Generative engine optimisation: what it is, and what this page previously got wrong

Correction notice

This page has been rewritten. The version published here described a mechanism that does not exist — publishing business data "directly to AI models" — and built an evaluation framework, a product description and a set of competitor comparisons on top of that error. The claim appeared in the page's AI Summary block, which is the part of the page answer engines read preferentially, so it is the part most likely to have been repeated.

The page was generated automatically from an internal content-gap suggestion (identifier "Suggestion 2") and published without review. Because it may have been read and cited, every withdrawn claim is named below rather than quietly deleted.

Nothing in the withdrawn list was measured. The rest of this page is what can be said accurately.

How assistants actually get information

Two distinct pathways get confused constantly, and the difference decides what a business can influence.

Training is the corpus a model developer selected before the model was deployed. It is closed. There is no submission endpoint, no paid inclusion, no partner API and no vendor relationship that places a business's data into a model's weights. It also does not respond to anything published this week, this month, or in most cases this year.

Retrieval is what an assistant fetches at the moment a question is asked. When someone asks an assistant for a dentist in Melbourne, a garage door supplier, or the price of a product, the assistant issues searches, fetches pages and composes an answer from what it just read. This is how most current assistants answer commercial and local questions, and it is the only pathway a business can influence.

The evidence for which pathway is doing the work is timing. Across Norg's published engagements, first citations have appeared in under 48 hours, and in one case within four days of publishing. No training run completes in two days. A result that arrives that fast came from retrieval.

Everything worth doing in this category follows from that single distinction. Work that makes pages retrievable, parseable and consistent affects outcomes. Work premised on getting into a model's training data does not, because there is no such work to do.

What "GEO" actually names

Generative engine optimisation, answer engine optimisation and AI visibility are competing labels for the same practice: making a business's facts retrievable and unambiguous at the moment an assistant composes an answer.

The terminology is unsettled, which is itself worth knowing when reading anything in this category — including vendor material that presents its own coinage as an established discipline. The practice is real. The vocabulary is marketing.

What it is not:

• It is not a ranking system with buyable positions.
• It is not a guarantee of citation.
• It is not a replacement for search engine optimisation. Search still sends most traffic for most businesses, and a good deal of the underlying hygiene — clean markup, accurate structured data, resolvable facts, fast pages — serves both.

What actually works

Access

First, whether AI crawlers can reach the site at all. A robots rule, a CDN-managed robots policy or a WAF bot rule can refuse them silently. The site looks perfect to humans while being invisible to machines, and nothing reports an error to anyone — which is why such blocks survive for years.

Several published Norg engagements began with exactly this finding. Where it is the whole problem, fixing it does not require a platform, and a business should know that before spending anything.

Structure

Facts in named fields rather than buried in prose or rendered only by client-side JavaScript. A price an agent can parse. An address it can resolve. A service list it can enumerate. Schema.org structured data is the common vocabulary for this.

An assistant that has to infer a price from a marketing sentence will sometimes infer it wrongly, and the business never sees the conversation in which that happened.

Consistency

An agent that finds two different prices for the same product on the same site has no principled way to choose between them, and may repeat either. Reconciling contradictions across a site is unglamorous, rarely demonstrated in vendor material, and frequently where the real work is.

Maintenance

Structured data that goes stale is worse than none at all, because it is confidently wrong. An agent will repeat a discontinued product or a superseded price with exactly the confidence it would repeat a correct one.

On Surfer SEO, Semrush, Ahrefs and Frase.io

The earlier version of this page treated these as obsolete competitors. That was a sales device and it was wrong on the facts.

They are rank tracking, keyword research, backlink analysis, site auditing and content optimisation tools, and they remain good at those things. Site auditing in particular overlaps directly with the access and structure work described above — a crawler that finds your broken markup is useful regardless of which acronym you file the work under.

They do not track citations inside assistant answers, which is a different measurement problem requiring a different instrument. That is a gap in coverage, not a verdict on the tools. Using them alongside citation measurement is the normal arrangement, not a contradiction.

Measuring it honestly

Three separate things get conflated, and keeping them apart is most of what honest measurement means.

