A company can have significant revenue, thousands of customers, strong Google rankings and years of press coverage and still be nearly invisible when buyers ask ChatGPT, Gemini or Google AI questions about its category.
That’s because brand awareness and AI visibility measure two different things.
Brand awareness tells you whether people know your company.
AI visibility tells you whether AI systems surface your company when people ask questions related to the problems you solve.
For a growth-stage technology company, the difference becomes especially important when you look at non-branded prompts.
Ask ChatGPT:
“What are the best platforms for X?”
“Which companies are leading in X?”
“What are the best alternatives to X?”
“What should I buy if I need X?”
“Which X platform is best for a mid-market company?”
Now remove your company’s name from the question.
Do you still appear?
If you don’t, you may have a brand that people know but an authority footprint that AI systems struggle to connect with the category.
That’s a very different problem.
What is machine-visible authority?
Machine-visible authority is the body of discoverable, relevant and credible evidence that helps AI systems understand what a company does, where it belongs and when it should be surfaced in an answer.
That evidence can exist on your website.
But much of it exists elsewhere.
Editorial coverage.
Product reviews.
Industry publications.
Comparison articles.
Expert commentary.
Customer discussions.
YouTube.
Research.
Relevant third-party websites.
The important distinction is that brand awareness exists in people’s heads. Machine-visible authority has to exist in information machines can retrieve and interpret.
A company can have plenty of the former and surprisingly little of the latter.
That’s how a $100 million technology company can find itself absent from an AI answer while a smaller competitor appears prominently.
Brand awareness and AI visibility are different things
Think about what creates traditional brand awareness.
Advertising.
Events.
Sponsorships.
Paid social.
Sales.
Email.
Word of mouth.
Customer relationships.
Retail distribution.
Press coverage.
These activities can build a very successful company.
But when somebody asks an AI system an unbranded question about your category, the system has a different task.
It has to construct an answer.
And increasingly, that process can involve retrieving information from the web.
OpenAI explains that ChatGPT Search can search the web and provide citations. OpenAI also says search results are ranked using multiple factors intended to surface relevant and reliable information.
Google describes a similar retrieval process for its generative search experiences. Its official guidance for AI Overviews and AI Mode says these features use techniques including retrieval-augmented generation to find relevant, current pages from Google’s search index.
That creates an important distinction for CMOs:
| Traditional brand strength | Machine-visible authority |
|---|---|
| People recognize your name | AI associates your name with the category |
| Customers know your products | AI can retrieve evidence about your products |
| You have media coverage | Relevant sources substantiate your expertise |
| You have website traffic | Your content answers important buyer questions |
| You rank for your company name | You appear for valuable non-branded prompts |
| You have a strong reputation | That reputation is represented across sources AI can find |
You can be strong on the left and weak on the right.
That’s where this problem starts.
The real test is what happens when you remove your brand name
One of the easiest mistakes in AI visibility measurement is testing branded prompts.
Ask:
“Is Company X a good cybersecurity platform?”
and you have already told the AI which company to discuss.
Ask:
“What are the best cybersecurity platforms for a 1,000-person company?”
and you’ve created a much harder test.
The second question requires the system to determine which companies belong in the answer.
That’s why non-branded prompts are so useful.
They show whether you have established enough category authority to be discovered rather than simply recognized.
Muck Rack’s methodology for its What Is AI Reading? research makes an interesting distinction here. Its researchers intentionally avoid predictable transactional branded prompts and focus on questions where outcomes are less predictable and potentially more valuable.
CMOs should make a similar distinction.
If your GEO dashboard says you have 85% visibility because you’ve loaded it with prompts containing your company name, congratulations. You’ve discovered that ChatGPT can talk about you when explicitly asked to talk about you.
That’s not particularly interesting.
I want to know what happens when you aren’t mentioned in the question.
Why can a well-known company disappear from non-branded AI prompts?
There usually isn’t one explanation.
For established growth-stage companies, I would investigate six areas.
1. Your reputation is stronger than your category association
People may know your company without clearly associating it with the category you want to own.
This happens surprisingly often.
Companies evolve.
Products expand.
Positioning changes.
A startup that entered the market selling one thing now sells six.
The company moved upmarket.
The messaging changed.
The category changed.
The website changed.
But the company’s broader information footprint didn’t change at the same speed.
You might describe yourself as an enterprise automation platform today while years of articles, reviews and third-party pages still describe you as a workflow tool.
A human who has followed the company understands the evolution.
A machine has to reconcile the evidence it can find.
This is where brand awareness can create false confidence.
Everybody inside your industry may know what you do.
The web may be telling a much messier story.
2. You have lots of press coverage, but it establishes the wrong things
This is particularly common with successful growth-stage companies.
You’ve raised several rounds.
You’ve hired notable executives.
You’ve expanded internationally.
Your founder gets interviewed.
You announce partnerships.
You publish predictions.
You’ve accumulated hundreds of media mentions.
That’s good PR.
But now look at those articles through a different lens.
What do they actually establish about the company?
Do they connect your company with the problems customers are asking AI about?
Do they explain what your product does?
Do they demonstrate expertise?
