The Death of the Click
The numbers are unequivocal:
| Metric | Evolution |
|---|---|
| CTR for top Google positions | -34.5% (AI Overviews) |
| AI-referred traffic | +357% in 1 year |
| Zero-click searches | 60% in 2024 |
| Traditional search volume | -25% projected by 2026 (Gartner) |
The paradigm is shifting. We no longer optimize for the click. We optimize for the citation.
When a user asks ChatGPT "What is the best Swiss White Shepherd breeder in France?", Google doesn't intervene. Neither do your meta descriptions. Nor your title tags.
What matters: being in the LLM's knowledge corpus. And if you're not there? You don't exist.
The Paradigm Shift: SEO vs GEO
Before diving into acronyms, let's understand the fundamental shift:
| Dimension | SEO (Human-Centric) | GEO (Agent-Centric) |
|---|---|---|
| Primary metric | Clicks, CTR, positions | Citations, mentions, attributions |
| Target | Google index (SERPs) | Context window (LLM) |
| Main tool | Keywords, backlinks | Knowledge graphs, entities |
| Desired outcome | Traffic to site | Automated purchase decision |
| Optimization | Title tags, meta descriptions | Schema.org, structured data |
| Authority proof | PageRank, Domain Authority | Verifiable E-E-A-T, trust signals |
| Timing | Indexation (days/weeks) | Real-time (continuous crawl) |
| Response format | Clickable link | Synthesized answer with source |
This isn't incremental evolution. It's a paradigm rupture.
In the SEO model, you optimize to be clicked. In the GEO model, you optimize to be cited as an authoritative source — even if the user never visits your site. This fundamental shift requires rethinking how we approach topical authority and E-E-A-T signals.
The 3 New Acronyms
AEO — Answer Engine Optimization
Microsoft's definition: "Optimizes content for AI agents and assistants (like Copilot or ChatGPT) so they can find, understand, and present answers effectively."
AEO focuses on data clarity.
What AEO requires:
- Precise structured data: No ambiguity, no vagueness
- Exhaustive product attributes: Every characteristic documented
- Real-time information: Data must be current
- Specificity: Generic is no longer enough
Concrete example (SEO vs AEO)
| Approach | Query |
|---|---|
| Classic SEO | "waterproof jacket" |
| AEO | "lightweight waterproof jacket, packable, with storage pocket, ventilated seams, reflective strips, 30D ripstop fabric" |
The AI agent needs machine-readable precision. It cannot guess what you haven't stated.
GEO — Generative Engine Optimization
Microsoft's definition: "Optimizes content for AI-generated search environments (like LLM engines) to make it discoverable, trustworthy, and authoritative."
GEO focuses on credibility.
What GEO requires:
- Authority signals: Press mentions, certifications, awards
- Verifiable social proof: Authentic reviews, sourced testimonials
- Explicit guarantees: Returns, warranties, commitments
- Distinctive editorial voice: A point of view, not a compilation
AEO vs GEO example
| Approach | Focus |
|---|---|
| AEO | "Swiss White Shepherd, LOF registered, MDR1 tested, hip/elbow dysplasia screening, sales contract, veterinary follow-up included" |
| GEO | "FCI World Champion 2021, 100+ international titles, SCC Breeding Commission member, recommended by the Société Centrale Canine, 4.9/5 on 87 verified reviews" |
AEO says what you are. GEO says why you should be trusted.
Vectorial Authority: Understanding LLM Logic
LLMs represent concepts in multi-dimensional vector space. Each word, entity, and concept occupies a position in this space.
Imagine an N-dimensional space where:
- The center of a concept (e.g., "quality dog breeding") is defined by the most representative entities
- Each site/brand/entity has a position relative to this center
- The closer you are to the conceptual center, the more authority you have
How to get closer to the center?
- Semantic consistency: Use exact domain vocabulary, not marketing synonyms
- Related entity density: Mention official organizations, recognized experts, major events
- Verifiable proof: Every claim should be traceable (dates, numbers, sources)
- Age and consistency: An entity with 10 years of coherent discourse has more "vectorial weight"
This is why a site that simply says "we are the best" will never have vectorial authority. It's connected to no knowledge graph nodes. It floats in semantic void. Proper entity registry management and deliberate connection to recognized entities becomes essential for LLM visibility.
