SEO Ranked Your Website in a List of Links. GEO Gets You Cited Inside the Answer.

For twenty years, the internet ran on a simple contract.

You typed a keyword into a search box. Google returned a page of ten blue links. If you ranked high enough, the user clicked, arrived at your website, and you had a chance to convert them. Your website was the destination, and SEO was the discipline of winning that slot in the list.

That contract is being rewritten in real time. The list of links is still there, but it is no longer where the decision happens. Increasingly the decision happens inside a synthesised answer, and your brand either appears in that answer as a cited source, or it does not appear at all.

The cleanest way to explain the shift to a client is this: SEO optimises your website to rank in a list of links. GEO optimises your brand to be the source cited inside the AI's answer.

One is about winning a click from a menu. The other is about being the ingredient inside the dish.

GEO.thejourney.hk

SEO MENUGEO is inside the ingredients

What Ranking in a List of Links Actually Means

Traditional search is a retrieval economy. The engine retrieves documents, ranks them, and hands the user off to a destination. The entire model assumes a handoff: the engine shows a list, the user scans it, chooses, clicks, and reads your page.

Success is measured in three numbers: ranking position, click-through rate, and traffic. If you ranked third and earned a 4% CTR, you won a share of attention you could monetise on your own site.

Strictly speaking, Google ranks pages, not companies. Its index holds your homepage, your service page, your blog post, not "you." But the practical effect, being visible at the moment a buyer searches, made that distinction feel academic for two decades.

AI engines collapse the distinction. They cite entities. They name your brand, or they name your competitor's. The unit of competition shifts from the page to the brand's presence in the model's information ecosystem. We unpacked that shift separately in Why GEO is not just SEO with a new name.

Why the Clicks Are Disappearing

When Google shows an AI Overview, the summary at the top of results, the top-ranking organic link loses a large share of its clicks. Seer Interactive tracked 3,119 queries across 42 organisations and found organic CTR on AI Overview queries fell 61%, from 1.76% to 0.61%, between June 2024 and September 2025. Paid CTR on those queries dropped 68%.

Ahrefs, tracking 300,000 keywords, found that by December 2025 an AI Overview correlated with a 58% lower CTR for the top-ranking page, nearly double the 34.5% decline it measured a year earlier. In Google's AI Mode, Semrush's clickstream data puts the zero-click rate at roughly 93%.

The evidence now comes from enough independent samples that the direction is not in dispute. What differs is magnitude, and that depends on what each study measured.

GEO.thejourney.hk

But here is the critical nuance: the click is not dead, it is conditional.

Seer found that brands cited inside AI Overviews earned 35% more organic clicks and 91% more paid clicks than brands that were not cited. The same research carries an honest caveat: "We cannot definitively prove that citation causes higher CTRs, it's equally possible that brands with stronger authority and higher baseline CTRs are simply more likely to be cited by Google's AI. What we can say with confidence is that queries where you're cited consistently outperform those where you're not, regardless of the causal direction."

The conclusion is not that traffic is dead. It is that the answer itself has become the gatekeeper. If you are not in the answer, the shrinking pool of clicks flows to someone else.

What Replaces the List: the Citation Economy

In a generative engine, ChatGPT, Gemini, Perplexity, Claude, Copilot, DeepSeek and the rest, there is no list of ten URLs to scan. There is one synthesised response. The AI may cite one, two, or three sources inline, often as small links or footnotes. The user reads the answer, and frequently never leaves the interface.

Winning now means inclusion, not position. The unit of value shifts from the link to the citation, and then to the mention: being part of the model's understanding of who the credible sources are in your category.

Retrieval and citation are two different decisions, made at two different moments, and most reporting treats them as one. That gap is the subject of Read, But Not Cited, and it explains why a page can be fetched, read end to end, and still not named.

GEO.thejourney.hk

This is not a marketing metaphor. It is a formal research problem. In August 2024, researchers published the first academic framework for Generative Engine Optimization at the ACM SIGKDD Conference, defining it as "the first novel paradigm to aid content creators in improving their content visibility in generative engine responses", and reporting visibility gains of up to 40%. The field is barely two years old and the rules are still being written, which is why brands that move early can build a durable advantage.

