Gloria K. SEO Specialist, AI Strategy & Performance Marketing

I have been in the SEO over 15 years. Ive seen the hacks and the results: first the hack works, then not, try another ‘hack’, same pattern. But even as the online visibility has changed, the actual work and underlying logic has not.

13.5.2026

I started doing SEO when keyword stuffing worked. Around me, an entire industry was doing the opposite. Fifteen years ago, you could rank a page by repeating the same phrase in every paragraph, buying a few hundred backlinks from a directory farm, and calling it a strategy. Then Google started systematically dismantling it, update by update, and suddenly the thing I had been doing out of basic self-respect became the professional standard everyone was scrambling to meet.

I have now watched fifteen years of this cycle. The shortcut works, then it does not, then the people who never relied on it are still standing.

How SEO Evolved into AEO and GEO: From 2011 to 2026

When I properly entered the SEO world, the dominant approach was still largely mechanical. Keyword density, link volume, and directory submissions. The relationship between effort and visible ranking result was almost transactional, and a lot of people had built comfortable businesses on it.

It was, to borrow from Austen, a truth universally acknowledged across the industry that a page in want of traffic must have a sufficient number of exact-match anchor texts. The people writing genuinely useful content were considered naive. Some of them were. Most of them just outlasted everyone who laughed at them.

Google has spent the last fifteen years systematically closing every shortcut that existed. Just look at the full timeline:

• Panda landed in 2011 and targeted thin, low-quality, and duplicate content. Content farms lost up to 40% of their visibility within weeks. Nobody saw it coming. Everyone claimed they had.

• Penguin followed in 2012 and penalized manipulative link schemes. Directory farms, link wheels, and paid link networks collapsed across affected verticals.

• Hummingbird arrived in 2013 and brought semantic search. Google stopped matching keywords and started matching intent. This is where keyword stuffing definitively died, though some corners of the industry took several more years to accept the funeral.

• RankBrain came in 2015 and introduced machine learning into the core ranking stack. Topical authority started mattering more than keyword density.

• BERT followed in 2019, improving how Google interpreted conversational and context-dependent queries. Writing for humans, which the sensible people had always done, suddenly had measurable ranking advantages.

• The Helpful Content Updates ran from 2022 through 2024. Google explicitly went after content written for search engines rather than people, including mass-produced AI content and programmatic text. A large portion of the cheap-content industry started quietly disappearing.

• AI Overviews rolled out globally in 2025. Organic click-through rates on affected queries dropped by approximately 61% (BrightEdge, 2026). Position one stopped meaning what position one used to mean.

• Gemini 3 became the default model powering AI Overviews on January 27, 2026. The percentage of AI-cited pages that also ranked in the top 10 organically dropped from 76% to 38% within weeks (Ahrefs, 2026). Ranking and being cited became two separate jobs almost overnight.

• The February through March 2026 update cluster brought three official Google updates inside five weeks: a Discover-only core update from February 5 to 27, a spam update from March 24 to 25 that finished in under 20 hours, and a broad core update from March 27 to April 8 affecting all languages, industries, and site types (Google Search Status Dashboard, 2026).

• The May 2026 core update began rolling out on May 21, two days after Google I/O, and is still completing as of this writing. It is global, affects all languages and site types, and arrived on the heels of the biggest Search announcements in years (Google Search Status Dashboard, 2026).

Google I/O in May 2026 confirmed the scale of what had shifted (to what it should have been all along). AI Mode now has over one billion monthly users. Queries are on average three times longer than traditional searches.

Position one is no longer the prize. Getting cited inside an AI response is.

Every one of these updates taught the same lesson: those who built on whatever “shortcut ” was working at the time got wiped out. The ones who built around what Google was actually trying to reward from the beginning (user intent) kept their traffic and kept their clients.

The cosmetology blogger from Tallinn turned out to have the right instinct all along.

I am not going to pretend that it does not feel at least a little satisfying.

Google’s Official Position: AEO and GEO Is Still SEO

On May 15, 2026, Google published its official guide on optimizing for generative AI features in Search. Their exact words on AEO and GEO: “From Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.” (Google Search Central, 2026)

I am not making this up. That is Google’s documentation. The article you are reading is built around the same conclusion, and I arrived at it fifteen years before Google decided to write it down.

Google also said in the same guide that llms.txt files, content chunking for AI, and rewriting content specifically for AI systems are unnecessary for Google Search. The tactics being sold as AEO and GEO hacks are, by Google’s own account, distractions. What they want is the same thing they have always wanted: useful content, clear structure, and a site that works properly.

The State of Search in 2026: Rankings, Citations, and AI Mode

If you are still measuring success only by traditional rankings, you are looking at half the picture. The other half lives inside AI Overviews, ChatGPT, Perplexity, and Gemini, and it is the half that is growing fastest.

Roughly 65% of global Google searches now end without a click aka zero-click search (Searchlab, 2026).

AI Overviews appear on 48% of search results pages, up 58% year over year (BrightEdge, February 2026).

62% of AI Overview citations come from outside the top 10 organic results, based on an Ahrefs study of 863,000 queries across six markets. Google holds approximately 91% of global search market share (StatCounter, 2026).

