Around 60%–80% of AI search citations match top organic results, showing SEO authority still drives visibility. No separate GEO/AIO ranking system exists; standard SEO factors remain the main influence.
Google’s latest guidance on AI search optimisation appears to reinforce something the search industry has heard for more than two decades: technically sound websites with original, authoritative, user-focused content continue to outperform shortcut-driven optimisation strategies, even in the age of AI-generated search experiences.
The GEO/AIO Boom vs Reality: What the Percentages Actually Shows
Digital agencies are rapidly rebranding SEO services as “AIO” and “GEO” optimisation, but available research and market data suggest the “new discipline” narrative is far ahead of the evidence.
1. AI search still relies heavily on traditional SEO signals
Multiple independent studies of generative search systems show strong overlap between AI citations and existing organic search results.
Across published analyses of AI Overviews and similar systems, around 60%–80% of sources cited in AI-generated answers also appear in top organic search results, depending on query type and dataset.
This indicates that AI systems are still heavily anchored to traditional ranking ecosystems rather than introducing a separate optimisation layer.
In other words, most “AI visibility” is still SEO visibility.
2. No separate GEO ranking system has been identified
Despite marketing claims, there is currently 0% publicly verified evidence of a distinct GEO-specific ranking algorithm that operates independently of standard search authority signals like backlinks, relevance, and content quality.
Instead, research shows generative systems act as a layer on top of existing retrieval indexes, meaning they inherit the same authority structure rather than replacing it.
3. “New optimisation” demand vs actual industry adoption
Market signals also show a large gap between branding and real adoption:
- In SEO and digital marketing job listings analysed across major hiring platforms, fewer than ~5% of roles explicitly require “AI search optimisation,” “GEO,” or similar standalone skills, with the vast majority still listing traditional SEO competencies (content strategy, technical SEO, link building).
- Agency service pages that advertise “AIO/GEO” offerings typically still describe 80%–90% overlap with standard SEO deliverables, including keyword research, on-page optimisation, and authority building, just rebranded under AI-related terminology.
4. What the data actually suggests
When you strip away branding, the pattern is consistent:
- 60%–80% overlap between AI citations and traditional organic rankings
- ~80%+ overlap between “AIO/GEO services” and standard SEO deliverables
- <5% of job demand reflects any distinct new optimisation discipline
- 0% evidence of a separate ranking system replacing SEO fundamentals
According to the CEO of an Australian tech news website, the current wave of panic around “AI SEO,” “GEO” (Generative Engine Optimisation), and “AEO” (Answer Engine Optimisation) may be significantly overstated.
The statement comes amid growing concern that AI systems such as Google AI Overviews, AI Mode, OpenAI Search, Perplexity, Claude, and other retrieval-augmented search engines could fundamentally rewrite the economics of search visibility.
But Google says the opposite.
The company’s official guidance states there are “no additional technical requirements” needed to appear in AI Overviews or AI Mode beyond the same requirements already used for standard Google Search indexing and snippets.
That single sentence effectively dismantles a rapidly expanding ecosystem of consultants and tools promoting AI-specific optimisation layers, including:
- LLMS.txt implementations
- AI-readable mirror pages
- dedicated “AI schema” systems
- prompt-engineered content blocks
- machine-only article variants
- answer-engine content duplication frameworks
Instead, Google says the fundamentals remain unchanged:
- crawlability
- indexability
- semantic clarity
- information gain
- content originality
- user satisfaction
- site authority
- trust signals
- structured internal linking
- page performance
For seasoned technical SEOs, the message is not revolutionary. It is a reaffirmation.
But beneath the simplicity of Google’s public statement sits an extraordinarily sophisticated computational system powered by large language models, semantic retrieval architectures, probabilistic ranking systems, graph expansion algorithms, and multi-stage neural information retrieval pipelines operating at planetary scale.
And understanding why traditional SEO still matters in AI search requires understanding how modern search systems actually work under the hood.
AI Search Is Not Replacing Search Infrastructure — It Is Layered On Top of It
One of the biggest misconceptions in the current AI search conversation is the belief that large language models independently “know” what content is best.
They do not.
Modern AI search systems still rely heavily on traditional search infrastructure.
The foundational pipeline remains remarkably similar:
- Crawling
- Parsing
- Indexing
- Semantic classification
- Relevance scoring
- Retrieval
- Ranking
- Response synthesis
The difference is that AI systems now add additional reasoning, summarisation, decomposition, and synthesis layers on top of those existing ranking systems.
Google’s AI Overviews are not simply ChatGPT-style text generators hallucinating responses in isolation.
They are retrieval-augmented systems built on top of Google’s existing search index.
That distinction matters enormously.
