<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Tom Kong on Medium]]></title>
        <description><![CDATA[Stories by Tom Kong on Medium]]></description>
        <link>https://medium.com/@wel3kxial?source=rss-f0bd763c4066------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*f8oesmWsrWCSQ8Aj</url>
            <title>Stories by Tom Kong on Medium</title>
            <link>https://medium.com/@wel3kxial?source=rss-f0bd763c4066------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 08:04:48 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@wel3kxial/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[GEO Technology Goes Viral!]]></title>
            <link>https://medium.com/@wel3kxial/geo-technology-goes-viral-dcee109a707c?source=rss-f0bd763c4066------2</link>
            <guid isPermaLink="false">https://medium.com/p/dcee109a707c</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[geo]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[traffic]]></category>
            <dc:creator><![CDATA[Tom Kong]]></dc:creator>
            <pubDate>Thu, 10 Jul 2025 18:38:11 GMT</pubDate>
            <atom:updated>2025-07-10T18:38:11.850Z</atom:updated>
            <content:encoded><![CDATA[<h3>GEO Technology Goes Viral!<br>Exclusive Analysis of Princeton University’s Authoritative Research + Practical Guide — Everything You Need to Know About Generative Engine Optimization in One Article.</h3><h3><strong>1. Introduction to Princeton University’s Authoritative and Pioneering Research on GEO</strong></h3><p>With the rapid development of large language models (LLMs, such as the well-known ChatGPT), search engines have quietly entered a brand-new era. Today, generative search engines like ChatGPT and Bing Chat can directly generate accurate and personalized answers for users by synthesizing information from multiple sources. This “ask a question, get an answer” experience is gradually replacing the traditional search method we’re used to — manually sifting through pages of results on Google.</p><p>However, while this transformation greatly benefits users, it presents a significant challenge for another group: websites and content creators. Unlike traditional search engines, which require users to click through links to visit different websites, generative engines summarize and present content directly within the response. Although this “one-stop information delivery” improves efficiency, it severely reduces the exposure opportunities for content creators. Small businesses and independent creators who rely on website traffic for income are now facing a sharp drop in traffic and revenue due to this trend.</p><p>What’s even more concerning is that generative engines typically operate as closed black-box systems. Outsiders have no visibility into how these engines retrieve, select, or synthesize content. This means content creators have virtually no control over how their content is quoted or displayed in generative responses — or even whether it’s referenced at all. In the face of this invisible and uncontrollable process, creators are becoming increasingly passive.</p><p>To address this new challenge, a research team at Princeton University introduced an entirely new concept: <strong>GEO (Generative Engine Optimization)</strong>. In simple terms, GEO empowers content creators to regain control in an AI-driven search environment. It provides a flexible set of strategies to optimize content so it’s more likely to be selected and cited by AI systems — ultimately increasing its visibility in generative answers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/980/1*dGTo2BWQs-DZrQaYgUwa2g.png" /></figure><p>Paper Link: <a href="https://arxiv.org/pdf/2311.09735">https://arxiv.org/pdf/2311.09735</a></p><p>To enable scientific evaluation of this optimization framework, the Princeton team built a large-scale benchmark dataset called <strong>GEO-bench</strong>. This dataset includes <strong>10,000 real user queries across various domains</strong>, along with corresponding web page sources. GEO-bench allows researchers to test GEO strategies in simulated real-world search scenarios with systematic and comprehensive performance assessments.</p><h3>The results are striking:</h3><ul><li>By applying GEO strategies, <strong>content visibility in generative search responses increased by up to 40% on average</strong>.</li><li>Content containing <strong>clear citations, references, and statistics</strong> saw the most significant gains in exposure.</li><li>In real-world testing — such as on the generative search platform <strong>Perplexity.ai</strong> — optimized content achieved <strong>a 37% boost in visibility</strong>.</li></ul><p>This research not only opens a new path for content creators to survive and thrive but also offers important insights for the design of future AI-powered information retrieval systems.</p><p>This article offers an exclusive in-depth breakdown of Princeton University’s findings, along with a practical guide to applying GEO in real-world settings.</p><h3>In simple terms, GEO is a framework that helps content creators adapt to the new era of generative search. Its core ideas are:</h3><ul><li>Treating generative engines as <strong>black boxes</strong> and applying <strong>black-box optimization</strong> techniques to them;</li><li><strong>Modifying how content is presented</strong> — including style, structure, and formatting — to make it more easily identified, understood, and quoted by AI systems;</li><li><strong>Maximizing the visibility and citation</strong> of content within AI-generated responses.