The empty space between rankings and citations is the challenge right now. Ranking position one does not guarantee a place inside an AI Overview. Topical depth, entity clarity, and citation-worthy structure do, but not 100%.

Ranking and citation are parallel games now, and playing only one of them is how you lose half your traffic without ever understanding why.

What Has Survived Every Google Algorithm Update Since 2011

Across fifteen years of working in iGaming, fintech, biotech, crypto, and e-commerce, five things have held up regardless of what Google released. I structure every project around these because they have outlasted every shortcut I have ever watched fail.

Clean technical architecture: logical URL hierarchies, fast loading speeds, structured data, no orphaned pages, crawl paths that make sense to a search bot. Not glamorous. Not optional.

Genuine subject expertise: content written by someone who actually knows the topic, structured for the person asking rather than the algorithm reading. Sounds obvious. The content industry has spent fifteen years proving it is not.

Editorially earned links: inbound link profiles built on relevance and real credibility. I have watched the paid link industry collapse twice. There will be a third time.

Stable Core Web Vitals: Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift all hold across mobile and desktop. A beautiful site that loads in six seconds is a beautiful site no one will wait for.

Internal linking that reflects real topical relationships: structure that mirrors how concepts actually connect, anchor-text games around exact-match keywords won’t work.

That’s how I have survived Panda, the Helpful Content Updates, the Gemini 3 rollout, and everything in between. The goal never changed. We still need to produce content FOR the users, and know what you write about. Wow. What a rebellious idea indeed.

AEO and GEO in 2026: What They Actually Mean for SEO

Answer Engine Optimization and Generative Engine Optimization describe the same shift. The terminology varies by who is writing, but the actual work is the same: optimizing content so AI systems can extract, synthesize, and cite it accurately when answering queries inside ChatGPT, Perplexity, Gemini, Google AI Overviews, and the rest.

I started focusing on AEO and GEO early because the direction was obvious. The same instinct that kept me ahead of every algorithm update told me this was the biggest structural shift since Google launched. I now run traditional SEO and AEO/GEO as parallel workstreams for every client, with shared technical foundations and separate optimization layers on top.

How to Structure Content for AEO and GEO Citations

AI models prefer content that places a concise, direct answer within the first 100 words of a section. Heading hierarchies need to be clear. Entities need to be named explicitly. Pronouns and assumed context are friction. The model is looking for extractable, attributable answers and it skips content that makes that work harder.

I structure every page around the question I expect a user to actually ask, then answer it cleanly at the top of the relevant section. I track question intent and entity coverage alongside traditional keyword volume now. The questions are often more valuable than the volume numbers.

Schema Markup for AEO and GEO: Is it Really as Important as it Seems?

Simple answer: no, it’s not “the most important”. But it’s important.

Four schema types do most of the practical work for AEO and GEO in 2026. Implementing all four on the appropriate content types improves both citation probability and the accuracy of what AI systems extract and attribute.

• FAQPage maps directly to the question-and-answer extraction logic that large language models use. It is the single highest-impact schema type for AEO right now, and consistently underimplemented across most sites I audit.

• HowTo outlines procedural and step-by-step content clearly for machine ingestion. If your site publishes any kind of process explanation and does not have HowTo schema, that is traffic you are leaving on the table.

• Article with named author connects content to a verified Person entity and strengthens the E-E-A-T signals that AI systems weight when selecting sources to cite. The author entity is not decoration. It is infrastructure.

Speakable flags specific text passages for high-confidence extraction. Adoption is still low, which means implementing it now gives an outsized advantage while the field is uncrowded. Google noted in May 2026 that structured data is not required for AI features but continues to support overall SEO performance (Google Search Central, 2026).

How to Measure AEO and GEO Performance Alongside SEO

Standard rank tracking is no longer enough. With only 38% of AI citations aligning with top-10 organic rankings (Ahrefs, 2026), position tracking will systematically underreport your actual visibility.

Semrush AI Toolkit tracks AI Overview presence and citation across target keywords. Ahrefs Brand Radar monitors brand mentions and citation patterns across LLM outputs. BrightEdge handles enterprise-scale AI Overview tracking. Otterly.

AI and Profound are dedicated AI search visibility platforms built for citation tracking across multiple LLM platforms. Google Search Console has included AI Overview data within the standard Web search type since June 2025, though it does not isolate AI-driven impressions from organic.

Beyond tools, I run manual citation sampling every month across different LLMs for each client’s target queries. Manual work is not scalable, but definitely it’s not optional if you want accurate data.

Why AEO, GEO, and SEO Are Still the Same Job

Treating SEO and AEO as the same job is the fastest way to underperform at both. They share technical foundations but diverge in editorial structure and measurement. That distinction matters in practice, even if the underlying principle is identical.

Google holds approximately 91% of global search market share, and total search impressions are up 49% year over year (BrightEdge, 2026). Users search more and click less. Traditional SEO is a stable market with rising quality requirements.

The AEO and GEO layer sits on top of that foundation and requires editorial adjustments that go beyond traditional ranking work. The framework I use starts with the same question it always has: what does the person asking this actually need, and how do I make it impossible for any system, human or machine, to misunderstand the answer.

That has been the job since I started writing cosmetology articles for an audience of real people who wanted real information. The systems reading those answers now are considerably more sophisticated. The principle is identical.