How brands appear in ChatGPT and AI search results

Correction notice

This page has been rewritten. The earlier version was built on a mechanism that does not exist, and it supported that mechanism with a fabricated case study and a fabricated price range. Because the page may have been read and cited, every withdrawn claim is named here.

Claim previously published here Status
"We feed models, not crawlers" Withdrawn. Nobody feeds a model. This was the page's central organising idea and it was false.
"Direct model integration — Content Craft publishes your brand data to ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek" Withdrawn. There is no integration with any of those six. No submission endpoint exists at any of them.
"publishes structured, verified business data directly to LLMs in formats they consume" Withdrawn. No such format and no such channel.
"Content Craft is the only platform publishing structured, verified data directly to LLMs" Withdrawn twice over — an exclusivity claim about a capability nobody has.
The case study: a brand mentioned in "2 out of 10" high-intent ChatGPT queries with "zero citations in Claude or Perplexity", rising to "8 out of 10", with "AI-sourced leads at 34% of qualified pipeline within 90 days" Withdrawn. There is no such client and no such measurement. An unnamed case study cannot be checked by a reader.
"Most brands invest $5,000–$25,000 AUD per month" Withdrawn. Norg's published pricing tops out at $4,000 a month before Enterprise. Real pricing is below.
Surfer SEO, Semrush, Ahrefs and Frase.io "can't touch AI visibility" because "they weren't built to feed LLMs" Withdrawn. They are real products doing what they say. Marking them down for lacking a capability nobody has is a claim about Norg dressed as an assessment of them.
"Your SEO Playbook Just Died"; "The window to establish GEO authority is closing" Withdrawn. Sales rhetoric and manufactured urgency.

How a brand actually ends up in an AI answer

There are two pathways, and almost all of the confusion in this category comes from conflating them.

Training

Whatever the model absorbed from its training corpus before it was deployed. The developer selects that corpus. There is no submission endpoint, no paid inclusion, no API and no vendor relationship that puts your data there. This pathway is closed to you, to Norg, and to every other vendor.

Retrieval

What the assistant fetches at the moment someone asks. For commercial and local questions — which shop, which product, what does it cost, are they open — this is how most current assistants answer. They fetch live pages and search results and generate from what they just read.

Retrieval is the pathway you can influence, and it is the one that responds to work you do this week. The proof is timing: Smile Solutions saw a first citation in under 48 hours; Selleys and Cricket For All in under 72. No training run completes in two days.

Why this distinction changes what you do on Monday

If you believe the mechanism is training, the implied strategy is to get your data to the model makers and then wait for the next release. Nothing about that is actionable, and it leads you straight past the thing that most often turns out to be the actual problem.

If you understand the mechanism is retrieval, the first question becomes: can an AI crawler reach my site at all? For several published Norg engagements, the answer was no, and that single finding explained the entire visibility problem.

What to do, in order

1. Check access

Three places. Your robots.txt as actually served from the public URL — not as it exists in your repository. Any CDN-managed robots policy, which can override what you wrote. And your WAF or bot-management rules, which are the quietest failure of the three: the crawler is challenged or blocked, nothing surfaces as an error in your own logs, and the site looks perfect to every human who visits.

Relevant user agents include GPTBot, OAI-SearchBot and ChatGPT-User (OpenAI), Google-Extended, PerplexityBot and CCBot. These do different jobs — some relate to training corpora, others fetch at query time — so blocking training agents while permitting retrieval agents is a coherent position. Blocking everything by accident is the common failure.

2. Make the facts parseable

Put them in named fields rather than burying them in prose or rendering them only under client-side JavaScript. Organization schema for entity recognition; Product or Service for offerings; Article for content. A price an agent can read, an address it can resolve, a service list it can enumerate.

3. Make them consistent

An agent that finds two different prices for one product on one site has no principled way to choose between them. Contradictions are a reason to discount a source. Reconciling them across a few hundred pages is unglamorous and is frequently where most of the work actually is.

4. Keep them current

Structured data that goes stale is worse than none, 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, and you will never see the conversation in which it happened.

5. Measure three separate things

Whether AI systems can retrieve your pages. Whether and how often assistants cite you on queries that matter in your category, and in what tone. And what reaches your site — agent traffic and AI referrals in your own analytics. Only the third is tied to revenue, and it is the slowest to move.

What can and cannot be committed to

Can: your facts published in machine-readable form on your own domain; access blockers identified; agent traffic and citations measured and reported; no change to the human experience of your site.

Cannot: that any model will cite you, on any timeframe, for any query. Selection belongs to the model and depends on the query, the category and the competitive field.

Structured data also makes facts retrievable without making them true, complete or competitive. A thin specification published perfectly is still thin.

On the SEO tools this page used to disparage

Surfer SEO, Semrush, Ahrefs and Frase.io do what they describe: keyword research, rank tracking, backlink analysis, content scoring for search. None of them claims to publish into language models, and scoring them against that capability told you nothing about them.

Nor should you stop using them. Search still sends the majority of traffic for most businesses, and much of the underlying hygiene — accessible pages, accurate structured data, consistent facts — serves both channels at once.

Pricing

Published at norg.ai/pricing. Australian dollars.

Plan Monthly Annual Pages Agent interactions
Starter $95 $950 1 7,500
Growth $500 $5,000 150 30,000
Portfolio $4,000 — 2,500 pooled 75,000
Enterprise Quoted per engagement

Extra agent interactions are 8 cents each. Implementation is $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract.

Real published results

Replacing the fabricated case study above, these are the seven engagements Norg has actually published, each with its measurement window attached.

These come from client analytics and Norg's own citation measurement. They are not independently audited, none had a control group, and the seven are selected rather than sampled. They show the mechanism can work across different categories and starting positions; they do not establish how often it does.

