The Next Scarce Thing in Advertising Might be a Human Being

By Beatrice Lhopitallier, Chief Revenue Officer, France at Utiq

There is a clear assumption underneath almost every digital advertising metric: that somewhere, eventually, there is a person. A page loads. Somebody clicks. We call it reach or engagement, and behind those words sits the same implied human. However, that assumption is getting weaker over time.

Bots have been part of the internet for years. What is changing is the variety of non-human activity around us. AI crawlers are consuming content at scale. Agents are starting to browse and act on behalf of people.

For advertising, this creates a a new paradigm: the existence of a new form of browsing delegated to an AI agent, which interacts with advertising but does not respond to the same stimuli as humans. If we cannot always tell whether an interaction came from a person, what exactly are we measuring?

We have always assumed there was a person there

Digital advertising has spent a lot of time debating whether an impression was viewable and whether a click was fraudulent. Both matter, although we have tended to take the human itself for granted. A browser produces signals and we infer a user. A click happens and, unless it is identified as invalid traffic, it may find its way into reporting.

AI makes those assumptions less comfortable.

DoubleVerify reported in late 2025 that AI bots accounted for up to 15% of clicks in unprotected media in some of its testing. That does not mean every one of those interactions was malicious. In fact, that is part of the point.

An agent comparing products for someone may be doing exactly what its user asked it to do. Neither is necessarily fraudulent. Neither is quite the same as a person seeing an advert either.

A bot does not have to be bad to change the value

There is a risk that the industry responds by putting all of this into the familiar bucket marked ‘fraud’. That would miss something important.

Fraudulent traffic of course needs to be removed. But an AI agent can change the meaning of a media metric without doing anything wrong, leaving the harder question as to whether an advertiser should measure and value that interaction in the same way as human attention.

Conversely, an agent interaction may actually have value, for example if it is researching a purchase for its user, where there is commercial intent somewhere behind it. This means the industry will need to find a way to account for these interactions, since they are ultimately linked to a real person; the interaction will be part of the AI agent’s response and will influence (perhaps even more so) the user’s purchasing behavior.

Human reach should become THE quality signal

We already have a language for media quality. We talk about attention, suitability, viewability and fraud. But in an internet where automated activity is growing, human authenticity may need to sit somewhere in that conversation. The reason is obvious when you strip away the technology. Marketers are ultimately trying to influence people.

A campaign can produce an impressive dashboard, while leaving lots of questions unanswered: how many real people did we actually reach? Was frequency controlled around a person or around devices? Were the audience signals deterministic or inferred?

At Utiq, our Authentic Audiences proposition is deliberately people-based: real people rather than impressions, using telco-powered, user-consented signals to support deterministic addressability across the premium open internet.

Utiq provides proof of a genuine human mandate and can be considered as a building block of trust for the digital market: proof that there is indeed a person behind a session, even if that session is being delegated to AI

That does not turn identity infrastructure into bot detection, and we should not pretend it does. But it points towards something advertisers may value more as the internet changes: a dependable connection between an audience signal and a real person.

This changes the meaning of identity

For years, the identity conversation has largely been about recognition – can we recognise someone and can we connect activity across browsers or devices? Those questions remain, but another one is joining them: what kind of actor are we recognising in the first place?

This becomes particularly interesting as AI moves further into both sides of advertising. Planning and buying will become more automated, while agents increasingly act for consumers. Machines will be making decisions about media while other machines navigate the environments where that media appears.

Somewhere inside all of that automation sits the person the advertiser ultimately wants to influence. Knowing when we have actually reached them becomes the valuable signal, as does controlling how often we reach them and measuring the outcome with some confidence.

The web can contain both people and agents. Advertising simply needs to get better at understanding the difference.

The most basic question may become the hardest one

There will be awkward edges to this. An AI agent acting for a consumer may be part of a genuinely valuable customer journey. New forms of measurement will probably emerge rather than forcing every machine interaction into today’s definitions.

But the direction feels clear.

As more activity becomes automated and AI-enabled – in a web increasingly populated by software  that acts, searches and clicks alongside us – certainty about human direct or indirect reach should become far more important. Whether a person clicks or sends an agent to click for them, knowing who’s really there isn’t a nice-to-have anymore. It’s the whole game.