Written by Julien Ricciarelli-Bonnal
14 August 2026
The Essentials
Companies are beginning to invest in tools that measure how visible their brands are in ChatGPT, Google AI Overviews and other AI-powered search environments. They want to know whether they are mentioned, recommended, positioned ahead of competitors and associated with the right topics. The problem is that this visibility remains difficult to connect directly to sales, leads or revenue. After impressions, followers and engagement rates, marketing may therefore be creating a new category of vanity metrics: reassuring indicators because they move, but whose commercial value remains uncertain. Visibility in AI probably already matters. The challenge is understanding what it actually means.
For years, digital marketing promised something traditional media could rarely offer with the same level of precision: measurement. A campaign generated impressions, clicks, sessions, conversions and, in the best cases, revenue that could be directly attributed. This ability to follow almost every stage of the journey helped establish the idea that a marketing action should eventually produce a number.
AI-powered search is changing that logic again. A user can ask ChatGPT which brands to consider, compare several options through Google AI Overviews, receive a recommendation, close the conversation and make a purchase a few hours later on Amazon, in a physical store or directly on a company website. The brand may have influenced the decision without generating a single identifiable click.
Marketing teams are therefore looking for new indicators. Tools such as Profound, Scrunch, Semrush or AirOps now try to measure how frequently a brand appears in generated answers, how it compares with competitors and which sources models use when making recommendations. Recent surveys suggest that many marketers have already started investing in this kind of AI visibility tracking.
Measuring that presence makes sense. Automatically treating it as proof of commercial performance makes much less sense.
Marketing Is Rediscovering a Problem It Thought It Had Solved
Visibility in traditional search followed a relatively familiar logic. A company appeared for a query, the user could click through to the website and eventually convert. Attribution was never perfect, but there was at least an observable continuity between exposure, visit and part of the commercial result.
In an AI-generated answer, that continuity can disappear. ChatGPT can mention a company without immediately sending the user to its website, while Google can summarise multiple sources directly inside the search experience. A recommendation may simply place a brand in the consumer’s mind, with the conversion taking place later and somewhere else.
This is already creating measurement problems for companies trying to understand whether AI visibility actually generates revenue. Some brands are combining traffic, conversions and signals from AI tools, while others are adding brand-search data, CRM information or attribution models in an attempt to reconstruct a journey that is no longer visible from start to finish.
AI-powered search therefore does not remove the need for measurement. It simply brings marketers back to a reality they had sometimes started to forget: a purchasing journey cannot always be divided neatly into a sequence of clicks.
Being Mentioned by ChatGPT Is Not Yet a Commercial Objective
The danger begins when the metric becomes the objective. As soon as a platform displays a visibility score, share of presence or ranking against competitors, teams naturally start trying to improve that number, just as companies once chased followers, impressions or engagement without always asking what those indicators actually produced.
A brand can increase its presence in LLM responses without every mention carrying the same value. Being listed among ten companies in response to a broad question is not equivalent to being recommended when a user is actively looking for a solution and is close to making a purchase. A dashboard can nevertheless treat both appearances as part of the same visibility score.
Context matters as well. A brand may be cited frequently because it already has strong historical awareness, because models can find a large volume of content about it or because it is regularly included in comparison articles. None of these factors guarantees that the recommendation is favourable, distinctive or persuasive enough to change a buying decision.
A mention is useful information. It is not yet evidence of a sale.
This is what makes some new AI metrics look uncomfortably similar to the vanity metrics that shaped the social media era. A rising number reassures teams, helps fill reports and creates the impression that a strategy is moving forward, but an indicator only becomes genuinely useful when the company knows what decision it is supposed to support.
Zero-Click Behaviour Makes the Link Between Visibility and Results Even Harder
The issue goes beyond ChatGPT. Google has been moving for years towards search experiences in which more of the answer is delivered directly inside the interface, while OpenAI has gradually expanded product discovery and comparison features inside ChatGPT. A brand can therefore receive value from visibility before the user ever leaves the platform.
For marketers, the attribution problem becomes almost impossible if the goal is to identify one perfect path. A person might discover a company through ChatGPT, search for the brand on Google, read Reddit, watch a YouTube video and buy three days later from a mobile device. Giving full credit to the last click would obviously be simplistic, but assigning an exact percentage to each previous exposure is often little more than reconstruction.
This is why triangulation is becoming useful again. Rather than waiting for one tool to provide a definitive answer, companies can compare changes in branded search, traffic from AI platforms, customer surveys, CRM data, conversions, awareness studies and statistical models. Traffic from ChatGPT to business websites has been growing rapidly, which confirms that the channel matters more than it did a year ago, but growth in referrals still does not solve the question of how much revenue AI-generated visibility actually influenced.
For companies trying to build a marketing strategy around indicators that are genuinely useful rather than simply reassuring, the answer may therefore involve accepting that part of the customer journey will remain invisible. Measurement still matters, but no single AI visibility score can solve an attribution problem that existed long before generative search arrived.
The “AI Score” Could Become the Next Number Everyone Shows in Meetings
Every generation of digital marketing has produced its favourite indicator. Facebook fan counts once served as proof of success, followed by impressions, reach, engagement and traffic, all of which were sometimes used as substitutes when it was more difficult to demonstrate a real commercial effect.
AI visibility could follow exactly the same path. A dashboard showing that a brand appears in 62% of sector-related answers looks immediately useful. It can be compared with the previous month, benchmarked against competitors and turned into an annual objective. The number has all the characteristics of a good KPI except perhaps the most important one: knowing what a ten-point increase actually changes for the business.
That does not make these tools useless. They can reveal topics where a brand is absent, identify the sources influencing generated answers, highlight reputation problems or show that a competitor is becoming more visible. The metric becomes far more valuable when it is treated as a diagnostic tool rather than automatic proof of performance.
The real risk would be repeating what companies have already done with social media: optimising what is easy to measure instead of what actually matters.
We May Need to Measure Less Precisely, but More Intelligently
Traditional search was never as perfectly attributable as dashboards sometimes made it appear. Cookies, multiple devices, word of mouth, offline media and long decision cycles already made customer journeys far more complicated than last-click models suggested.
AI-powered search simply makes that complexity harder to ignore. When a recommendation influences a decision without generating an immediate visit, marketers need to return to methods they sometimes pushed aside in favour of analytics: changes in brand demand, surveys, correlations, geographic testing, marketing-mix modelling and the interpretation of several weak signals rather than one perfectly traceable journey.
Companies can therefore track their presence in ChatGPT, Gemini and other AI environments, but they should ask one question before turning that data into a KPI: what do we want to learn from it? If the objective is to understand reputation, share of voice or the brand’s ability to appear in certain purchase situations, the metric can already be useful. If it is supposed to prove direct return on investment, the standard of evidence needs to be much higher.
Visibility in AI is likely to become a genuine marketing issue because these tools are increasingly involved in discovery, comparison and decision-making. But the next major challenge may not simply be appearing in generated answers. It may be avoiding the temptation to turn that presence into another magic number that everyone follows because it exists, before anyone has established what it is actually worth.
We support companies that want to build a marketing strategy around indicators that are genuinely connected to commercial objectives.
Written by Julien Ricciarelli-Bonnal
14 August 2026