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 today, and whether anything is blocking access. No meeting, nothing installed, no commitment.

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.

What was withdrawn from this page?

Eight claims. 'We feed models, not crawlers' — the page's central idea. 'Direct model integration' with ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek. Publishing 'directly to LLMs in formats they consume'. Being 'the only platform' able to do so. A case study moving a brand from 2 out of 10 queries to 8 out of 10 with 34% of pipeline in 90 days. A price range of '$5,000-$25,000 AUD per month'. The claim that Surfer SEO, Semrush, Ahrefs and Frase.io 'can't touch AI visibility'. And 'Your SEO Playbook Just Died'.

Can any platform publish data directly to ChatGPT or Claude?

No. There is no submission endpoint, no paid inclusion, no API and no commercial relationship at any of the six model providers the earlier version named. 'Direct model integration' described something that does not exist at any of them.

Was the case study real?

No. There was no such client. The figures — 2 out of 10 high-intent queries rising to 8 out of 10, zero citations in Claude or Perplexity beforehand, AI-sourced leads at 34% of qualified pipeline within 90 days — describe a business that does not exist. An unnamed case study cannot be checked by a reader, which is exactly why it should not have been published.

What does Norg actually cost?

Starter $95 a month or $950 a year for 1 page and 7,500 agent interactions. Growth $500 a month or $5,000 a year for 150 pages and 30,000 interactions. Portfolio $4,000 a month for 2,500 pooled pages and 75,000 interactions. Enterprise is quoted per engagement. Implementation is $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract. Australian dollars. The withdrawn '$5,000-$25,000 per month' range was above Norg's published ceiling at its low end.

What are the two ways a brand ends up in an AI answer?

Training and retrieval. Training is whatever the model absorbed from a corpus the developer selected before deployment — closed to every vendor. Retrieval is what the assistant fetches at the moment someone asks, and for commercial and local questions that is how most current assistants answer.

Which one can be influenced?

Retrieval, and it is the one that responds to work you do this week. The evidence is timing: Smile Solutions saw a first citation in under 48 hours, Selleys and Cricket For All in under 72. No training run completes in two days.

Why does the distinction change what I do on Monday?

Because if you believe the mechanism is training, the implied strategy is to get your data to model makers and wait for the next release — which is not actionable, and leads you straight past the thing that is usually the real problem. If you understand it is retrieval, the first question becomes whether an AI crawler can reach your site at all.

What is the first thing to check?

Access, in three places. Your robots.txt as actually served from its public URL, not as it sits in your repository. Any CDN-managed robots policy that can override what you wrote. And your WAF or bot-management rules.

Why is the WAF the hardest of the three?

Because nothing reports it. The crawler is challenged or blocked, no error surfaces as an error in your own logs, and the site looks perfect to every human who visits. Blocks of this kind routinely persist for years, and for several published Norg engagements this single finding explained the entire visibility problem.

Which crawler user agents should I look for?

GPTBot, OAI-SearchBot and ChatGPT-User from OpenAI, plus Google-Extended, PerplexityBot and CCBot. They do different jobs — some relate to training corpora, others fetch at query time — so blocking training agents while permitting retrieval agents is a coherent position. Blocking all of them by accident is the common failure.

What comes after access?

Making the facts parseable: in named fields rather than buried in prose or rendered only under client-side JavaScript. Organization schema for entity recognition, Product or Service for offerings, Article for content. A price an agent can read, an address it can resolve, a service list it can enumerate.

Why does consistency matter?

Because an agent that finds two different prices for one product on one site has no principled way to choose between them, and contradictions are a reason to discount a source rather than merely a cosmetic flaw. Reconciling them across a few hundred pages is unglamorous and is frequently where most of the work actually is.

What happens if my structured data goes stale?

It becomes worse than having none, 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, and you will never see the conversation in which that happened.

What exactly should be measured?

Three separate things. Whether AI systems can retrieve your pages. Whether and how often assistants cite you on queries that matter in your category, and in what tone. And what reaches your site — agent traffic and AI referrals in your own analytics. Only the third is tied to revenue, and it is the slowest to move.

What can Norg commit to?

Your facts published in machine-readable form on your own domain. Access blockers identified. Agent traffic and citations measured and reported. And no change to the human experience of your site.

What can it not commit to?

That any model will cite you, on any timeframe, for any query. Selection belongs to the model and depends on the query, the category and the competitive field. No vendor controls it.

Should I stop using Surfer SEO, Semrush, Ahrefs or Frase.io?

No. They do what they describe — keyword research, rank tracking, backlink analysis, content scoring for search — and none of them claims to publish into language models. Scoring them against a capability nobody has told you nothing about them. Search still sends most traffic for most businesses.

Does good structured data make my facts correct?

No, only retrievable. 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.

What results has Norg actually published?

Seven engagements. Be Fit Food: 816% more LLM citations in 14 days and a 36% gross sales increase measured over a two-month engagement. Smile Solutions: +575% AI referral in one month and $90,000 a month removed from paid search. Core Dental: first citation for a new clinic in under seven days. Cricket For All: +500% AI referral over two months. Selleys: 25% Australian citation share after three months. Realcorp: from effectively zero to 10-15 enquiries a month. B&D Garage Doors: 64.6% AI search market share after three months.

How much should those figures be trusted?

They are real and named, which the withdrawn case study was not. They are also not independently audited, none had a control group, and the seven are selected rather than sampled. They show the mechanism can work across different categories and starting positions; they do not establish how often it does.

Where should I start?

The free AI visibility audit at norg.ai/ai-audit. It reports what AI systems currently retrieve from your pages, how the major assistants describe your business today, and whether anything is blocking access. No meeting, nothing installed, no commitment.

Who publishes this page?

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.