SEO vs. GEO: the Difference in One Table

SEO optimises a page for a ranked list. GEO optimises a brand for inclusion in a synthesised answer. The win conditions, the assets, the signals and the measurement all diverge from there.

The deepest difference is the asset being optimised. SEO optimises a page. GEO optimises a brand's presence in the model's information ecosystem, which includes your website, but also your structured data, your third-party mentions, your entity consistency, and the technical accessibility of your content to retrieval bots.

The Four Workstreams of GEO

If SEO breaks down into keyword tracking, content writing, technical SEO and backlinks, GEO has a parallel four-part structure. This is the most useful way to show a client that GEO is not a mysterious new black box. It is a disciplined workstream with clear deliverables.

1. Prompt tracking and improvement replaces keyword tracking. You monitor which prompts trigger your brand, which competitors appear instead, and whether you are cited, merely mentioned, or absent. The improvement work is to raise citation share on the prompts that matter to your pipeline. Seer's own conclusion points the same way: report on share of voice in AI citations, branded search lift and assisted conversions, rather than traffic and clicks alone.

2. Answer-first content and topical authority replaces content writing. GEO content is written to be extracted: a direct, self-contained answer immediately under each subheading, clean HTML tables for comparative data, explicit definitions, and deep topical coverage. The goal is extractability and trustworthiness, not keyword density. The overlap with answer engine optimisation is deliberate: the paragraph that wins an answer engine is the one a generative engine can quote.

3. AI crawler access and machine-readability replaces technical SEO. Cloudflare now publishes the ratio of how often each platform crawls a site versus how often it refers a visitor. In its June 2025 sample, Anthropic's crawler made nearly 71,000 HTML requests for every referral, while search-engine crawlers sat at the opposite end of the same table. That asymmetry is why publishers block AI bots, and why blocking them is also a decision to opt out of being cited. It is also a question of intent: in Cloudflare Radar's Q3 2026 reading, 47.6% of AI crawler requests were for training, only 8.3% for search, and 3.7% served a live user request. Most AI crawling cannot cite you, because most of it is not looking for something to cite.

GEO.thejourney.hk

4. Entity authority and citation signals replaces backlinks. SEO builds links to a URL. GEO builds entity consistency so the model can verify who you are: your LinkedIn presence, industry directories, verified profiles, third-party mentions and consistent naming across the web all function as corroboration signals. The goal is not just a strong page, but a recognisable, verifiable entity the model is comfortable citing. This is why digital PR and directory hygiene are GEO workstreams, not side projects.

What the Data Says, and What It Does Not Say Yet

Credibility in this field comes from separating correlation from causation. Three honest limits are worth stating plainly.

The click is declining, but not uniformly. AI Overviews reduce CTR for top-ranking pages, yet cited brands earn more clicks than non-cited brands. Traffic is being redistributed toward cited sources, not destroyed.

AI referral traffic is growing, but it is still small and converts differently. Referrals from ChatGPT grew 206% year on year, yet the largest academic sample to date, 973 e-commerce sites in a working paper, found them about 0.2% of sessions, roughly 200 times smaller than Google organic, and converting below organic and affiliate traffic. AI visibility is stronger for discovery and brand preference than for last-click revenue. The honest pitch is to own the answer before the click, not to replace your whole funnel.

Structured data correlates with citations, but adding it alone does not cause them. Ahrefs found that AI-cited pages are roughly three times more likely to carry JSON-LD. It then tested causation directly, tracking 1,885 pages that added JSON-LD against 4,000 control pages, and found no meaningful uplift on any platform. Schema is baseline hygiene, not a magic lever.

The pattern is consistent: the answer layer is becoming the first moment of brand discovery, and citation is the new visibility unit. But the mechanics are fluid, so the right strategy is iterative, measured at the prompt level, and adapted per engine.

Why This Matters More in Asia, and for Cross-Border Brands

The AI engine landscape is not monolithic. DeepSeek is now the fifth most-visited AI product in the world by web traffic, ahead of Perplexity, with Kimi and Qwen both inside the global top 25, according to Exploding Topics' October 2026 ranking.

