Artificial intelligence is changing the purchasing journey faster than the systems used by brands to measure it. The consumer can discover a product, compare it with alternatives, evaluate its characteristics and form a preference within a generative search engine or an assistant without necessarily visiting the website of any of the companies involved. For marketing, the problem posed by this behavior is considerable: How to attribute value to an influence that no longer always produces a click?

That is one of the main conclusions of “From Clicks to Influence: The New Economic Model of AI-Driven Commerce”, a study prepared by Emarketer in collaboration with Partnerize based on a survey carried out in April 2026 among 100 marketing managers in the United States, with director positions or higher and belonging to companies with at least $20 million in annual revenue.

The data show a transition that is still poorly resolved. 91% of marketers believe that AI-powered search and discovery is already changing the way their organizations approach marketing. However, measurement systems continue to largely respond to an architecture designed around traffic, clicks and observable conversions. The consequence is a growing distance between the places where influence occurs and those where companies are able to recognize it.

The consumer moves faster than the measurement

Traditional channels continue to concentrate a good part of brand discovery activity. 68% of the managers consulted point out social platforms among the channels they depend on to promote the discovery of products and services, and 53% mention traditional search engines. Results generated by AI already appear for 33%, while AI chatbots are still at 8%.

But consumer behavior points to a more profound change. According to forecasts collected by Emarketer, 106.2 million people will use generative search in the United States this year and 79.6 million will use generative AI for shopping-related activities, which represents an increase of 25%. Another study cited in the report indicates that 64% of AI users worldwide have already used these tools to compare products and 53% to discover a new brand.

Nate Elliott, principal analyst at Emarketer, also points out a phenomenon that expands the scope of this transformation: the user does not even need to deliberately modify their habits to expose themselves to responses generated by artificial intelligence. According to Advanced Web Ranking data collected in the report, more than 60% of Google commercial searches already include AI Overviews.

The click, therefore, loses part of the prominence that it has had for decades as a signal to reconstruct the consumer’s journey. Research, comparison and consideration can occur in environments that synthesize information from different sources before there is a direct visit to the brand.

There appears one of the most significant data of the study. Only 8% of organizations can track the end-to-end influence of AI discovery on revenue and conversions, while only 15% have a reliable way to measure the revenue generated by consumers who have been influenced by AI outside of their own brand environments.

The distance compared to the rest of the channels is notable: AI leads connected television by 21 percentage points and organic social networks by 23.
The explanation is related to the moment of the tour in which it acts. Elliott notes that while AI may seem like an evolution of search to many professionals, consumers are often turning to it at much earlier stages. This reduces precisely the signals that traditionally allowed performance to be measured: clicks, leads and conversions.

Uncertainty is already affecting confidence in the tools available. 54% of marketers believe that the rise of AI discovery has weakened their confidence in their current attribution model, compared to just 14% who disagree.

From the attribution problem to the economic problem

The question takes on another dimension when we look at who produces the information that AI systems draw on. Media, affiliation websites, creators, communities and other sources can participate in the construction of a purchasing decision even if the user never directly visits those contents. The report identifies a paradox here: the value of certain content to influence the consumer can increase at the same time that its ability to generate the traffic with which that influence was traditionally monetized decreases.

The last-click model is especially problematic in that scenario. A source may contribute to a brand appearing in a model recommendation, but the company’s attribution system may be unable to recognize that intervention.
And there is still a greater distance between recognizing it and remunerating it. Only 13% of marketers currently have a clear mechanism to connect AI visibility to partner compensation. Another 22% can detect influence, but lack a way to translate it into economic value.

This circumstance introduces a question that transcends analytics: if certain content helps a brand gain presence in AI systems, who should receive the economic value generated by that influence and under what criteria?

The budget begins to follow the influence

The industry is beginning to respond by reallocating resources. If traffic from traditional search engines declines as a result of AI discovery, 51% of marketers plan to shift budget toward content and SEO over the next two years.
38% would be allocated to performance-based partnerships linked to visibility or results; 23%, to influencer and creator marketing; 21%, to the brands’ own channels; 13%, to affiliation programs and partnerships; and 11%, to media commerce networks.

The movement responds to a new dynamic of discovery. If generative systems provide answers instead of simply offering links, a brand’s presence also depends on reliable content and external voices capable of generating authority and relevance around it.
Max Willens, principal analyst at Emarketer, notes that certain affiliate content sites routinely appear among the domains most cited by language models, while content generated by creators and users on established domains can quickly change a brand’s visibility in the eyes of AI.

The publishers themselves are reacting. According to another study included in the report, 37% of publishers and advertisers worldwide are expanding distribution through affiliation and partnerships to diversify their traffic sources in the face of the growth of AI and clickless discovery tools.

The research also raises a competitive derivative. Two-thirds of marketers express some concern that publishers or content partners may prioritize competitors that are better able to measure and reward click-free or AI-influenced impact.
The hypothesis is relevant because it transforms measurement into a possible incentive mechanism. If a publisher or creator knows that one company can financially recognize their contribution to visibility within generative systems and another cannot, that difference could end up conditioning where they dedicate their resources.

The attribution infrastructure would thus no longer be limited to explaining what happened after a campaign. It could also influence what relationships a brand maintains and, with them, its future ability to be part of the sources that fuel AI discovery.

The problem is that companies start from a fragmented infrastructure. 41% of the managers consulted use several systems without having a single source of truth to measure and report marketing performance. 24% mainly use internal data warehouses or business intelligence tools; 14%, attribution platforms; and 10%, web analytics tools. It is not surprising, therefore, that 73% plan to increase their investment in analytics or adopt new technology this year.

From measuring clicks to measuring influence

The report does not present a standardized solution – in fact, it recognizes that such standardization does not yet exist – but it does identify four priorities for companies:

• First audit your presence on the different AI surfaces
• Identify partners whose loss would pose a greater risk
• Experiment now with attribution models capable of covering the entire customer journey
• Test compensation mechanisms even if measurement is still imperfect

Among the compensation possibilities he points out are negotiated bonuses linked to AI appointment tracking, multi-touch models that give part of the credit to partners located in high stages of the funnel, or content licensing agreements that reward visibility itself and not just the click. The document emphasizes that none of these models is fully consolidated.

And the underlying question is broader than finding a new metric. For much of the digital age, marketing built its economic system around signals: an impression, a click, a visit, a lead or a sale. AI introduces a layer of intermediation capable of generating influence without necessarily producing any of them. And this displacement forces us to reconsider how its value is distributed. The attribution battle in the age of AI may actually end up being an argument about something much more fundamental: who contributed to a purchasing decision and who deserves to get paid for doing so.