Advertisers are beginning to target AI agents instead of only human audiences — but the industry still lacks a clean way to prove whether a sponsored message actually changed what the machine recommended or what the person eventually bought.

WHAT’S HAPPENING

A new advertising market is forming around AI agents.

Instead of placing an ad only in front of a person browsing a website, startups are experimenting with ways to place sponsored information where AI systems can retrieve, interpret and potentially incorporate it into their answers.

That creates a measurement problem.

Traditional digital advertising can often follow a relatively visible path:

Person sees ad → clicks ad → visits site → buys product.

Agentic advertising introduces more steps:

AI retrieves sponsored information → AI evaluates it → AI generates a recommendation → Human sees the recommendation → Human or AI completes the purchase.

Companies including OpenAds and Oasy are experimenting with referral codes, attribution systems and performance-based models intended to connect AI-generated recommendations with downstream actions. But the market still does not have a universally accepted way to determine whether a specific advertisement actually influenced an AI system’s recommendation.

The Interactive Advertising Bureau is already working on measurement and technical standards for the emerging agentic-advertising market, including new approaches for measuring AI visibility and infrastructure for agent-driven advertising.

WHY IT MATTERS

Digital advertising spent decades learning how to measure human attention.

AI agents could change the audience.

If people increasingly ask AI systems to research products, compare companies and recommend what to buy, advertisers may try to influence the software making those comparisons before the human ever sees an advertisement.

But measuring that influence is much harder than measuring a click.

If an AI recommends one product over another, advertisers need to know why.

Was it the sponsored information?

Independent reviews?

Pricing?

Previous training data?

Another website?

The AI’s own ranking process?

Without reliable attribution, advertisers cannot easily determine whether they are paying for meaningful influence or simply paying to appear somewhere inside an AI’s information stream.

WHO BENEFITS

Publishers could gain a new source of revenue by making sponsored information available to AI agents.

Advertising technology companies could build an entirely new measurement industry around AI visibility, referrals and machine-generated recommendations.

Brands that learn how to structure accurate, useful and machine-readable information could also gain an advantage as AI systems become more important in product discovery.

And consumers could benefit if AI agents make comparison shopping faster and more useful.

But only if commercial influence remains transparent.

WHO LOSES

Traditional advertising models built around impressions, page views and clicks could lose importance if fewer people visit websites directly.

Publishers could also lose leverage if AI systems consume their information without sending meaningful traffic back.

Advertisers face another risk: paying for exposure they cannot prove affected anything.

The larger concern belongs to consumers.

An advertisement shown to a human usually looks like advertising.

Sponsored information influencing an AI recommendation may be much harder to see.

If an AI assistant tells someone:

“This is the best product for you,”

the user may interpret that as independent analysis even if commercial information helped shape the recommendation.

That makes disclosure increasingly important.

WHAT HAPPENS NEXT

The advertising industry will probably try to create a new attribution chain built specifically for AI.

Instead of measuring only:

Impressions → Clicks → Conversions

the market may begin measuring:

AI access → AI citation → AI recommendation → Human action → Transaction

Standards organizations are already moving in that direction, but the difficult question will remain:

How do you prove that an advertisement influenced the machine rather than simply appearing somewhere in the information the machine considered?

That problem becomes even more important as AI agents begin doing more than recommending products.

Eventually they may compare prices, negotiate purchases and complete transactions themselves.

At that point, advertisers will no longer be competing only for human attention.

They will be competing to influence the intelligence acting on the human’s behalf.

And the advertising industry’s next measurement problem will be simple to state — but difficult to solve:

Prove the machine was influenced.