AI Interfaces Now Own Discovery-Phase Attention — The Open Web Does Not
Executive Summary
Users now spend 75% of discovery-phase online time inside AI interfaces, consuming answers without loading a publisher page. This is not a performance problem with your display campaigns — it is the inventory base disappearing. The immediate move is auditing which brand knowledge assets currently earn AI citations, then restructuring the rest for selection by AI interfaces before Google, Perplexity, and OpenAI close the unmonetised window within 12–18 months.
The Signal
Open web dwell time has collapsed to 15 minutes per hour of online activity — meaning users now spend 75% of their online search and discovery time inside AI-driven interfaces rather than navigating to publisher pages. The shift reflects a structural change in information consumption: AI prompt interfaces deliver consolidated, sourced answers that eliminate the click-through journey traditional search monetisation depends on. This is not a trend line — it is an already-realised state. The open web, as an advertising surface, is no longer where attention lives during the discovery phase of the user journey.
What Changed
AI interfaces now serve as the primary discovery layer for a majority of online information-seeking — a function previously owned by search engines that routed traffic to the open web. What is new is not AI-assisted search, but AI as the terminal destination: users retrieve, evaluate, and act on information without ever surfacing on a publisher page. The open web's role in upper-funnel attention capture is structurally diminished, not cyclically weakened.
Why It Matters
The economics of open-web advertising are broken at the structural level, not the performance level — and that distinction matters enormously for how you respond. CPM-based display, pre-roll, and contextual advertising all presuppose that users travel to publisher pages. When 75% of discovery-phase attention is absorbed inside AI interfaces before a single page loads, the inventory those formats depend on simply does not exist in sufficient volume to sustain current spending levels. Publishers lose yield. DSPs lose match rates. The entire programmatic stack built on cookie-synced, page-load-triggered identifiers loses its activation surface. What becomes newly possible is direct-to-interface brand presence — the equivalent of owning shelf space inside the AI layer itself. Brands that structure their content, data, and authority signals to surface favorably inside AI-generated answers gain a discovery channel that currently operates outside the auction-based pricing floor. That is a temporary arbitrage, and the window is closing as platforms including Google, Perplexity, and OpenAI build monetisation layers on top of it. The deeper strategic pressure is this: AI interfaces consolidate trust. When a user receives a synthesised answer, the source citations that do appear carry disproportionate authority — because they survived curation. Brands that become those cited sources gain not just visibility but implicit endorsement from the interface itself. That is qualitatively different from a paid impression. The obsolete model is spending budget to intercept attention on the open web. The emerging model is structuring knowledge assets to be selected by AI as the authoritative answer. Those are fundamentally different marketing disciplines, and most organisations are still funded and staffed for the first one.
Marketing Impact
media
Open-web display and programmatic budgets lose their activation surface at the discovery phase. Media plans built around CPM-based reach on publisher inventory are now funding impressions that occur after the decision-relevant moment has already passed inside the AI interface — a structural mismatch, not a targeting problem.
martech
Cookie-synced, page-load-triggered identity infrastructure has no trigger event when discovery happens inside a closed AI interface. Audience activation, frequency capping, and attribution models built on URL-level signals are partially blind to the majority of upper-funnel activity, requiring a rearchitecture around first-party and declared-intent data signals.
research
Brand tracking and awareness measurement methodologies assume open-web exposure as a baseline condition. When 75% of discovery-phase attention runs through AI interfaces that produce no impression logs, conventional brand lift studies and reach-and-frequency models are measuring an incomplete — and increasingly unrepresentative — slice of actual consumer exposure.
product marketing
Positioning, claims architecture, and content structure must now be engineered to surface favorably inside AI-generated answers. The operative discipline shifts from SEO-optimised landing pages to knowledge-graph authority: structured data, cited sources, and entity disambiguation that make a brand the answer the AI selects rather than a result the user chooses.
The Exploit
Opportunity
Redirect 15–20% of open-web display spend toward an AI-visibility program: structured knowledge assets, authoritative long-form content, and entity-markup infrastructure designed to surface inside AI-generated answers on Google AI Overviews, Perplexity, and ChatGPT. Brands that own cited-source status inside AI interfaces capture discovery-phase attention at effectively zero CPM — a channel that auction pricing has not yet reached.
Risk
AI interface algorithms are opaque and unstable; content optimised for citation today can be deprioritised by a model update tomorrow. Redirecting display budget creates short-term reach gaps that performance metrics will flag immediately.