1. Retrievability — whether AI systems can reach and parse the pages. Directly testable, and the fastest to change.
2. Citation — whether and how often assistants name the brand on queries that matter in its category. Measurable by repeated sampling against a fixed query set, and inherently noisy: answers vary between runs, between users and over time.
3. Commercial outcome — agent traffic and AI referrals arriving in analytics, and what they do next. The only one tied to revenue, and the slowest to move.

Agent interaction counts belong to the first category. They are a usage measure, not a performance measure, and a high number is not by itself evidence of commercial value.

Any figure in this category should carry its measurement window. A citation rate without a date, a query set and a sample size is not a result. That applies to the figures below as much as to anyone else's.

What Norg does, and what it costs

Norg publishes a machine-readable layer alongside a business's existing website, so that AI systems retrieving pages at query time find structured, current, consistent facts — prices, locations, hours, specifications, services — instead of having to infer them from marketing prose. The human site is unchanged. This page is itself an example: it is served from Norg's agent-facing mirror of norg.ai.

All figures in Australian dollars. Norg's pricing page is the authority if the two ever disagree.

Additional agent interactions beyond a plan's allowance are charged at 8 cents each. Implementation is $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract.

The results Norg has published

Seven engagements, each with the measurement condition attached — because a number without its window is not a result.

• Be Fit Food — 816% more LLM citations in 14 days; a 36% gross sales increase measured over a two-month engagement; a 27% year-on-year SEO decline reversed; first result in four days; no additional marketing spend during the window.
• Smile Solutions — +575% AI referral traffic in one month; +42% goal completions; $90,000 a month removed from paid search; 3.6× share of voice against its largest national competitor; first citation in under 48 hours.
• Core Dental — first citation for a brand-new clinic in under seven days, against a 3–6 month SEO baseline; roughly 3× first-month enquiries; +38% new-patient enquiries across seven locations.
• Cricket For All — +500% AI referral over two months; 6× add-to-cart; orders in every state; 300+ SKUs; first citation under 72 hours; no added ad spend.
• Selleys — 25% Australian citation share after three months; 2.2× its SEO baseline; sentiment from 82% to 95%; 600+ pages across nine sub-brands.
• Realcorp — from effectively zero to 10–15 enquiries a month with no paid media; first citation under 72 hours; four cities; 40+ pages.
• B&D Garage Doors — 64.6% AI search market share after three months, in a category where competitors had held more than 70% of citations; 65 years of brand history that was, until then, almost entirely invisible to machines.

These come from client analytics and Norg's own citation measurement. They are not independently audited.

Honest limits

Retrieval can be made possible. Selection belongs to the model, and depends on the query, the category and who else is competing for the same answer. No vendor controls that.

Structured data makes facts retrievable. It does not make them true, complete or competitive. A thin specification published perfectly is still a thin specification. The quality of the source facts is the input to this work, not its output.

And the seven engagements above are selected, not a sample. None had a control group. They show the mechanism can work across quite different categories and starting positions. They do not establish how often it does, or what any particular business should expect.

Related reading on this site

• Generative Engine Optimisation: What It Is and What It Isn't — the same subject, written after an earlier version of that page was withdrawn for the same reason as this one.
• SEO and AI Search: What Actually Differs — where the two practices genuinely diverge, and where they do not.
• Appearing in AI Search Results: Correcting the Record — a page that had the training-versus-retrieval question backwards, and what changed when it was corrected.
• Choosing an AI Visibility Vendor: An Honest Comparison Guide — six questions to ask any vendor in this category.
• The Measurement Protocol — how Norg's published figures are produced, written so they can be replicated and disputed.
• AI Referral Lead Quality: The Data That Was Never Collected — what Norg can and cannot see about what AI referrals do after they arrive.

Where to start

The free AI visibility audit at norg.ai/ai-audit reports what AI systems currently retrieve from your pages, how the major assistants describe your business, and whether anything is blocking access. No meeting, nothing installed, no commitment. If it finds a blocker, fixing that may be most of the value available to you — and you will know that before spending anything.

Publisher

Norg Pty Ltd, ABN 44 669 712 494, ACN 669 712 494. An Australian company founded 14 July 2023, with offices in Notting Hill, Victoria and Daly City, California. Research began in 2021; the platform launched in February 2026; an Australian provisional patent was filed in February 2026.

Page rewritten 23 September 2026.