A brand's visibility in ChatGPT does not automatically translate into DeepSeek or Kimi. Retrieval patterns, source preferences, the index, and the language of the prompts all differ. English-language optimisation does not reach a Chinese-language question, and a citation in one engine does not carry to the next. That is the same fragmentation we mapped across the Western engines in The Great Fragmentation.

Cross-border GEO is therefore not one strategy but a portfolio of engine-specific ones. A Hong Kong brand targeting both English- and Chinese-speaking buyers has to track citations across two ecosystems, often in two languages. In our own client conversations here this year, that is the most common gap: teams have benchmarked ChatGPT carefully and have never once asked what Qwen says about them.

GEO.thejourney.hk

The opportunity is especially relevant here because the answer layer compresses the advantage incumbents held in the link economy. A smaller brand that is consistently described and corroborated across the web can be cited ahead of a larger one that has never bothered to be legible to a model.

The Bottom Line

SEO is not obsolete. You still need to be crawlable, to have relevant content, and to earn links. But SEO alone now optimises for a shrinking surface: a list of links that fewer users scan and fewer still click. That is why our SEO and GEO workstreams now run as one programme, with AEO between them as the discipline of writing the sentence an engine will quote.

GEO is the discipline of optimising for the growing surface: the synthesised answer where your buyer makes a first judgement about who to trust.

The brands that win the next three years will not choose between the two. They will run both, because the buyer uses both, often in the same journey: a query in an AI tool, a synthesised recommendation, then a return to Google to verify. If your brand is absent from the AI answer, you may never enter the verification set at all.

You cannot afford to be missing from the answer that shapes the shortlist before the click ever happens.

→ If this is the first piece you have read here, start with Why GEO is not just SEO with a new name, then What Is AEO? The Answer Layer, Explained. Unfamiliar terms are defined in the GEO and AI visibility glossary.

→ Want your own read-versus-cited picture? Ask for a free GEO Compass scan, or see how we work and the results we have delivered.

→ Want the numbers for your own site? Create a free account and open the live GEO dashboard, or see how our SEO, GEO and AEO services fit together.

Sources

Every external figure above traces to a study below. Where credible datasets disagree, we say so in the text rather than quoting whichever number flatters the argument.

• Seer Interactive (September 2025). AIO Impact on Google CTR: September 2025 Update — the 61% organic and 68% paid CTR decline, and the 35% / 91% advantage for cited brands.

• Ahrefs (February 2026). Update: AI Overviews Reduce Clicks by 58% — 300,000 keywords; 58% lower CTR for the top-ranking page, up from 34.5%.

• Semrush (2025). Google AI Mode's Early Adoption and SEO Impact — the roughly 93% zero-click rate in AI Mode.

• Semrush (2026). ChatGPT traffic analysis: 17 months of clickstream data — outbound referrals growing 206% between January 2025 and January 2026.

• Schulze et al., SSRN working paper (2026). ChatGPT Referrals to E-Commerce Websites — 973 e-commerce sites, $20 billion revenue; referrals about 0.2% of sessions and converting below organic, as reported by Search Engine Land.

• Ahrefs (May 2026). We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved. — the correlation across 6 million URLs and the controlled test showing no uplift.

• Cloudflare (June 2025). The crawl before the fall: AI's impact on referrals — crawl-to-refer ratios by platform, including Anthropic at roughly 70,900:1.

• Technology Checker (Q3 2026), from Cloudflare Radar. AI Crawler Statistics — crawl-purpose split through September 2026: 47.62% training, 8.26% search, 3.66% user action.

• Exploding Topics (October 2026). Most Popular AI Tools Ranked — DeepSeek at #5 by traffic, with Kimi and Qwen inside the global top 25.

• Aggarwal et al. (2024). GEO: Generative Engine Optimization — the founding GEO paper, presented at ACM SIGKDD.

• GEO The Journey (2026). Own daily prompt tracking across ten engines, including DeepSeek, Kimi and Qwen. Methodology available on request.