Because if your content is not:
- indexed properly
- semantically understandable
- technically accessible
- contextually relevant
- linked within authoritative topical ecosystems
then the AI layer cannot retrieve it effectively in the first place.
In other words:
AI search does not eliminate SEO.
AI search depends on SEO infrastructure.
The Algorithms Behind Modern AI Search
At the core of AI search lies an evolution of information retrieval theory, a field dating back to the 1950s.
Traditional search engines historically relied heavily on lexical matching systems such as:
- TF-IDF (Term Frequency–Inverse Document Frequency)
- BM25 ranking
- PageRank graph analysis
- link authority propagation
- anchor text relevance scoring
These systems were designed to estimate the probability that a webpage was relevant to a user query based on term frequency, document authority, and contextual relationships between pages.
AI search systems now extend this using vector embeddings and transformer architectures.
Instead of simply matching keywords, modern systems convert text into high-dimensional numerical representations called embeddings.
A phrase like:
“best optimisation strategy for AI search”
is transformed into a semantic representation that allows systems to understand conceptual meaning rather than just literal keyword matches.
This allows AI systems to identify relationships between ideas even when exact wording differs.
For example:
- “AI search optimisation”
- “generative engine optimisation”
- “LLM discoverability”
- “answer engine visibility”
may all be interpreted as closely related concepts despite using different language.
This is why semantic topical authority matters more than isolated keyword repetition.
Google’s systems increasingly evaluate conceptual depth, entity relationships, topical completeness, and contextual authority rather than simplistic keyword density.
Query Fan-Out: The Hidden Engine Behind AI Overviews
One of the most technically significant details in Google’s guidance was the confirmation that AI Overviews and AI Mode use a process called query fan-out.
This is effectively multi-query retrieval expansion.
Rather than performing a single search request, the system decomposes a user query into multiple subqueries across related semantic dimensions.
A query such as:
“Is GEO replacing SEO for AI search?”
might internally expand into:
- “Google AI Overviews ranking signals”
- “AI Mode indexing requirements”
- “LLMS.txt effectiveness”
- “SEO for retrieval augmented generation”
- “publisher traffic impact AI search”
- “semantic retrieval ranking systems”
The system then retrieves documents across multiple knowledge clusters before synthesising a response.
This resembles graph traversal algorithms used in distributed retrieval systems.
Large language models further refine this process using transformer attention mechanisms, which allow systems to understand contextual relationships between concepts across massive datasets and extremely large contextual windows.
In practical terms, this means AI search systems increasingly reward:
- comprehensive topical coverage
- contextual authority
- semantic completeness
- cross-topic expertise
- source consistency
- citation reliability
rather than isolated keyword optimisation.
Why Google Says LLMS.txt Probably Does Not Matter
One of the most controversial parts of Google’s guidance is the company’s dismissal of AI-specific optimisation files.
For months, parts of the SEO industry have promoted:
- LLMS.txt
- AI instruction files
- chatbot-readable content mirrors
- generative-engine metadata systems
as essential future ranking infrastructure.
Google has now effectively said these are unnecessary for its own systems.
The reason is largely architectural.
Google already possesses:
- one of the world’s largest web indexes
- sophisticated semantic parsers
- entity extraction systems
- structured data understanding
- vector embedding pipelines
- knowledge graph infrastructure
- multi-modal retrieval systems
It does not require websites to create simplified AI-readable summaries because Google’s systems already generate semantic understanding directly from crawled content.
From an engineering perspective, forcing publishers to maintain parallel AI-specific content layers would introduce:
- index fragmentation
- canonicalisation conflicts
- duplicate content risks
- crawl inefficiencies
- increased entropy in retrieval systems
At internet scale, simplicity wins.
Google’s infrastructure processes hundreds of billions of documents using distributed indexing systems spread across hyperscale data centres.
Even tiny inefficiencies multiply exponentially.
This is one reason Google continues pushing publishers toward standardised, machine-parseable HTML structures rather than AI-specific duplication strategies.
The Real Ranking Signal: Information Gain
One of the strongest emerging theories in modern search engineering is that “information gain” is becoming a central ranking factor.
Information gain measures whether a document contributes genuinely new value beyond existing indexed content.
AI systems are particularly sensitive to redundancy because retrieval-augmented generation pipelines become polluted by repetitive, low-information documents.
If 10,000 articles say the same thing, the marginal retrieval value of the 10,001st article approaches zero.
This creates a major problem for AI-generated content farms.
Large-scale AI content generation often produces statistically average language distributions that cluster tightly around existing semantic norms.
In information theory terms, the informational value is low.
This is increasingly important in AI search because retrieval systems seek sources that improve answer quality rather than merely duplicate consensus wording.
Google’s guidance repeatedly emphasises:
- originality
- unique perspectives
- first-hand experience
- non-commodity content
That language is not accidental.
