavily.
Earn third-party mentions
AI training data includes news sites, directories, review platforms, and industry publications. When other credible sources mention your business in context, AI models learn that association. Guest articles, industry awards, case studies in trade publications, and genuine reviews all contribute.
Keep your HTML clean and accessible
Semantic HTML, proper heading hierarchy, alt text on images, and fast page loads are not just good web practice — they are how AI crawlers and retrieval systems extract content. A well-structured page is easier for both humans and AI to understand.
Audit your AI visibility regularly
Ask ChatGPT, Claude, Gemini, and Perplexity questions your customers would ask. Note whether your business appears. Track which competitors get cited instead. This is not a one-time exercise — AI models update their knowledge continuously.
A practical GEO checklist for your business
Start here. Each item takes less than an hour and has measurable impact:
- Add Schema.org LocalBusiness or Organization markup to your homepage
- Ensure your business name, address, and description are identical across all profiles
- Write a clear, factual "About" section with specific services, locations, and differentiators
- Add FAQPage schema to your most common customer questions
- Audit your business across ChatGPT, Claude, and Gemini — note gaps
- Publish one in-depth article on your core expertise (not a sales pitch — genuine expertise)
- Claim and complete your Google Business Profile with accurate categories
- Ensure your site has proper heading hierarchy (h1 → h2 → h3, not random)
This is not a set-and-forget exercise
GEO is an ongoing discipline. AI models retrain, retrieval systems update, and competitor content evolves. The businesses that treat GEO as a regular practice — not a one-time project — will maintain their visibility advantage.
Why we teach this in our workshops
GEO is one of the practical skills we cover in our AI workshops. We don't just explain the theory — participants implement structured data, run live citation audits, and build an AI visibility plan for their own business during the session.
Whether you are a sole trader wanting to show up when someone asks "best [your service] near me" or a team lead preparing your marketing department for AI-first search, these techniques are immediately actionable.