How AI Is Shifting Value in Advertising—from Attention to Transactions

The rapid adoption of artificial intelligence (AI) by consumers and businesses is acting as a growth engine, a source of discontinuity, and an accelerant for shifts already underway across the advertising ecosystem. McKinsey’s "AI in Advertising Survey," fielded in February 2026, collected 182 responses from US-based advertising agency and marketing leaders — including C-level executives, presidents, vice presidents, and directors — at companies reporting annual marketing spend from $5 million to more than $5 billion. Three-quarters of respondents expect AI to increase total media spend, and one-third believe it will drive at least a 10 percent increase in return on ad spend (ROAS).

At the same time, spending may continue to move away from the open web as direct media deals that bypass agencies and programmatic intermediaries gain share.

The old model and how it’s changing

For the past two decades, digital advertising largely operated on a simple model: brands paid to access consumer attention via search results, webpages, or ad slots, while a growing set of tech and data intermediaries connected buyers of attention to sellers. That system assumed a segmented consumer funnel, human decision-making, and fragmented data.

Now that model is shifting. More consumers use AI tools to research and decide what to buy, and over time agentic purchasing decisions may occur with limited active human input. Advertisers already sense the change: more than 50 percent report AI has reshaped discovery and consideration.

Success today is increasingly about being surfaced, recommended, and selected by the systems that shape what consumers see and buy — not only about securing data-driven impressions with brand-controlled creative. Drawing on proprietary research, a survey of advertising decision-makers, and expert interviews, this analysis examines where value is moving across the ecosystem and how leaders can reposition. Because the technology and its effects evolve rapidly, the perspective reflects the market as of June 2026.

Three dominant ways AI is transforming the ecosystem

1. Discovery is shifting from the open web into AI-driven environments

• Over 50 percent of Google searches now include an AI-generated overview, and 20–50 percent of open-web search traffic could be at risk by 2030.
• The larger shift is occurring inside walled gardens: social and tech platforms that combine discovery, buying, measurement, audience data, and transaction visibility. On major integrated platforms, AI ranking now shapes most consumer attention: more than 50 percent of content viewed on Instagram and over 95 percent of watch time on TikTok comes from algorithmic feeds rather than followed accounts.
• Meta reported AI-driven recommendations increased time spent on Facebook by 5 percent in Q3 2025 and U.S. Instagram Reels watch time by more than 30 percent year over year.

2. Two distinct shifts around the ad transaction

• On the advertiser side, more than 90 percent of advertisers report using AI to plan media, set budgets, optimize targeting, and generate creative.
• On the consumer side, shopping agents and AI assistants rank products, compare alternatives, and sometimes complete transactions. Selected early deployments of AI-enabled shopping and recommendation experiences have delivered up to 60 percent higher conversion rates.
• Over the next few years, 10–35 percent of e-commerce transactions could be initiated, influenced, or completed through AI-native experiences, though the balance between standalone shopping agents and retailer-integrated AI tools remains uncertain.

3. Value concentrating in AI-native platforms and walled gardens

As attention, decision-making, buying, conversion, and measurement migrate to integrated environments, value may concentrate among AI-native platforms and walled gardens that bundle these capabilities into a single AI-managed product, reducing the role of intermediaries (exchanges, ad networks, traditional agencies) that historically linked buyers and signals separately.

This shift could increase advertisers’ dependence on platform-provided data and make measurement more opaque; 42 percent of advertisers in the survey cite reliance on “black box optimization” systems as a key risk to their media investment strategies. Advertisers may also lose visibility and control over where ads run across products, placements, and channels within a single platform.

What the survey shows about current reallocation of spend

• AI is expanding total ad spend: nearly three-quarters of companies surveyed expect total media spend to increase in the next 12 months, driven by new use cases (such as AI shopping agents) and AI-enabled ROAS gains. One-third of advertisers expect ROAS to increase by more than 10 percent because of AI.

• Hirdetők AI-driven formatsba fektetnek (Advertisers investing in AI-driven formats): more than 50 percent report investing in ads embedded in AI-generated answers as brands test formats in AI-assisted commerce and recommendation-driven environments. Much of this spend is reallocated rather than fully incremental: roughly 40 percent of reallocated AI-driven spend shifts away from traditional search and the open web toward platforms offering stronger data, measurement, and transaction capabilities.

• Direct buying is getting more attractive: direct ad buying nearly tripled between 2019 and early 2024, and AI is likely to reinforce that trend as discovery, buying, and measurement integrate and advertisers increasingly bypass agencies and programmatic intermediaries to work directly with walled gardens and AI-native platforms.

Four emerging trajectories and a shared challenge

The authors outline four possible future trajectories — from a base case to increasingly disruptive outcomes — that differ in speed and concentration of value. Across all scenarios the central challenge is positioning for outcomes where control over data, decision-making, and transactions determines advantage.

Legacy models based on traffic, manual execution, and fragmented measurement are under pressure, while platforms and ecosystems with stronger data, transaction visibility, and integrated measurement capabilities are capturing growing value.

