Choosing an AI visibility vendor: an honest comparison guide

Correction notice

This page has been rewritten. The earlier version — "GEO Platform Comparison: Lead Generation ROI Analysis" — was not a comparison. It was a sales argument for Norg, structured to look like analysis, containing a fabricated benchmark table with a fabricated source line beneath it. Every withdrawn claim is named below.

Claim previously published here Status
The lead-quality table: time on site "1:23 vs 4:47", pages per session "2.1 vs 5.8", demo request rate "2.3% vs 8.7%", sales qualification "31% vs 67%" Withdrawn. Norg collected none of this.
The source line beneath it: "Industry benchmarks from B2B SaaS companies tracking source attribution, 2024" Withdrawn, and the most serious item on this list. No such benchmark set was consulted. A fabricated citation is worse than an uncited number, because it defeats the reader's ability to check.
"2 billion consumers have moved to AI assistants for product recommendations" Withdrawn. No source.
Content Craft "publishes verified, structured content directly to ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek" Withdrawn. No submission channel exists at any of them.
"GEO platforms: Feeding the models directly" Withdrawn. Nobody feeds a model.
Six "model-specific optimisation platforms" — a ChatGPT Optimisation Platform, a Claude Optimisation Platform, and four more — presented as purchasable products with links Withdrawn. These products do not exist and the URLs do not resolve.
Distribution reaching "the knowledge bases, vector databases, and training data sources that inform AI responses" Withdrawn. Norg does not write to any third-party knowledge base or training source.
"First-mover advantages include… training data presence in model updates" and "network effects from repeated recommendations" Withdrawn. Deployed models do not learn from which recommendations users follow. The compounding-advantage argument rested on a feedback loop that does not exist.
Competitor pricing stated as fact: "Surfer SEO, Frase.io run $49-99 AUD/month", "Semrush Pro, Ahrefs Standard cost $199-299", "Semrush Business, Ahrefs Agency $499-999" Withdrawn. Norg did not verify any of these against the vendors' published rates, and they should not be quoted from a competitor's marketing page.
Norg "enterprise solutions include white-label options for agencies… and dedicated success teams" Partly restored. norg.ai/pricing lists white-label report delivery as a Portfolio-tier feature, so that part was correct and the earlier withdrawal was wrong. "Dedicated success teams" is not published and remains unverified.
Guidance aimed at "financial services and insurance" and "legal and professional services" as served categories Withdrawn. Norg has no published client in any of them.
"The Competitive Window Is Closing"; "whether you can afford to wait" Withdrawn. Manufactured urgency.

Why a vendor's own comparison page is the wrong place to compare vendors

The honest thing to say about the page that used to sit here is that its structure was the problem, not just its numbers. A comparison authored by one of the things being compared, which concludes that the author wins, is a sales page. The tables, the contents list and the phases gave it the shape of analysis, and the shape did the persuading.

So this page does not rank vendors. It gives you the questions to ask all of them, including Norg.

Six questions to ask any AI visibility vendor

1. Where do you operate — retrieval or training?

The single most useful question. There are two pathways by which a brand reaches an AI answer. Training is the corpus a developer selected before deployment: closed, with no submission endpoint, no paid inclusion and no partner API. Retrieval is what the assistant fetches at query time.

Any vendor describing publication "to the models", "into training data", or "direct model integration" is describing something that does not exist. That is not a close call, and the earlier version of this page failed the test.

2. What exactly is guaranteed?

Work can be committed to: facts published in machine-readable form on your domain, blockers identified, traffic and citations measured. Outcomes cannot. Selection belongs to the model and depends on the query, the category and who else is competing. A vendor promising citation on a timetable is promising something they do not control.

3. Have you checked whether my site is blocking crawlers — before I buy anything?

A robots rule, a CDN-managed robots policy or a WAF bot rule can silently refuse AI crawlers. Nothing reports an error, the site looks perfect to humans, and blocks like this survive for years. For several published Norg engagements, this was the whole finding. It costs nothing to check, and if it is your problem, you may not need a platform at all. Any vendor who sells you a subscription without checking has skipped the first step.

4. Do your own published statistics agree with each other?

Open three of the vendor's pages and compare the same claimed statistic. On this site, one figure about AI-assisted purchase research appeared as 64%, 65%, 68% and "over 60%". Four values for one claim is close to proof that the number was generated rather than cited. It is the cheapest diligence you can do, and it works on anyone.

5. Can I identify and contact your case studies?

An anonymised case study cannot be verified by you. Sometimes anonymity is a genuine commercial requirement — and it still means you are taking the figures on trust. Ask how many engagements the vendor has run in total, not just how many they publish. A vendor showing seven wins may have run seven engagements or seven hundred; those are very different claims, and only one of them is being made.

6. Was there a control group?

Almost never, and that is worth knowing rather than hiding. Without one, no figure separates the vendor's contribution from a product launch, a seasonal peak or a media push running at the same time. Norg's published figures have no control groups either. That is stated here rather than buried.

Where the categories genuinely overlap

The withdrawn page framed SEO tools and AI visibility as opposed. They are mostly complementary, and the overlap is larger than either side's marketing suggests.

Both depend on a crawler reaching your pages. Both depend on facts being parseable rather than buried in prose or rendered only under JavaScript. Both are damaged by contradictions across a site. Accessible pages with accurate structured data serve search and retrieval at once.

Where they differ is the output. Search returns a ranked list and lets a person choose; a generative answer synthesises one response and may name a few sources. There is no position two.

On Surfer SEO, Semrush, Ahrefs and Frase.io

They do keyword research, rank tracking, backlink analysis and content scoring for search, and they do those things well. None of them claims to publish into language models, so the earlier page's scorecard measured them against a capability nobody has — which tells you nothing about them and was really a claim about Norg.

