31 August 2026

AI visibility is the new marketing metric. But what exactly are we measuring?

In summary

• What: AI visibility tools are emerging to measure how often and how prominently brands appear across generative AI platforms.
• Why now: Marketers are already using these tools to inform content, search and broader marketing strategies.
• The challenge: Different AI models and measurement methodologies can produce very different results for the same brand.
• What marketers should do: Use AI visibility as a useful signal, but understand how the number has been produced before making decisions from it.

AI visibility is quickly becoming the latest metric marketers are being told they need to track.

Brands are investing in tools to understand whether they appear when consumers ask ChatGPT, Gemini, Claude or Perplexity for recommendations. Some are already changing content strategies, SEO investment and creator activity in an attempt to improve those results.

But there is a fairly significant problem: your brand’s AI visibility depends on how you measure it.

Recent analysis by Fractl found just 11% of brands in its dataset were referenced across ChatGPT, Claude and Gemini. Another 12% appeared across two, while 77% were referenced by only one model.

A brand can therefore look highly visible in ChatGPT and virtually disappear in Gemini. Change the prompts being tested, the model, the frequency or even the definition of a “mention”, and the picture can change again.

Yet marketers are increasingly being presented with neat dashboards and precise-looking AI visibility scores that can make this emerging area of measurement look far more settled than it actually is.

And money is beginning to follow those numbers.

In the US, Zoom, for example, has been experimenting with creator activity as part of its efforts to influence how the company appears in AI search, as it looks to shift perceptions of the brand beyond video meetings towards its broader AI workplace offering.

Once AI visibility starts influencing content strategy, creator investment, search activity and eventually budgets, the reliability of the measurement behind those decisions becomes much more important.

The industry is now starting to address that problem.

Following IAB Australia’s inaugural AI & Search Summit in Sydney in July, which put AI discovery and measurement firmly on the local agenda, the US-based IAB released new standardised guidelines in August for measuring brand and publisher visibility across AI-powered discovery platforms.

For Australian marketers, this isn’t a minor measurement issue. As AI becomes part of how consumers research brands, products and services, local teams need to understand what AI visibility metrics can, and can’t, reliably tell them.

The IAB said more than 20 companies now sell AI visibility measurement tools, using different methodologies that can produce different answers for the same brand or publisher.

Before marketers start using AI visibility to inform investment decisions, there is a more fundamental question to answer: can we actually trust the number we’re being given?

AI visibility has an old measurement problem

Anyone who has spent enough time around digital advertising has seen versions of this story before.

A new channel emerges. New measurement providers follow. Different methodologies develop and eventually advertisers start asking why the numbers don’t match.

Viewability went through it. Attribution continues to go through it. Attention measurement has faced similar questions.

AI visibility may now be entering its own version of that cycle, only much faster.

Generative AI is fast becoming another way people search for information, research products and compare brands. And as that behaviour grows, so too does the number of tools claiming to measure which brands are actually showing up.

But measuring AI visibility isn’t as straightforward as checking where a webpage ranks in search.

Why measuring AI isn’t the same as measuring search

With traditional search, marketers are familiar with many of the basic measures. Rankings can be observed. Impressions and clicks can be measured. Search and analytics platforms can provide additional signals about what happened afterwards.

Generative AI works differently.

An AI response can change depending on the question, model, context, location and user. Ask the same question again and you may not get exactly the same answer.

That leaves marketers with some fairly basic questions that aren’t always easy to answer.

Which prompts are being tested? How are they selected? Which models are being measured? And what actually counts as visibility?

Does a brand simply need to be mentioned? Does it need to be recommended? Is being cited as a source different from appearing in the answer itself? Does the position or sentiment of the mention matter?

Those decisions can materially change the number a marketer ultimately sees.

A visibility score needs context

A dashboard showing that a brand has “35% AI visibility” looks reassuringly precise.

But that number doesn’t mean much without context.

A score based on 50 commercially relevant prompts is very different from one based on 5,000 broader prompts. Measuring ChatGPT alone is different from measuring ChatGPT, Gemini, Claude and Perplexity together.

