# Arthur Dosik > Arthur Dosik is an AI Consultant, SEO and GEO Strategist with 20+ years of experience in search and digital marketing. He helps businesses and individuals implement practical AI strategies spanning Generative Engine Optimization, AI-powered search, prompt engineering, content strategy, and workflow automation. He currently works for Leah (formerly ContractPodAi) and consults independently. ## Pages - [Home](https://arthurdosik.com/): Overview of Arthur Dosik's consulting, expertise, and the four focused service lines available to clients. - [Services](https://arthurdosik.com/services/): Single-page overview of every service with links to each deep-dive service page. - [Generative AI and Prompt Engineering](https://arthurdosik.com/generative-ai-prompt-engineering/): Prompt engineering fundamentals, model selection, reusable prompt libraries, and evaluation for production-grade AI workflows. - [SEO and GEO Strategy](https://arthurdosik.com/seo-geo-strategy/): Integrated SEO and Generative Engine Optimization including technical audits, AI Overviews strategy, LLM visibility, and citation share. - [AI Workflow Automation](https://arthurdosik.com/ai-workflow-automation/): n8n, Make.com, and Zapier automations with AI in the loop, covering design, monitoring, and guardrails. - [Content Strategy and AI Search](https://arthurdosik.com/content-strategy-ai-search/): Topic architecture, citation-ready formats, entity modeling, and AI-assisted editorial workflows built for generative search. - [Insights](https://arthurdosik.com/blog/): Longform writing and field notes on GEO, SEO, AI search, and prompt engineering. - [Prompt Library](https://arthurdosik.com/prompts/): Free, ready-to-use prompts for GEO audits, content rewrites, AI workflow scoping, and productivity systems. - [Free AI SEO and GEO Tools](https://arthurdosik.com/tools/): Hub for the free, AI-powered tools below: visibility checker, readiness scorecard, answer grader, content brief generator, title and meta optimizer, and prompt optimizer. - [AI Competitor Showdown](https://arthurdosik.com/competitor-showdown/): Free tool that asks a frontier AI model a buyer question and ranks your brand against named competitors, showing who AI recommends first, the gaps, and the fixes. - [AI Visibility Checker](https://arthurdosik.com/ai-visibility-checker/): Free interactive tool that scores technical GEO signals and runs a live brand-in-answer test for one buyer question. - [AI Search Readiness Scorecard](https://arthurdosik.com/ai-readiness-scorecard/): Free tool that fetches a URL and uses AI to grade six GEO citation signals, then returns a prioritized action plan. - [Answer Readiness Grader](https://arthurdosik.com/answer-readiness-grader/): Free tool that grades how readily a passage of content can be quoted by AI answer engines, with an optional AI rewrite layer. - [Content Brief Generator](https://arthurdosik.com/content-brief-generator/): Free tool that uses AI to produce a full GEO content brief from a target keyword: intent, outline, entities, FAQs, and internal links. - [Title and Meta Optimizer](https://arthurdosik.com/title-meta-optimizer/): Free tool with a live SERP preview plus AI-written title tag and meta description variants. - [Prompt Optimizer](https://arthurdosik.com/prompt-optimizer/): Free tool that uses AI to rewrite a rough prompt into a structured, model-ready prompt for a chosen goal and target model. - [About Arthur Dosik](https://arthurdosik.com/about/): Background, career timeline, expertise areas, and credentials of Arthur Dosik, AI Consultant and SEO/GEO Strategist. ## Machine-readable files - [Summary LLM file](https://arthurdosik.com/llms.txt): Concise site overview with key pages, articles, and prompts. - [XML sitemap](https://arthurdosik.com/sitemap-index.xml): Canonical crawl list with lastmod timestamps. ## Contact - Email: arthur@arthurdosik.com - Website: https://arthurdosik.com - LinkedIn: https://www.linkedin.com/in/arthurdosik - YouTube: https://www.youtube.com/@AIwithArthur - Book a consultation: https://calendly.com/adosik21/30min ## Articles (full text) # What Is llms.txt? Examples, the Spec, and an Honest Take for 2026 URL: https://arthurdosik.com/blog/what-is-llms-txt/ Published: 2026-06-26 Updated: 2026-06-26 Author: Arthur Dosik Tags: GEO, AI Search, SEO llms.txt is having a moment. It shows up in GEO checklists, agency pitch decks, and "rank in ChatGPT" listicles, usually described as the new robots.txt for AI. Here is the uncomfortable part: Google has said plainly that it does not need AI-specific files. No major AI engine has committed to reading llms.txt. And yet I publish one on this site, I keep it updated, and I think you probably should too. Those positions are not contradictory. They just require knowing what the file