The Move
By October 2026, have your content and SEO teams audit which brand knowledge assets currently appear as AI citations across the top three interfaces, then restructure the bottom 80% using entity markup, primary-source data, and authoritative depth — measuring success by citation rate in AI answers, not page traffic. Own this in the head of content strategy, not paid media.
First-Mover Advantage
Gains
Brands that achieve cited-source status before monetisation layers activate inherit the trust signal of AI endorsement — a qualitative authority advantage that paid placement cannot replicate once bidding opens.
Risks
AI interface algorithms are opaque and unstable; content optimised for citation today can be deprioritised by a model update tomorrow. Redirecting display budget creates short-term reach gaps that performance metrics will flag immediately.
Window
The unmonetised window likely closes within 12–18 months — the signal is Google, Perplexity, or OpenAI announcing sponsored placement inside AI answer units, at which point organic citation becomes structurally harder to sustain.
Winners & Losers
Winners↑
AI-native content and knowledge teams
Organisations that structure content as structured data, authoritative Q&A, and citable knowledge assets are positioned to surface inside AI-generated answers — a discovery channel that currently carries no auction-based price floor. The mechanism is selection, not interruption: AI interfaces reward depth, sourcing quality, and topical authority, which means teams built around knowledge engineering rather than page-view optimisation gain compounding visibility advantages. The immediate move is auditing content libraries for AI-citability and reallocating editorial resource toward formats AI interfaces preferentially surface.
Authoritative vertical publishers with first-party data
The publishers who survive the discovery collapse are those whose content earns citation inside AI answers rather than relying on click-through volume — meaning deep subject-matter authority, strong editorial standards, and verifiable sourcing become the new yield drivers. The mechanism is curation: when AI interfaces select sources, high-trust vertical publishers appear disproportionately, converting their citation presence into audience re-engagement through direct relationships rather than search referral. These publishers should accelerate direct subscription and newsletter infrastructure to monetise the audience that does click through, as that cohort is now self-selected and higher intent.
Retail media networks with closed-loop purchase environments
Retail media networks operate inside walled gardens where the discovery-to-purchase journey is already self-contained, insulating them structurally from open-web attention collapse. As upper-funnel discovery migrates into AI interfaces and lower-funnel conversion migrates into commerce environments, the mid-funnel open web loses its connective role — and retail media networks capture the commercial intent that surfaces after AI-driven discovery concludes. The exploitation play is positioning retail media as the natural successor to performance display: commerce-adjacent, intent-rich, and immune to the page-load dependency that is breaking programmatic economics.
Losers↓
Open-web programmatic media buying teams
The entire activation model — cookie-synced identifiers, page-load triggers, contextual targeting against publisher content — presupposes that users navigate to publisher pages, and 75% of discovery-phase attention is now consumed before that navigation occurs. Match rates fall, addressable inventory shrinks, and CPM floors erode as remaining open-web inventory concentrates among a smaller, less representative user pool. The defensive response is accelerating reallocation toward closed-environment inventory — retail media, CTV, social platforms — while auditing how much current display spend is reaching audiences who no longer exist at that point in the funnel.
Publisher display and programmatic yield teams
Publishers dependent on programmatic display revenue face a structural demand collapse, not a cyclical softening — shrinking session volume means fewer impressions, lower fill rates, and deteriorating CPMs as DSPs register declining match rates against the user pools that matter. The mechanism is supply-side: when discovery-phase users never load a publisher page, the high-value moments that justified premium CPMs simply do not occur. The only viable defensive path is accelerating the transition to subscription revenue, owned audience monetisation, and direct brand partnership deals that price on authority and trust rather than impression volume.
Strategic Outlook
The next 18 months will separate organisations that treat this as a media efficiency problem from those that recognise it as a structural reallocation of where marketing capability needs to live. Expect accelerated consolidation among mid-tier publishers as programmatic yield collapses further — the same dynamic that gutted print classified revenue, compressed into a shorter cycle. Google will attempt to defend its monetisation position by deepening AI Overview ad integration, but this creates a ceiling on answer-layer inventory, not an open market. Perplexity and OpenAI will move faster to commercialise citation placement, effectively creating a new premium tier of brand presence that functions like sponsored shelf position inside a trusted reference. Brands that have already invested in structured content authority — detailed product data, third-party citations, entity-level credibility signals — will find that investment compounds rapidly as the window closes on unmonetised AI visibility. Those still debating the brief will pay auction prices to enter a market that early movers accessed for the cost of content production.