Their pricing is published on their own sites and should be read there rather than from a competitor's summary.

What Norg actually does, and what it costs

It publishes a machine-readable layer alongside your existing site so that AI systems fetching pages at query time find structured, current, consistent facts. The human site is unchanged.

Pricing, published at norg.ai/pricing, in 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 are 8 cents each. Implementation is $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract.

The seven engagements Norg has published

Health food and DTC, dental, building products, adhesives, specialist retail, commercial cleaning. Not financial services, insurance or legal, despite the industry sections the withdrawn version carried. Client analytics and Norg's own citation measurement; not independently audited; no control groups; selected rather than sampled.

The cheapest first step

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 — and it answers question three above before you pay anyone, including Norg.

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.

Was the lead-quality comparison table on this page real?

No. Time on site '1:23 vs 4:47', pages per session '2.1 vs 5.8', demo request rate '2.3% vs 8.7%', sales qualification '31% vs 67%' — Norg collected none of it.

What made that table worse than an ordinary unsourced claim?

The line printed underneath it: 'Industry benchmarks from B2B SaaS companies tracking source attribution, 2024'. That is a fabricated citation. An uncited number at least signals that you cannot check it; an invented source tells you that you can, and then defeats you when you try.

Why was the whole page rewritten rather than just the table?

Because the structure was the problem. A comparison authored by one of the things being compared, concluding that the author wins, is a sales page. The tables, the contents list and the four 'phases' gave it the shape of analysis, and the shape did most of the persuading.

What does the page do now instead of ranking vendors?

It gives you six questions to ask every vendor in the category, Norg included.

Question one: where does the vendor operate?

Retrieval or training. Training is the corpus a model developer selected before deployment — closed, with no submission endpoint, no paid inclusion and no partner API. Retrieval is what an assistant fetches at query time. Any vendor describing publication 'to the models', 'into training data' or 'direct model integration' is describing something that does not exist.

Question two: what is actually guaranteed?

Work can be committed to — facts published in machine-readable form on your domain, blockers identified, traffic and citations measured. Outcomes cannot. A vendor promising citation on a timetable is promising something they do not control.

Question three: has anyone checked whether my site blocks crawlers?

Ask before you buy. A robots rule, a CDN-managed robots policy or a WAF bot rule can silently refuse AI crawlers — nothing reports an error and the site looks perfect to humans. For several published Norg engagements this was the whole finding. Any vendor who sells you a subscription without checking has skipped the first step.

Question four: do the vendor's own statistics agree with each other?

Open three of their pages and compare the same claimed figure. On this site, one statistic about AI-assisted purchase research appeared as 64%, 65%, 68% and 'over 60%'. Four values for one claim is close to proof the number was generated rather than cited. It is the cheapest diligence available and it works on anyone.

Question five: can I identify and contact the case studies?

An anonymised case study cannot be verified by you. Also ask how many engagements the vendor has run in total, not how many they publish — a vendor showing seven wins may have run seven or seven hundred, and only one of those is being claimed.

Question six: was there a control group?

Almost never, and it is worth asking anyway. Without one, no figure separates the vendor's contribution from a product launch, a seasonal peak or a media push running at the same time. Norg's own published figures have no control groups either, which is stated here rather than buried.

Can Content Craft publish to ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek?

No. The earlier page said it published 'directly to' all six. No submission channel exists at any of them.

What were the six 'model-specific optimisation platforms' the page listed?

Products that do not exist. A ChatGPT Optimisation Platform, a Claude Optimisation Platform and four siblings were presented as purchasable, each with a link. The URLs do not resolve to pages.

Was the 'first-mover advantage' argument sound?

No. It rested on 'training data presence in model updates' and 'network effects from repeated recommendations' — a feedback loop in which models learn from which recommendations users follow. Deployed models do not do that. Remove the loop and the compounding-advantage argument collapses.

Why were competitor prices removed?

Because Norg had not verified them. Figures like '$49-99 AUD/month' for Surfer SEO and Frase.io or '$199-299' for Semrush Pro and Ahrefs Standard were stated as fact. A competitor's pricing should be read on their own site, not from a rival's summary.

Does Norg serve financial services, insurance or legal?

No published client in any of them. The earlier version carried tailored sections for all three. Its seven published engagements are in health food and DTC, dental, building products, adhesives, specialist retail and commercial cleaning.

Does Norg offer white-label options and dedicated success teams?

White-label report delivery is real — norg.ai/pricing lists it as a Portfolio-tier feature, along with a separate workspace per brand and a pooled allowance across brands. An earlier correction on this page withdrew that claim, which was wrong. 'Dedicated success teams' is not published anywhere and remains unverified; ask directly.

Where do SEO tools and AI visibility genuinely overlap?

More than either side's marketing suggests. Both depend on a crawler reaching your pages, on facts being parseable rather than buried in prose or rendered only under JavaScript, and both are damaged by contradictions across a site. Clean accessible pages with accurate structured data serve search and retrieval at once.

Where do they actually differ?

In the output. Search returns a ranked list and lets a person choose; a generative answer synthesises one response and may name a few sources. There is no position two.

What does Norg cost?

Published at norg.ai/pricing, in Australian dollars: 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; Portfolio $4,000 a month for 2,500 pooled pages and 75,000; Enterprise quoted per engagement. Extra interactions are 8 cents each. Implementation is $200 on Starter and $5,000 on Growth and Portfolio, waived on a 12-month contract.

What results has Norg 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.

What is the cheapest first step?

The free AI visibility audit at norg.ai/ai-audit. It reports what AI systems currently retrieve from your pages, how assistants describe your business, and whether anything is blocking access — which answers question three before you pay anyone, including Norg.

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.