---
id: "claim-marketing-new-audience"
type: "claim"
source_timestamps: ["§ Shift 3: Marketing Has a New Audience"]
tags: ["marketing", "google-overviews", "algorithms"]
related: ["concept-algorithmic-audience", "entity-product-insight", "entity-google-overviews", "quote-first-customer-algorithm"]
confidence: "high"
testable: true
speakers: ["Graham Kenny", "Ganna Pogrebna"]
sources: ["geo"]
sourceVaultSlug: "hbr-seg-geo"
originDay: 3
articleStem: "hbr-ext-11-llms-overtaking-search"
sourceUrl: "https://hbr.org/2026/03/llms-are-overtaking-search-heres-how-to-adjust-your-online-presence"
sourceTitle: "LLMs Are Overtaking Search. Here’s How to Adjust Your Online Presence."
---
# AI Overviews Have Changed the Fundamental Role of Marketing

**Claim (confidence: high; testable):** The arrival of AI Overviews has done more than alter search results — it has changed the *role* of marketing itself.

**Stated evidence (preserve exactly):** Because AI synthesizes recommendations directly, [[entity-google-overviews]] appear for **78% of core product queries** for affiliate site [[entity-product-insight]], causing a **67% traffic decline on historically high-value pages**. Brands must now persuade the **algorithms that mediate** interactions, not just the human end-users. Marketing is no longer *solely* about influencing human perception — see the [[concept-algorithmic-audience]] concept and its defining quote [[quote-first-customer-algorithm]].

**External grounding + caveat (enrichment):**
- **Supported:** McKinsey explicitly tells brands to add **GEO** alongside SEO and treat AI platforms as crucial discovery touchpoints ('improve visibility and sentiment on AI summaries and platforms'). Semrush recommends AI-visibility toolkits and treating LLMs as a distinct channel. Agentic-SEO practitioners argue E-E-A-T signals, machine-readable facts, and `/ai.json` endpoints are prerequisites to appearing in LLM answers.
- **Caveat:** 'First customer is the algorithm' is a useful lens but an over-generalization — McKinsey still frames the algorithm as an *intermediary/front door*, with humans as the ultimate audience. Product Insight's exact 78%/67% figures are case-specific, not published cross-industry averages.