Do they compare your company with alternatives?
Do they establish why your product belongs among the leading solutions in the category?
Or do they mostly establish that you raised money, hired people and grew?
Those aren’t the same thing.
The distinction matters because third-party sources appear frequently in AI citations.
Muck Rack’s May 2026 What Is AI Reading? research analyzed more than 25 million links across ChatGPT, Claude and Gemini and found 84% of citations came from earned media, while journalism accounted for 27%. Paid and advertorial sources represented just 0.3%.
For a mature communications program, that should trigger an audit.
Don’t just count your coverage.
Ask what your coverage proves.
3. Your website was built for people who already know you
This is another side effect of success.
Early-stage companies often spend enormous amounts of time explaining the problem.
Established companies frequently stop.
Their websites become increasingly product-centric.
Solutions.
Features.
Customers.
Resources.
Book a demo.
Request pricing.
Everything assumes the visitor has already arrived.
But AI discovery frequently begins earlier.
A buyer might ask:
“How do companies solve X?”
Then:
“What software can solve X?”
Then:
“Who are the leading providers?”
Then:
“Which is best for a company like mine?”
Then:
“Company A vs. Company B?”
If your content only becomes useful at question five, you’re absent from much of the journey.
This is why we map GEO against the entire buyer journey at Proper Propaganda, from education and discovery through evaluation, validation, purchase and post-purchase.
Google’s own guidance points in a similar direction. Its guide to optimizing for generative AI features recommends producing valuable, unique content and says established SEO fundamentals remain important to AI Overviews and AI Mode.
The objective shouldn’t be producing hundreds of pages for AI.
It should be making sure genuinely useful answers exist for the questions customers are actually asking.
4. You’re famous for your brand, not necessarily the problem
This distinction deserves more attention.
Search volume for your company name can be high.
Your direct traffic can be excellent.
Your branded search results can be immaculate.
None of that automatically means you’ll dominate:
“Best software for X.”
There is some useful evidence here.
Ahrefs analyzed 75,000 brands across ChatGPT, Google AI Mode and AI Overviews and found branded web mentions correlated strongly with AI visibility, at roughly 0.66 to 0.71 across the platforms studied.
By comparison, traditional metrics weren’t always as closely correlated. For ChatGPT specifically, Ahrefs reported correlations of 0.352 for branded search volume and 0.266 for Domain Rating. YouTube mentions showed the strongest correlation in the study, at roughly 0.737.
These are correlations, not proof of causation.
But they raise an interesting point.
Traditional indicators of brand strength don’t necessarily map neatly onto AI visibility.
A brand can be searched frequently while another brand is discussed more widely in the places and contexts relevant to AI answers.
Those are different footprints.
5. Your competitors have become easier to understand
This one can sting.
Your company might be bigger.
Your competitor might be clearer.
They’ve built comparison pages.
They’ve created category guides.
Their executives consistently talk about the same handful of problems.
Their positioning is repeated across the web.
Reviewers describe them consistently.
Journalists understand where to place them.
Their website answers the obvious questions.
Their product pages contain concrete facts.
Their PR program reinforces their category.
Their YouTube footprint is strong.
Their brand appears repeatedly around the subjects customers research.
Machines don’t know that you have 30% more revenue.
They encounter information.
The competitor may simply have created a clearer body of evidence about where it belongs.
6. You haven’t built authority around the prompts that matter commercially
This is where I think a lot of GEO programs go wrong.
They monitor visibility globally.
I care much more about visibility against commercial intent.
Suppose your company appears frequently for:
“What is [category]?”
That’s useful.
But suppose it almost never appears for:
“Best [category] platforms.”
“Best [category] software for enterprise.”
“Alternatives to Competitor X.”
“Which [category] provider is best for European companies expanding into the US?”
Now we have a different problem.
You’re visible.
You’re just invisible at the moments where visibility may matter most.
That’s why Share of Answer without prompt segmentation can be misleading.
A 40% Share of Answer across low-value educational prompts could be less commercially interesting than 15% visibility across the prompts immediately preceding a buying decision.
Context matters.
AI visibility is becoming large enough to deserve CMO attention
This isn’t happening in some obscure corner of search.
Google said in June 2026 that AI Overviews had more than 2.5 billion monthly active users, while AI Mode had surpassed one billion monthly users.
Google has also continued adding links, sources and discovery features to its generative search experiences.
Meanwhile, platforms such as Ahrefs now explicitly measure AI mentions, citations, impressions and AI Share of Voice alongside more traditional search metrics. The measurement infrastructure is catching up with the behaviour.
For CMOs, that means AI visibility is starting to become measurable in ways that look much more like an actual marketing channel.
How to tell whether you have a brand-awareness problem or an AI-authority problem
Here’s the diagnostic I would run.
Test 1: Branded visibility
Run 20 to 30 questions containing your company name.
For example:
“What does Company X do?”
“Who competes with Company X?”
“Is Company X good for enterprise customers?”
You should perform strongly here.
If you don’t, you may have a fundamental entity, content or reputation problem.
Test 2: Category visibility
Now remove your name.
Run questions such as:
“What are the leading companies in X?”
