AI citations are becoming the new share of voice metric

Three developments this week share the same fault line: the path from awareness to purchase is being restructured by AI at a pace that has outrun most brands' measurement infrastructure. Comscore data on AI discovery fragmentation, Digiday's reporting on the evolving commerce funnelfunnelThe customer journey from awareness to purchase, typically Awareness, Interest, Consideration, Decision, Action, with prospects narrowing at each stage.View full definition →, and Forrester's framing of the agentic enterprise shift all point to the same pressure point for CMOs. The tools used to measure marketing performance were designed for a world where each stage of the funnel left a trackable footprint. That world is receding.

ChatGPT is losing AI discovery share to Gemini and Claude

Digiday reported this week on Comscore data showing ChatGPT losing ground to Gemini and Claude as the primary interfaces through which consumers discover brands, products, and information via AI. More consequentially, AI citations are emerging as a distinct visibility metric that publishers and brands are beginning to track separately from organic search rankings.

For CMOs, this development has two immediate implications. First, the assumption that AI-driven discovery means ChatGPT exposure is already wrong. A brand optimised only for OpenAI's citation patterns may be invisible in Gemini's responses entirely, given that each model draws on different training data, weights authority differently, and surfaces sources through different logic. Second, and more structurally, citation volume in AI responses is becoming a proxy for share of voiceshare of voiceYour brand's share of total advertising or conversation volume in your category, measured against competitors over a defined period.View full definition → in a channel that carries no click data, no impressionimpressionThe total number of times an ad or piece of content is displayed, regardless of clicks. Each display counts as one impression, even to the same person.View full definition → counts, and no standard measurement currency yet.

The practical question is how to build content and authority signals that travel across models rather than being optimised for one.Understanding how content authority and structure influence AI citation patterns is no longer a technical SEOSEOSearch Engine Optimization: the practice of improving your pages' natural (unpaid) rankings in search engine results pages to attract more organic traffic.View full definition → conversation. It belongs on the CMO agenda. Watch whether Comscore formalises AI citation as a tracked metric category in its next methodology update. If it does, media agencies will reprice accordingly.

The commerce funnel is no longer a sequence you can measure from top to bottom

Digiday also covered a related shift this week: shoppers no longer need to be inside an owned-retailer environment to complete a purchase. New purchasing pathways are multiplying across social platforms, AI interfaces, and retail media inventory, which means the conversion event is detaching from the discovery event in ways that traditional last-click or even multi-touch models cannot reconstruct.

This is not a theoretical concern. When a consumer discovers a product through a Claude response, researches it through a Google AI Overview, and purchases it through a social commerce checkout, which touchpoint gets credit? The answer under most current attributionattributionA framework for assigning credit to the touchpoints that contributed to a conversion, so you can measure which channels and interactions actually drive results.View full definition → setups is the last one, which is almost certainly the least responsible for the decision.

The Digiday piece frames this as a retail media strategy problem, and it is, but the upstream question is measurement architecture. CMOs who are still running omnichannelomnichannelAn integrated approach connecting all customer touchpoints (physical, digital, mobile) into a seamless experience, with shared data and consistent context across channels.View full definition → programs without a unified measurement layer are flying partially blind.Getting attribution right across these fragmented purchase pathways requires moving past channel-level reporting toward incrementalityincrementalityThe share of results (sales, conversions, revenue) that only happened because of a marketing action, not what would have occurred anyway.View full definition → testing and media mix approaches that can account for touchpoints that leave no direct digital trail.

Nothing about this problem is new in principle. What is new is the speed at which AI-mediated discovery is adding touchpoints that are genuinely untrackable through conventional tagging.

Google's ad tech remedies land at exactly the wrong moment for Google

Digiday's ad tech briefing this week noted that regulators are forcing open Google's programmatic infrastructure at the precise moment when the platform battle has moved to AI. The remedies being imposed on Google's ad tech stack were designed to address a market structure that existed before AI began absorbing search intent directly.

For CMOs, the practical implication is that the regulatory relief competitors and advertisers might have expected from a more open programmatic market may not materialise in the way anticipated, because the value is shifting out of programmatic display and into AI-mediated discovery where Google, through Gemini, remains a primary player. The competitive battlefield changed during the litigation.

This is worth watching rather than acting on immediately. The ad tech remedy process will play out over months. What CMOs should note is that budget allocations built around a post-remedy programmatic resurgence may need to be revisited if AI interfaces continue capturing the discovery intent that programmatic display once converted.

The signal underneath: Forrester's agentic framing deserves more attention than it is getting

The development that received the least coverage relative to its implications this week was Forrester's characterisation of the agentic enterprise shift as the largest change in business software since cloud computing.

Forrester is not talking about chatbots or copilot features. The research describes business applications becoming agentic, meaning AI that initiates actions, manages workflows, and interacts with other systems without waiting for a human prompt. Applied to marketing, this means campaign optimisation, audience segmentationsegmentationDividing a market into distinct groups of customers who share similar needs, characteristics or behaviours, so each group can be served with a tailored approach.View full definition →, and bid management that operates continuously rather than in reporting cycles.

The measurement problem this creates is less discussed than the efficiency gain. When an AI agentAI agentSoftware that pursues a goal on its own: it plans steps, uses tools and takes actions with limited human input.View full definition → is both running a campaign and optimising it in real time, the attribution question becomes recursive: who or what made the decision that produced the outcome? CMOs who adopt agentic marketing infrastructure without updating their performance governance frameworks will find themselves unable to explain results to a CFO, let alone learn from them.

The Forrester framing is from an analyst firm with no direct commercial stake in the specific tools, which makes it a more reliable signal than the agentic SEO workflows being promoted by tool vendors this week, several of which (including content from Semrush, a vendor selling SEO and analytics tools) describe agentic AIagentic AIAgentic AI refers to AI systems that pursue goals autonomously by planning, taking actions through tools, and adapting based on results, with minimal step-by-step human direction.View full definition → as if citation volume automatically translates to traffic. Semrush's own study data, to their credit, flags that citations do not reliably drive clicks, though that finding sits awkwardly in content whose purpose is to sell citation-tracking tools.

The through-line across this week is that marketing's measurement infrastructure is about a cycle behind where the activity is happening. AI citations have no standard metric. Agentic campaigns have no clear attribution owner. The commerce funnel now includes touchpoints with no tag. CMOs who treat this as a tooling gap will keep buying new dashboards. The ones who treat it as a governance gap will redesign how performance is defined, owned, and reported before the next budget cycle.