ACO — Agentic Commerce Optimization
This is the most recent. And the most disruptive.
ACO optimizes for AI agents that make purchases.
Not humans who search. Agents who execute transactions.
The Protocols at Play
ACP (Agentic Commerce Protocol) — OpenAI + Stripe
- Standard for agent-to-business transactions
- The agent can negotiate, compare, purchase
- Already operational in ChatGPT plugins
UCP (Universal Commerce Protocol) — Google
- Expanded, interoperable version
- Integration with Google Shopping, Merchant Center
- Vision: a universal agent for all transactions
What ACO requires:
- Transactional APIs: The agent must be able to act, not just read
- Machine-readable pricing: Prices, availability, terms in JSON
- Verified identity: The agent must know who it's dealing with
- Automated fulfillment: From order to delivery
ACO Customs: Critical Files
ACO introduces new configuration files that AI agents consult before interacting with your site. These are the customs of agentic commerce.
/.well-known/ai-plugin.json
This file, popularized by ChatGPT plugins, describes your site's capabilities for AI agents:
{
"schema_version": "v1",
"name_for_human": "OWAG - Dog Breeders",
"name_for_model": "owag_breeders",
"description_for_human": "Find certified dog breeders in France",
"description_for_model": "API to search breeders by breed, region, certifications. Returns puppy availability, prices, health tests.",
"auth": { "type": "none" },
"api": {
"type": "openapi",
"url": "https://owag.fr/.well-known/openapi.yaml"
},
"capabilities": ["search", "compare", "contact"],
"logo_url": "https://owag.fr/logo.png",
"contact_email": "api@owag.fr"
} AI-Specific robots.txt Directives
The robots.txt evolves to include AI agent-specific directives:
# AI Agents - Access Authorized
User-agent: ChatGPT-User
Allow: /api/
Allow: /products/
Allow: /availability/
User-agent: PerplexityBot
Allow: /
User-agent: ClaudeBot
Allow: /
# Traditional Crawlers
User-agent: Googlebot
Allow: /
# Commerce Agents - API Access
User-agent: AgentCommerce
Allow: /api/v1/
Allow: /.well-known/ ACO Preparation Checklist
- Create /.well-known/ai-plugin.json describing your capabilities
- Document your API in OpenAPI/Swagger format
- Add robots.txt directives for AI agents
- Expose real-time data via JSON endpoints
- Implement lightweight authentication for verified agents
- Monitor agent requests in your logs
Schema.org: The New Lingua Franca
Microsoft explicitly recommends these Schema types:
| Type | Usage |
|---|---|
Product | Product characteristics |
Offer | Price, availability, conditions |
AggregateRating | Verifiable average rating |
Review | Individual reviews |
Brand | Brand identity |
ItemList | Collections, catalogs |
FAQ | Structured Q&A |
Organization | Entity behind the site |
Person | Authors, experts |
Structured data is no longer "nice to have". It's the condition for existence in the AI ecosystem.
A site without Schema.org in 2026 is like a site without HTML in 2000. Technically functional. Practically invisible.
Entity Interconnection
The power of modern Schema.org lies in interconnection via @id:
{
"@context": "https://schema.org",
"@graph": [
{
"@type": "Organization",
"@id": "https://owag.fr/#organization",
"name": "OWAG",
"founder": { "@id": "https://selimreggabi.com/#person" },
"award": ["FCI World Champion 2021", "100+ international titles"]
},
{
"@type": "Person",
"@id": "https://selimreggabi.com/#person",
"name": "Selim Reggabi",
"jobTitle": "Breeder & SEO Expert",
"memberOf": { "@id": "https://owag.fr/#organization" },
"award": ["FCI World Champion 2021"]
}
]
} Entities reference each other, creating a mini knowledge graph that LLMs can traverse. This interconnected approach is central to the Doctrine Mesh methodology, where every entity reinforces the authority of others within the same semantic network.
