Why the Shift From “Search” to “Ask” Changes Everything
Over the last twenty-five years, I have had the privilege of serving as a Chief Marketing Officer, Chief Data Officer, Chief Digital Officer, and Chief Innovation Officer. Across every one of those roles, one principle never changed: the consumer sat at the center of the experience.
Channels evolved. Technologies came and went. But the model remained human-centric.
About a decade ago, I started speaking about what I called the Proxy Web, a future state where humans would no longer sit at the center of every interaction. Instead, algorithms, personal assistants, and intelligent systems would increasingly act on our behalf. At the time, it felt conceptual. Abstract. Directionally interesting.
Fast forward to today, and that future has arrived.
In my current role as a Consumer AI leader at EY, working across hundreds of enterprise programs, I see a clear behavioral inflection point emerging across industries, categories, and geographies. Buying behavior is shifting. Discovery is shifting. Decision-making is shifting.
We are moving from searching to asking.
And as agentic commerce accelerates, this shift is not incremental. It is foundational.
From SEO to GEO: A Strategic Reframe
Most organizations still approach this moment through a familiar lens: Search Engine Optimization. Tweak the technical stack. Update metadata. Adjust content cadence.
That framing is already outdated.
What we need now is Generative Engine Optimization, not as a marketing tactic, but as an enterprise capability.
GEO is not simply about ranking in AI-generated answers. It is about how your organization is represented, interpreted, and trusted by large language models and autonomous agents acting on behalf of consumers, employees, and partners.
Here is the critical mistake I see repeatedly: treating GEO as a technical site-structure problem.
It is not.
GEO sits at the intersection of data governance, technical infrastructure, marketing and content strategy, public relations, and analytics. It requires a holistic re-examination of how an organization shows up in an AI-mediated world.
Let me break it down.
1. Internal Data and Governance: The Foundation
AI models do not know your brand. They infer it.
That inference is only as strong as the grounding data available to them. If your internal sources are fragmented, outdated, or contradictory, the AI’s synthesis of your brand will be as well.
Key considerations include:
- Entity Source of Truth
Identify the canonical source for core entities such as executives, product specifications, pricing, and positioning. Is there a single system feeding your website, LinkedIn presence, and public knowledge bases, or a patchwork of conflicting sources? - Hallucination Risk Audits
Proactively test how models like Gemini or ChatGPT describe your company today. When facts are wrong, trace the error back to the internal or public documentation creating that confusion. - Claim-to-Evidence Mapping
Every major marketing claim must be anchored to a public, verifiable source. If you say “number one in sustainability,” there needs to be a URL with data that an AI can confidently cite as proof. - Narrative Ownership (RACI)
Who owns the brand’s AI narrative? Increasingly, organizations are shifting this responsibility from SEO teams to Corporate Communications, and for good reason.
2. Technical Infrastructure: The Pipeline
Traditional search crawlers follow links.
Generative engines extract meaning.
That distinction matters.
- llms.txt Implementation
An llms.txt file provides a clean, markdown-based map designed specifically for AI crawlers. If you have not built one, you are leaving interpretation to chance. - Schema Coverage and Depth
JSON-LD for Organization, Product, FAQPage, and Person is table stakes. Completeness matters. SameAs links to social profiles, third-party mentions, and authoritative sources increase confidence. - Information Gain
AI engines deprioritize derivative content. If your material simply rehashes what already ranks elsewhere, it will be ignored. Original data, frameworks, and insights win. - Crawlability versus Renderability
Critical information buried behind heavy JavaScript may never be seen by high-speed AI crawlers. Accessibility is no longer just a UX concern. It is an AI visibility issue.
3. Marketing and Content Enablement: The Output
Content strategy must evolve from keyword-centric to question-centric.
AI systems are optimizing for answers, not pages.
- Answer Nugget Density (My personal favorite AI term)
High-value pages should lead with a concise forty to eighty word direct answer. If the AI needs to scroll to understand your value, you have already lost. - Interrogative Headings
Replace generic H2s like “Our Process” with prompts users actually ask, such as “How does Brand integrate with SAP?” - Comparison Frameworks
Structured tables for pros and cons and feature comparisons outperform dense narrative text in AI extraction workflows. - Multimodal Metadata
Images and videos need context-rich tagging. Gemini and other multimodal systems reason across formats, but only if you give them something to work with.
4. PR and External Grounding: The Authority Layer
In a GEO world, mentions matter more than links.
AI engines place disproportionate trust in high-authority third-party validation.
- Earned Media Share of Voice
Track how often your brand appears in best-of lists and category roundups, not just backlinks. - Digital PR Seeding
Target publications that serve as trusted training data sources for large language models, including major business media, industry analysts, and research firms. - Wikipedia and Wikidata Health
Neutral, factual, and well-cited entries are not optional. They are foundational grounding sources for most major models. - Executive Footprint
Subject-matter experts should have visible, verified presences across knowledge graphs and professional platforms. Thought leadership is now machine-read as well as human-read.
5. Measurement and Analytics: The Feedback Loop
You cannot manage what you do not measure, and click-through rate is no longer the metric that matters most.
In a zero-click AI world, new signals emerge:
- Citation Rate
How often does your brand appear as a cited source in relevant AI responses? - Narrative Alignment
When an AI summarizes your company, does the tone and value proposition align with your actual go-to-market strategy, or is it drifting? - Share of Model
How frequently are you mentioned versus competitors in unbranded category discovery prompts?
Universal Commerce Protocol (UCP)
Another signal I am watching closely is Google’s move toward a Universal Commerce Protocol and the introduction of Merchant Center brand agents. This is not a feature release. It is a directional statement about where commerce is headed.
Google is positioning AI not as a discovery layer that points users to brands, but as an active participant in the buying journey. As that happens, brands are no longer discovered solely through pages and links. They are represented through structured identities, trusted data feeds, and governed product truths that AI systems can reason over and act on.
Merchant Center brand agents make this explicit. Pricing, availability, policies, attributes, and fulfillment signals are becoming first-class inputs into AI decisioning. In that world,
Generative Engine Optimization is not about visibility. It is about readiness. GEO becomes the discipline that ensures your brand is accurately understood, confidently recommended, and correctly transacted by AI systems acting on behalf of consumers.
The Bottom Line
Generative Engine Optimization is not a marketing optimization exercise. It is an enterprise readiness challenge.
As AI agents increasingly act as intermediaries between brands and buyers, organizations must shift from optimizing for clicks to optimizing for trust, clarity, and grounding.
The companies that win will not be the ones chasing algorithms.
They will be the ones architecting their data, narratives, and authority for an AI-mediated future.
The proxy web is no longer coming.
It is here.