</li></ul><h3>2. GEO Requires an Entirely New Visibility Evaluation Framework</h3><p>In traditional search engines like Google, it’s easy to evaluate how a website performs: rankings are clear — who’s in 1st, 2nd, or 3rd place is visible at a glance. These engines typically present direct links or content snippets from original sources, following a <strong>linear and quantifiable</strong> logic. As a result, traditional SEO focuses heavily on rankings: the higher your page appears, the more exposure it gets.</p><p>But with <strong>generative engines</strong>, things get far more complex.</p><p>As shown in Figure 3 of the Princeton paper, generative engines don’t just list links — they <strong>generate a complete, synthesized answer</strong>, blending and summarizing content from multiple websites. Your content might be quoted only briefly, or interwoven with text from other sources. The <strong>position, manner, length, and impact</strong> of the citation can vary greatly.</p><p>A vivid analogy:</p><blockquote><strong><em>Traditional search engines are like à la carte dining — you choose and eat one dish at a time.</em></strong><em><br></em><strong><em>Generative engines are like a buffet plate — AI mixes selections from various kitchens into one dish, and the user consumes the final mix without seeing the original menu.</em></strong></blockquote><p>In this new context, simply asking “Was I cited?” or “How many times?” is no longer enough. We need to <strong>assess the overall visibility and influence of a web page within an AI-generated response</strong>, through dimensions such as:</p><ul><li><strong>Position of citation</strong>: Is the content used at the beginning, middle, or buried at the end of the answer?</li><li><strong>Length of citation</strong>: Is it just one sentence, or a full paragraph?</li><li><strong>Contribution score</strong>: How crucial is this content to the completeness of the answer?</li><li><strong>Relevance</strong>: How closely does the cited content match the user’s original query?</li></ul><p>What’s more, every content creator has <strong>different goals</strong>:</p><ul><li>Some may want their content <strong>featured at the beginning</strong> of the response to establish authority;</li><li>Others may prefer being <strong>cited for specific stats or key data points</strong>;</li><li>Some may aim to <strong>contribute across multiple subtopics</strong> in a comprehensive answer.</li></ul><p>That’s why GEO proposes a <strong>flexible visibility evaluation framework</strong>, allowing content creators to <strong>customize how they define and measure success</strong> in generative environments.</p><p>In short:</p><ul><li>If you want to <strong>lead the answer</strong>, focus on <strong>concise summaries and strong opening cues</strong>;</li><li>If you want to be <strong>quoted for authoritative data</strong>, emphasize <strong>statistics and source attribution</strong>;</li><li>If you want to be <strong>embedded throughout the answer</strong>, <strong>cover multiple angles or subtopics</strong>.</li></ul><p>With this <strong>black-box optimization approach</strong>, GEO doesn’t just tell you <strong>whether your content was used</strong> — it teaches you <strong>how to make AI prefer using your content</strong>.</p><h3>3. Evaluating the Effectiveness of GEO: What Kind of Content Is Most Likely to Be Cited by AI?</h3><p><strong>Princeton University’s Research on the Effectiveness of Different GEO Optimization Strategies Reveals Striking Insights</strong></p><p>The Princeton research team systematically tested various <strong>GEO optimization strategies</strong>, and the results were quite revealing:</p><h3>🔴 Traditional SEO Tactics Don’t Work Anymore</h3><ul><li>One key finding is that <strong>old-school tactics</strong> like <strong>keyword stuffing</strong> — once common in SEO — <strong>perform poorly</strong> in generative engines.</li><li>Simply cramming in keywords no longer influences AI-generated results effectively.</li></ul><h3>✅ What Does Work: New GEO Techniques</h3><p>Instead, newly proposed <strong>GEO-specific strategies</strong> showed clear advantages:</p><ul><li><strong>Statistics Addition</strong> — embedding concrete data and figures</li><li><strong>Quotation Addition</strong> — adding direct quotes or citable statements</li></ul><p>These techniques significantly improved visibility and influence in AI-generated responses by making the content <strong>more interpretable, authoritative, and valuable</strong> to the model.</p><h3>📈 Key Metrics from the Study</h3><p>Two core evaluation metrics were used:</p><p><strong>MetricExplanationPosition-Adjusted Word Count</strong>Measures not only how much content is quoted, but also <em>where</em> it appears (early placement = more impact)<strong>Subjective Impression Score</strong>Assesses the <em>perceived importance and relevance</em> of quoted content from a user’s point of view</p><h4>Results:</h4><ul><li>The <strong>best-performing optimization strategy</strong> improved <strong>position-adjusted word count by 41%</strong> compared to the baseline.</li><li>It also improved the <strong>subjective impression score by 28%</strong>.