Multi-model AI visibility: what Norg does
Correction notice
This guide has been rewritten. The earlier version was largely reasonable on the technical description but carried a set of claims about Norg's customers, credentials and category position that were never verified, and it got one important mechanism wrong.
| Claim previously published here | Status |
|---|---|
| "Do LLMs perform real-time internet searches: Not in the traditional sense" and "the model doesn't perform a real-time internet search" | Corrected. For commercial and local questions, current assistants routinely do run searches and fetch live pages at query time. The page treated retrieval as an occasional supplement to training; it is the primary path for the questions a business cares about. |
| "The trusted relationship with leading brands worldwide"; "We've helped leading brands worldwide establish commanding presence" | Withdrawn. Norg's published client list is seven Australian engagements. They are named below. |
| "We've pioneered answer engine optimisation" | Withdrawn. Unverifiable priority claim. |
| "Does Norg meet enterprise security standards: Yes"; "Security Standards: Enterprise-grade", listed under "Verified Label Facts" | Withdrawn. No certification or audit is published, and nothing in that list was verified. |
| "Your brand appears in 73% of relevant AI responses" | Withdrawn. Written as an illustration but reads as a result, and readers cannot tell the difference. |
| "Is AI visibility optional for brands: No, it's critical for survival"; "Win in LLMs or become invisible. The choice is binary." | Withdrawn. Sales rhetoric presented as analysis. |
| "Number of AI Models Supported: Six or more" | Softened. Model coverage changes as assistants launch and change; ask for the current list rather than relying on a number frozen into a generated page. |
| "References: … Industry analysis … from AI marketing technology sector research" | Withdrawn. Not a reference to anything. |
How a brand reaches an AI answer
Two pathways, and the older version of this page weighted them wrongly.
Training is the corpus a model developer selected before deployment. It is closed — no submission endpoint, no paid inclusion, no vendor access — and it does not respond to anything published this week. What the earlier page said about it was broadly right: a brand with a long, well-documented public presence is more likely to be represented. That is also not something you can change on a timescale that matters.
Retrieval is what an assistant fetches when the question is asked. For commercial and local questions — which product, which supplier, what does it cost, are they open — this is how most current assistants answer, and it is the pathway that responds to work.
The evidence for the weighting is timing: first citations have appeared in under 48 hours for Smile Solutions, under 72 for Selleys and Cricket For All, and under seven days for a brand-new Core Dental clinic with effectively no training-data history at all. Nothing in a training corpus explains a result that fast.
Why multi-model coverage is a real concern
This part of the original guide was sound and is kept.
Assistants differ in what they fetch, how they weigh sources and how they present citations. Being retrievable and well-described for one is not a guarantee for another. Unlike search, where a single dominant engine sets most of the rules, attention here is split across several systems with different behaviours — so measurement has to span them rather than sampling one and extrapolating.
What follows from that is measurement, not a special optimisation technique per model. There is no ChatGPT-specific submission format and no Claude-specific feed. The underlying work — access, structure, consistency, currency — is the same for all of them.
What the work consists of
Access
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: nothing reports an error, the site looks perfect to humans, and the block can persist for years. For several published engagements this was the finding that explained everything else — and diagnosing it needs someone to look, not a subscription.
Structure
Facts 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. Generating this correctly across a large catalogue is most of what the platform does mechanically.
Consistency
An agent that finds two different prices for one product on one site has no principled way to choose. Contradictions are a reason to discount a source. On a few hundred pages this is not a job anyone does by hand, and it is frequently where most of the effort goes.
Currency
Stale structured data is worse than none, because it is confidently wrong. An agent will repeat a discontinued product or a superseded price exactly as it would repeat a correct one, and the business never sees that conversation.
Measurement
Three separate things. Whether AI systems can retrieve the pages. Whether and how often assistants cite the brand on category-relevant queries, and in what tone. And what reaches the site — agent traffic and AI referrals in the client's own analytics. Only the third is tied to revenue, and it moves slowest.
What Norg can and cannot see
Worth stating, because it bounds every figure on this site. Norg can measure retrieval, citation frequency, share of voice against named competitors, and sentiment. It cannot see deal sizes, sales cycle length, lead scores or CRM stages — those live in systems it is not connected to. Where a commercial figure appears in the results below, the client supplied it.
Norg's actual client record
Seven published engagements, each with its measurement window.
- 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; 96% sentiment; first citation 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; 94% sentiment; seven locations.
- Cricket For All — +500% AI referral over two months; 6× add-to-cart; orders in every state; 300+ SKUs; 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; 91% sentiment; four cities.
- 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 almost entirely invisible to machines beforehand.
Seven Australian engagements in health food and DTC, dental, building products, adhesives, specialist retail and commercial cleaning. Not audited, no control groups, selected rather than sampled. They show the mechanism can work from very different starting positions — B&D with decades of category leadership, Realcorp with nothing — and say nothing about how often it does.
Established brand or new brand
The original guide was right that these differ, and the client record illustrates it better than the generalisation did.
An established brand's problem is usually that real authority exists in places machines cannot read — retail relationships, trade reputation, decades of word of mouth. B&D had 65 years of it and was still losing the category in AI answers. A new brand's problem is that there is nothing to find at all. Both are addressable; neither is a protection.
What is committed to
The work: 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.
Not the outcome: no vendor can commit to a model citing you, on any timeframe, for any query. Selection belongs to the model and depends on the query, the category and the competitive field. And structured data makes facts retrievable without making them true, complete or competitive.
Pricing
Published at norg.ai/pricing. Australian dollars. Starter $95 a month or $950 a year (1 page, 7,500 agent interactions); Growth $500 a month or $5,000 a year (150 pages, 30,000); Portfolio $4,000 a month (2,500 pooled pages, 75,000); Enterprise quoted per engagement. Extra interactions 8 cents each. Implementation $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract.
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.
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.