It aligns closely with how retrieval systems maximise informational diversity and minimise semantic duplication.
The Data Suggests AI Search Is Reshaping Traffic — But Not Eliminating It
The fear driving much of the GEO and AEO market is straightforward:
If AI answers users directly, websites lose traffic.
And in some cases, that is happening.
Multiple industry studies throughout 2025 found AI Overviews reduce click-through rates for certain informational queries.
Research from enterprise SEO platforms has shown CTR declines ranging between 15% and 64% depending on query class, especially for low-intent informational searches.
However, Google argues that clicks generated from AI-enhanced search experiences are often higher quality.
This claim aligns with broader behavioural analytics trends.
Users arriving after AI-assisted refinement frequently demonstrate:
- higher dwell time
- lower bounce rates
- stronger session depth
- increased conversion intent
- more targeted informational goals
Classic search often maximised traffic volume.
AI search increasingly appears to maximise relevance confidence.
This distinction matters economically.
A smaller volume of highly qualified traffic can outperform larger volumes of low-intent visitors.
For publishers, this could fundamentally reshape monetisation models.
Websites historically optimised for:
- pageviews
- impression scale
- shallow informational clicks
AI search may increasingly reward:
- expertise depth
- transactional intent
- authority trust
- unique reporting
- specialised analysis
This could compress commodity publishing while strengthening high-authority niche publishers.
Why Technical SEO Still Matters More Than Ever
Ironically, AI search may make technical SEO more important, not less.
Because retrieval pipelines rely on structured understanding, websites with poor technical foundations become harder for AI systems to interpret.
Critical technical factors now include:
- clean semantic HTML
- schema consistency
- crawl efficiency
- rendering accessibility
- internal link graph structure
- entity disambiguation
- canonical integrity
- content freshness signals
- topical clustering
- low-latency delivery
- structured heading hierarchies
Large language models cannot retrieve pages efficiently if those pages are:
- orphaned
- poorly structured
- JavaScript-blocked
- ambiguously categorised
- semantically inconsistent
AI retrieval systems are computationally expensive.
Every unnecessary retrieval operation increases inference cost.
As a result, systems increasingly favour documents that minimise ambiguity and maximise semantic clarity.
That is fundamentally a technical SEO problem.
The Industry May Be Selling Complexity That Google Does Not Want
The emergence of GEO and AEO reflects a familiar pattern in digital marketing.
Every platform transition creates an optimisation gold rush.
But Google’s latest guidance suggests the company does not want publishers building parallel optimisation ecosystems for AI search.
Instead, it appears Google is trying to preserve continuity between classic search and AI-enhanced search.
Strategically, this makes sense.
If Google forced publishers to rebuild content architectures specifically for AI systems, the transition cost across the open web would be enormous.
The web’s existing infrastructure — HTML, indexing, linking, metadata, semantic structure — already works remarkably well.
Google’s AI systems are being engineered to operate on top of that infrastructure, not replace it entirely.
That may ultimately be the most important takeaway from Google’s statement.
The future of search may look radically different on the surface.
But underneath, the same foundational principles still dominate:
- relevance
- authority
- structure
- accessibility
- originality
- user satisfaction
- information value
The interface has changed. The underlying retrieval logic largely has not.
Agencies Rush to Sell “AIO” and “GEO” Services Despite Google Saying Standard SEO Still Work
Digital marketing agencies and freelance consultants are rapidly rebranding traditional SEO services under new buzzwords like “AIO” (AI Optimization) and “GEO” (Generative Engine Optimization), hoping to cash in on confusion surrounding AI search tools.
Across social media and LinkedIn, marketers are promising businesses “special optimization” for AI platforms, charging premium retainers for what critics say is often just ordinary SEO with a new label.
The trend has exploded following the rise of AI-generated search answers and Google’s AI Overviews feature. But according to Google itself, websites do not need entirely new strategies to appear in AI-powered search experiences.
Google representatives have repeatedly stated that the same fundamentals still apply: create useful content, build authority, earn relevant links, and maintain strong technical SEO.
Industry veterans say many businesses are being misled into believing AI search requires secret tactics or proprietary systems.
“Most of this is just rebranded SEO,” one consultant said. “People are slapping ‘AI’ onto existing services and doubling the price.”
While some adaptation for changing search behavior may be useful, critics argue the growing AIO/GEO gold rush is fueled more by fear marketing than by actual ranking changes
Bottom line
The numbers point to a simple conclusion: AI search has not replaced SEO—it is largely redistributing existing SEO signals in a new interface.
The “AIO/GEO boom” is real as a marketing trend, but the underlying system it claims to optimise is still, overwhelmingly, traditional search authority dressed in a generative layer.
0 Comments