How ecosystem players are exposed and how they can adapt

• Agencies: from execution to orchestration—or disintermediation

  • Agencies are highly exposed: roughly half of media spend now flows through direct channels, and core agency tasks (planning, buying, reporting, creative production) are precisely the knowledge work AI can disrupt quickly.
  • Agencies shifting from execution to orchestration help clients design AI-enabled workflows, govern automated decisions, interpret platform signals, and prove outcomes. Early leaders invest in proprietary AI and data assets, governance offerings (brand safety, AI-bias, compliance), and outcome-linked pricing tied to verified business results.

• Ad tech: from intermediation to specialized infrastructure

  • Ad tech’s middle layer is compressing as AI-native buying agents can replicate cross-platform optimization in a single workflow, potentially eliminating use cases for certain SSPs, exchanges, and undifferentiated middleware. As AI automates buying, optimization, and measurement — and 82 percent of advertisers plan to buy AI ad formats directly in the next 12 months — intermediation layers based on cross-platform arbitrage shrink.
  • Ad tech can remain relevant by serving as connective tissue across walled gardens and AI platforms, offering neutral cross-platform identity and measurement, independent AI performance audits, validation services, and secure clean-room partnerships.

• Publishers: from scale to survival of the most differentiated

  • Declining search traffic and the rise of AI-generated answers accelerate the erosion of open-web economics. Click-driven, scale-dependent publishing may become untenable; long-tail, SEO-dependent sites without subscription models are most exposed.
  • Publishers’ credible paths include licensing content to AI platforms for training and answer data, building generative engine optimization (GEO) capabilities to ensure visibility inside AI answers, and pooling authenticated audiences and premium inventory into curated ecosystems.

• Commerce media networks (CMNs): from inventory to decision advantage

  • CMNs sit closest to transactions; AI strengthens their position by enabling closed-loop measurement that tracks the customer journey and purchase visibility. As AI assistants collapse the shopping experience, the most valuable placement may be inclusion in an automated answer, agent-generated comparison set, or default selection based on structured product data and measurable outcomes rather than bidding alone.
  • CMNs are expanding omnichannel offerings, operationalizing conversational shopping experiences, agent-readable product feeds, and closed-loop measurement tied to online and in-store purchases, increasingly pricing against verified sales lift rather than impressions.

• Walled gardens: from media platforms to decision platforms

  • Established walled gardens are well positioned to extend beyond attention monetization into commerce and agent offerings. TikTok is expected to clear roughly $30 billion in commerce gross merchandise value by 2028; YouTube has made commerce a top priority; and Meta is integrating AI agents across its apps. How aggressively these platforms expand will shape the ecosystem’s terms.

• AI-native platforms: from interface advantage to advertising model

  • AI-native platforms tied to large language models (LLMs) already capture a growing share of consumer discovery: product research increasingly happens within ChatGPT, Gemini, Perplexity, and Claude alongside traditional search engines. Their conversational interfaces and agent layers give them privileged access to decision moments and rich behavioral data.
  • What they lack is a mature advertising model. Sponsored answers, agent-readable claims, affiliate commerce, and bidding into recommendations are all being tested. Platforms face a strategic choice: build walled gardens, partner with CMNs and publishers, or remain neutral interfaces optimized by brands via GEO.

Early moves — such as OpenAI’s experiments with commerce experiences and monetizing high-intent interactions — indicate AI platforms are exploring a more direct role in commercial journeys, but whether this evolves into integrated transactions and end-to-end measurement remains an open question.

Four actions leaders can take now

Leaders can focus on: assessing exposure across multiple possible futures; identifying expanding versus narrowing value pools and reallocating investment toward proprietary data, direct relationships, closed-loop measurement, and AI-enabled commerce; prioritizing assets that are hard to copy; and building capabilities quickly through partnerships, simplified workflows, interoperable measurement and data infrastructure, and reorganized teams around strategy, optimization, and AI-enabled decision support.

Across players, the same levers — data, products and services; business model; talent; governance and trust; partnerships; and customer focus — will determine competitive advantage. Specific starting actions differ by role: agencies must decide what to productize and how to price outcomes; ad tech must choose where to invest in neutral infrastructure; publishers must pick a differentiation path; CMNs must decide whether to compete for impressions or agent eligibility; walled gardens must determine how far to extend into commerce and agents; and AI-native platforms must decide what kind of advertising business they want to be, if any.

Conclusion

As AI becomes more embedded in media buying, advertising grows more automated, integrated, and concentrated among fewer walled gardens and AI-native platforms. The central challenge for leaders is retaining influence within the systems shaping discovery, recommendation, and transactions. Companies with stronger data, measurement, and transaction capabilities are likely to capture a disproportionate share of value, while others may compete on compressed margins inside ecosystems they do not control. The key question is who will set the rules for the next AI-driven advertising model and who will have to play by rules set by others.

Authors and acknowledgements

Jack Trotter is a partner in McKinsey’s Denver office; Marc Brodherson is a senior partner in the New York office, where Quentin George is a distinguished partner; and Aparna Srinath is an associate partner. The authors thank Arianna Sanchez, Gabriel Borsuk, Kaitlyn Maher, Shreya Jaggi, and Veronica Retana for contributions. The article was edited by Daniel Eisenberg, an executive editor in the New York office.