</li></ul><p>To ensure fair comparison, Princeton normalized the evaluation scores across different strategies — allowing direct insight into which methods truly outperform others.</p><p><strong>From Table 2 of the paper</strong>, Princeton’s research team conducted a focused comparison between websites with different rankings on traditional search engines (like Google) to examine how their visibility changed in <strong>generative search</strong> after applying GEO optimizations.</p><h3>🔍 Key Finding: GEO Benefits Are Especially Strong for Lower-Ranked Sites</h3><ul><li><strong>High-ranking websites</strong> (e.g., those that already appear on Google’s first page) did see a <strong>moderate improvement</strong>in content visibility after applying GEO techniques — but the increase was relatively stable.</li><li><strong>Mid- to low-ranking websites</strong> (e.g., those on page 3 or beyond in traditional search) experienced <strong>significant boosts in visibility</strong> when optimized with GEO strategies.</li></ul><h3>🧠 Why This Happens:</h3><ul><li>Traditional SEO favors those who already rank high — <strong>your position determines your traffic</strong>.</li><li>But <strong>generative engines don’t strictly follow original rankings</strong>. Instead, they focus on <strong>content relevance, quality, and citation value</strong>.</li><li>So even if your site ranks low in Google, if your content is <strong>well-structured, authoritative, and semantically aligned</strong>, it can still be <strong>prioritized in an AI-generated response</strong>.</li></ul><h3>🚀 The Big Takeaway:</h3><blockquote><strong><em>GEO opens the door for underdogs.</em></strong><em><br>Small brands, independent creators, or low-ranking sites now have a </em><strong><em>real opportunity to “leapfrog” the competition</em></strong><em> — not by chasing traditional rankings, but by </em><strong><em>becoming AI-friendly</em></strong><em>.</em></blockquote><p>This marks a <strong>fundamental shift</strong> in how visibility is earned online.</p><h3>From Paper Table 4: GEO Strategies Work Differently for Different Content Types and Industries</h3><p>The research also shows that the <strong>effectiveness of GEO tactics varies by domain</strong>. That is, different types of websites benefit from <strong>different optimization strategies</strong>.</p><p>Here’s a simplified summary of those findings:</p><h3>🎯 Final Thought:</h3><p>There is no “one-size-fits-all” strategy in GEO.</p><blockquote><em>The key is to </em><strong><em>align your content optimization with the expectations of generative models</em></strong><em> — and </em><strong><em>tailor your approach to your domain</em></strong><em>.</em></blockquote><p>This makes GEO not just a technique, but a <strong>customizable framework</strong> for content success in the AI-driven search era.</p><p>The research team also presented numerous <strong>real-world GEO optimization examples</strong>, demonstrating a powerful insight:</p><blockquote><strong><em>You don’t need to rewrite everything. Even small tweaks can significantly boost a page’s visibility in generative engines.</em></strong></blockquote><h3>🔧 GEO Optimization Involves Two Main Types of Edits:</h3><h4>1. ✅ Additions — Make Your Content More Informative and AI-Friendly</h4><p>These enhancements focus on <strong>adding precision, authority, and clarity</strong> to help AI models better understand and trust your content:</p><ul><li><strong>Add authoritative statistics</strong> (e.g., “According to the WHO, 72% of…”)</li><li><strong>Include direct citations or quotes</strong> (e.g., “Dr. Smith notes that…”)</li><li><strong>Clarify technical terms or keywords</strong> so they’re easier to semantically link</li><li><strong>Explicitly mark sources or references</strong>, making attribution clearer</li></ul><p>📌 <em>Goal: Strengthen the informational density and reliability of your content.</em></p><h4>2. ❌ Deletions — Remove the Noise, Keep the Signal</h4><p>These edits strip away fluff that can confuse or dilute the value of your message for AI systems:</p><ul><li>Remove <strong>irrelevant or tangential content</strong></li><li>Cut <strong>redundant sentences</strong> or vague generalizations</li><li>Eliminate <strong>filler words or ambiguous language</strong></li></ul><p>📌 <em>Goal: Make your content lean, focused, and easier for the AI to extract meaningful elements.</em></p><p>The research team also conducted <strong>real-world testing</strong> on a <strong>commercial generative engine platform with millions of active users</strong>, and the results once again confirmed the effectiveness of GEO strategies. This proves that <strong>GEO isn’t just a lab theory — it works in the wild</strong>, helping content creators secure more visibility and attention in practical, high-traffic environments.</p><h3>🌐 Key Insight for the Digital Content Ecosystem:</h3><blockquote><strong><em>GEO is the first line of defense for content creators</em></strong><em> in the era of generative search — <br>a way to adapt to change, protect visibility, and preserve influence.</em></blockquote><p>As generative engines continue to reshape how users consume information — <strong>favoring synthesized, AI-curated answers over traditional links</strong> — content creators must shift their strategies from chasing search rankings to <strong>earning selection by AI</strong>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dcee109a707c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>