And being mentioned somewhere in an AI response is very different from being recommended as the answer to someone’s question.

The IAB is attempting to bring more consistency to this through what it calls the 4 Ps of AI Visibility: presence, prominence, portrayal and persuasion.

But even with a framework, some of those things are easier to measure than others.

Louder’s Digital Ad Specialist Daniel Lim says presence is the clearest.

“Presence is relatively straightforward: did the brand appear or didn’t it? Prominence is a little harder, but you can still look at things like where the brand appeared and which competitors appeared alongside it,” he says.

Where it gets more complicated is portrayal and persuasion.

Portrayal looks at how a brand has been represented. But assessing whether that representation is positive, negative or accurate can itself involve judgement.

“You can end up asking one AI model to judge what another AI model has said about a brand,” Lim says. “The problem is there isn’t always one agreed definition of what an accurate representation of that brand should look like.”

Persuasion is similarly difficult.

A click can tell you someone engaged with a link. It can’t necessarily tell you what convinced them to click, or even whether the AI response was positive about the brand in the first place.

“With a search ad, the advertiser knows what someone saw before they clicked because they wrote the ad,” he says.

“With AI, you don’t necessarily know what was said before the link was served. The AI could be critical of a brand and still provide a link to it. Someone clicking that link doesn’t automatically mean they’ve been persuaded.”

That’s an important distinction. The framework gives the industry a common way to start talking about AI visibility. It doesn’t mean every part of the measurement has been solved.

Useful doesn’t mean absolute

That doesn’t make AI visibility reports useless.

Lim’s view is that some insight is still better than none.

“The fact that different AI visibility reports don’t always agree isn’t a reason to ignore them,” he says. “It’s a reason to understand how the numbers were produced.”

That’s probably the most useful way for marketers to think about AI visibility today.

The data can help show whether competitors are appearing more frequently, which sources are influencing AI responses, where a brand may be absent from important category conversations and how that picture changes over time.

What it shouldn’t necessarily be treated as is an absolute measure of performance.

If a visibility score moves from 32% to 38%, for example, that doesn’t automatically mean a strategy is working. Some of that movement could come from changes in prompts, models or responses.

The IAB framework is worth paying attention to, but the name attached to a framework shouldn’t replace marketers’ own judgement.

As Dan puts it: “Use the numbers, but understand how they were built and keep asking what they’re actually telling you. Time will tell which measures hold up.”

Visibility is only the beginning

There is also a bigger question here.

For most marketers, simply appearing in ChatGPT isn’t the end goal.

What ultimately matters is whether that visibility changes behaviour.

Does it generate referral traffic? Increase branded search? Influence consideration? Lead to conversions?

And as AI platforms move beyond answering questions towards taking actions on behalf of users, website traffic itself may not always be the right outcome to measure.

That’s where AI measurement eventually needs to go: beyond whether a brand appeared towards understanding whether that appearance contributed to a meaningful business outcome.

What marketers should be asking now

AI visibility measurement isn’t going away.

As ChatGPT, Gemini and other AI platforms become a bigger part of product discovery and consumer decision-making, brands will understandably want to know how they appear within them.

But before adding another score to the marketing dashboard, marketers should ask what sits behind it:

• Which AI platforms and models are being measured?
• How are prompts selected and are they representative of real customer behaviour?
• What actually counts as being “visible”?
• How are citations, recommendations and sentiment treated?
• How is variability between AI responses handled?
• Can the results be reproduced?
• Is the metric directional, or robust enough to inform investment decisions?
• Can it be connected to traffic, conversions or other business outcomes?

Get in touch

Get in touch with Louder to discuss how AI-driven discovery is changing measurement, and how your organisation can build a clearer picture of brand visibility and performance across emerging AI platforms.

Want more Louder content?
Add Louder as a Preferred Source in Google Search to see more of our articles when you search and subscribe to our newsletter to receive the latest industry updates straight to your inbox.