“What are the best platforms for X?”
“Who makes X?”
“What software should I use for X?”
This measures whether AI associates you with the category.
Test 3: Problem visibility
Go one step earlier.
Ask the questions a buyer might ask before knowing what kind of product they need.
“How can a company solve X?”
“What’s the best way to reduce X?”
“How do companies manage X?”
“What technology helps with X?”
This tells you whether you’re visible around the problem itself.
Test 4: Evaluation visibility
Now move toward purchase.
“Best X for enterprise companies.”
“Company A vs. Company B.”
“Alternatives to Company A.”
“Which X platform has the best Y?”
“Which X provider is best for companies in Europe?”
These are high-value prompts.
Test 5: Citation footprint
Finally, stop looking at your company for a moment.
Look at the sources.
Which websites are being cited?
Which publications appear repeatedly?
Which comparison pages matter?
Which YouTube channels appear?
Which expert sources surface?
Which competitor-owned pages appear?
Then ask:
Are we represented in those sources?
That’s where GEO becomes actionable.
A useful visibility matrix for growth-stage CMOs
Put the results into four buckets.
| Low AI visibility | High AI visibility | |
|---|---|---|
| High brand awareness | Machine authority gap | Category leader |
| Low brand awareness | Discovery problem | Emerging AI-native challenger |
For established growth-stage companies, the upper-left box is particularly interesting.
High awareness. Low AI visibility.
You don’t necessarily need more people to know the company exists.
You need a better representation of the company’s authority across the information environment AI systems encounter.
That’s a different communications brief.
How do you build machine-visible authority?
Once you’ve identified the gap, there are five areas I’d concentrate on.
First, map your prompt universe.
Don’t optimize for “AI.”
Identify the actual questions customers ask across education, discovery, evaluation, validation and purchase.
Second, map the sources surrounding those prompts.
Which third-party sources consistently appear?
This is where PR and GEO begin to overlap.
Third, audit your category associations.
Look at how your company is described across your website, editorial coverage, review sites, YouTube, industry publications and other influential sources.
Is the market seeing a consistent company?
Fourth, fill genuine content gaps.
Create the best answer you can for questions where useful information is currently missing.
Google explicitly recommends valuable, unique, non-commodity content for its generative search features.
Give it some.
Fifth, build third-party authority around the gaps.
This could involve editorial coverage.
Product reviews.
Original research.
Expert commentary.
Industry comparisons.
Relevant podcasts.
YouTube.
Community participation.
The tactic depends on where the authority gap actually exists.
Why PR becomes more important for established companies
There’s a tendency to think of GEO as a technical discipline.
Some of it is.
Your website needs to be accessible. Your content needs to be discoverable. Your pages need to be indexable. Google says the normal technical requirements for Search continue to apply to AI Overviews and AI Mode, with no special AI-specific technical requirement needed.
But technical optimization can only take you so far.
You can’t put schema on your website and manufacture an independent reputation.
You can’t rewrite a title tag and suddenly create five years of category authority.
You can’t publish a page declaring yourself one of the best companies in your industry and expect independent systems to treat the claim as settled fact.
Some authority has to be earned elsewhere.
That’s why the Muck Rack data is so interesting. Across three editions of its research since July 2025, earned media represented between 82% and 89% of AI citations, with 84% in the May 2026 study.
For communications teams, GEO creates a new reason to care about earned authority.
Media coverage can do more than reach an audience on publication day.
It can contribute to the body of third-party information surrounding a company, product and category.
The CMO dashboard needs another line
Most established growth-stage CMOs already track some combination of:
Brand awareness.
Share of voice.
Organic traffic.
Branded search.
Media coverage.
Pipeline.
CAC.
Conversion.
Revenue.
I’d add another category:
AI visibility.
And underneath it I’d track at least:
Share of Answer
How frequently are we included across the prompts that matter?
Non-Branded Prompt Coverage
How often do we appear when the buyer doesn’t mention us?
Competitive Share of Answer
How does our visibility compare with direct competitors?
Citation Footprint
Which sources are shaping the answers?
Category Association
Which categories, attributes and problems are we associated with?
Recommendation Rate
When we appear, are we actually recommended?
The objective isn’t to replace your existing marketing metrics.
It’s to identify a blind spot they may not show you.
Being famous isn’t the same as being findable by AI
This is the larger shift.
For the last 20 years, digital marketing was heavily concerned with helping people find your website.
AI search introduces another layer.
Increasingly, the machine is doing some of the research first.
It searches.
It retrieves.
It compares.
It summarizes.
It recommends.
And sometimes the customer encounters your brand only after that process has already started.
For established growth-stage companies, that creates an uncomfortable possibility.
You can have a strong brand and a weak machine-visible reputation.
Your customers know you.
Your investors know you.
Journalists know you.
Your category knows you.
But when a buyer asks an AI system an unbranded question about the problem you solve, you disappear.
That’s worth investigating.
Because the important question for the next phase of search isn’t simply:
“How well known are we?”
It’s also:
“When nobody mentions our name, does the web contain enough credible evidence for AI to find us anyway?”
For a growth-stage CMO, that’s the test that matters.