Why LLMs Are Already Crawling Us
OWAG.fr log analysis (January 2026):
| Bot | Frequency | Targeted Pages |
|---|---|---|
| ChatGPT-User | Daily | Breed pages, breeders |
| PerplexityBot | Daily | Health guides, articles |
| ClaudeBot | Weekly | Entire site |
| GoogleBot | Continuous | Entire site |
| Bingbot | Daily | Product pages |
| GPTBot | Weekly | Educational content |
LLMs no longer rely solely on their training corpus. They crawl in real-time. They verify information. They compare sources.
What the Logs Reveal
- Most crawled pages by AI: Those with the most structured data
- Frequency correlated with freshness: Updated pages get recrawled
- Crawl depth: AI follows internal links — strategic internal linking still matters
- Preference for structured data: Pages with JSON-LD are crawled 3x more
If your site isn't being crawled by these bots, ask yourself: why don't you deserve their attention?
The Citation Test
Here's the ultimate test of your AEO/GEO optimization:
- Open ChatGPT, Claude, or Perplexity
- Ask a question about your domain of expertise
- Observe: are you cited?
If not, your traditional SEO might be excellent. But you're invisible to 30% of information searches (and growing).
Our Result
For "best Swiss White Shepherd breeder France", OWAG is cited by ChatGPT and Perplexity. This isn't luck. It's 10 years of E-E-A-T construction + structured optimization.
Citation Test Methodology
For rigorous testing:
- Use multiple LLMs: ChatGPT, Claude, Perplexity, Gemini
- Vary formulations: Direct question, comparative, exploratory
- Test in incognito: Avoid personalization bias
- Document results: Date, model, exact prompt, response
- Repeat monthly: Authority evolves with model updates
What You Need to Do Now
Priority 1: Schema.org Audit
Verify every important page has:
OrganizationorLocalBusinesswith complete contact infoArticleorProductbased on content typePersonfor authors with qualificationsAggregateRatingif you have reviews
Priority 2: GEO Signals
Explicitly document:
- Your certifications and accreditations
- Your awards and recognitions
- Your press mentions and citations
- Your verifiable partnerships
Priority 3: AEO Precision
For each offer/product/service:
- List ALL relevant attributes
- Use precise vocabulary, not marketing speak
- Update in real-time
- Structure in JSON-LD
Priority 4: Prepare for ACO
If you sell something:
- Expose your prices in structured data
- Document your terms (returns, warranties)
- Create /.well-known/ai-plugin.json
- Prepare API endpoints (even simple ones)
- Identify automatable workflows
- Add agent directives to robots.txt
The Evolution of SEO: Timeline
| Period | Dominant Paradigm | Key Metric |
|---|---|---|
| 2000-2010 | Keywords | Keyword density |
| 2010-2015 | Links | PageRank, backlinks |
| 2015-2020 | Content | Time on page, engagement |
| 2020-2024 | E-E-A-T | Authority signals |
| 2024-2026 | AEO/GEO | AI citations |
| 2026+ | ACO | Agentic transactions |
Each transition hasn't killed previous paradigms. It has integrated and surpassed them. You still need keywords, links, content, E-E-A-T. But it's no longer sufficient.
Conclusion
SEO isn't dead. It's evolving.
The click paradigm (being clicked in SERPs) yields to the citation paradigm (being cited by AI).
The sites that will dominate tomorrow are those that:
- Exist in the knowledge graph — recognized entities, not anonymous pages
- Deserve to be cited — verifiable topical authority, not hollow claims
- Are machine-readable — structured data, not marketing prose
- Are accessible to agents — APIs, config files, real-time endpoints
Traditional SEO remains important. But it becomes a necessary, not sufficient condition.
AEO + GEO + ACO = the new triptych of digital visibility.
— Selim Reggabi, January 2026
Sources
- Microsoft Advertising Guide to AEO & GEO (PDF, August 2025)
- SearchEngineJournal: Agentic Commerce
- Ahrefs: AI Overview Impact Study (2025)
- Gartner: Future of Search Report (2024)
- OpenAI: Plugin Manifest Specification (2025)
- OWAG.fr server logs (internal analysis, January 2026)