actually is, who reads it today, and what it can and cannot do for you. That is this post. ## TL;DR - **What it is:** llms.txt is a proposed standard, published by Jeremy Howard in September 2024, that gives AI systems a clean, curated markdown index of a website's most important content. - **Google ignores it:** Google's documentation states there is no need to create special machine-readable files for AI Overviews or AI Mode, so llms.txt does nothing for Google AI Overview visibility. - **ChatGPT does not commit to it:** There is no published evidence that ChatGPT reads llms.txt, and the documented path to ChatGPT visibility is allowing OAI-SearchBot in robots.txt and being crawlable. - **Who actually uses it:** AI developer tools and coding assistants fetch llms.txt files to load documentation context, which is why companies like Anthropic and Supabase publish them. - **My verdict:** I publish one because it costs an hour, carries zero risk, and forces a clean entity definition of the business, but real AI citations still come from crawlable, structured, authoritative content. ## What is llms.txt? llms.txt is a proposed standard, [published by Jeremy Howard in September 2024](https://llmstxt.org/), for giving large language models a curated, LLM-friendly index of a website. The file lives at the root of your domain, the same way robots.txt does, and it is written in plain markdown. The reasoning behind it is solid. AI systems work with limited context windows, and your actual website is a hostile environment for them: navigation, scripts, cookie banners, footers, and layout markup wrapped around the content that matters. llms.txt hands the machine a clean map instead: here is who we are, here is what matters, here are direct links with descriptions. Note what it is not. It is not an access-control file like robots.txt, which tells crawlers what they may not touch. llms.txt is the opposite: an invitation and a guide to your best content. ## What does an llms.txt file look like? The [spec](https://llmstxt.org/) defines a simple markdown structure, in this order: 1. **An H1** with the site or project name. The only required element. 2. **A blockquote** summarizing what the site is and who it serves. 3. **Optional context paragraphs** with anything an AI should know upfront. 4. **H2-delimited sections** containing link lists. Each entry: a markdown link, a colon, a one-line description. 5. **An optional "Optional" section** for secondary links that can be dropped when a shorter context is needed. A minimal example, abbreviated from [the one on this site](/llms.txt): ```markdown # Arthur Dosik > Arthur Dosik is an AI Consultant, SEO and GEO Strategist with 20+ > years of experience in search and digital marketing. ## Pages - [SEO and GEO Strategy](https://arthurdosik.com/seo-geo-strategy/): Integrated SEO and GEO including AI Overviews strategy and citation share. ## Articles - [GEO is Just SEO With a Rebrand](https://arthurdosik.com/blog/geo-is-just-seo-with-a-rebrand/): Why GEO is not a new discipline, and where to start. ``` That is the whole format. If you can write a README, you can write an llms.txt. ### llms.txt examples worth studying - **[Anthropic](https://platform.claude.com/docs/llms.txt)** publishes one for the Claude developer docs: a long, sectioned link index that AI coding tools can pull into context. - **[Supabase](https://supabase.com/llms.txt)** uses theirs as a hub pointing to per-product full-text files like llms/js.txt and llms/cli.txt, a nice pattern for large doc sites. - **[Mine](/llms.txt)** covers a consulting site: services, articles, prompt library, and contact, with the blockquote doing the entity-definition work I would want any AI to get right. Notice who is on that list. Developer documentation and expertise sites, where the consumer is often an AI tool fetching context on demand. That is not an accident, and it leads to the honest part. ## Does Google use llms.txt? No. Google's documentation on AI features says there is [no need to create special machine-readable files](https://developers.google.com/search/docs/appearance/ai-features) for AI Overviews or AI Mode. Google's AI surfaces are built on the regular search index, the one your existing pages are already in. So let me say it without hedging: publishing llms.txt will do nothing for your [Google AI Overview visibility](/blog/how-to-show-up-in-google-ai-overviews/). Any agency selling llms.txt as an AI Overviews tactic is selling you a file Google has explicitly said it does not need. ## Does ChatGPT read llms.txt? There is no published evidence that it does. OpenAI documents [its crawlers](https://developers.openai.com/api/docs/bots) clearly: GPTBot gathers training data, OAI-SearchBot powers ChatGPT search visibility, ChatGPT-User handles live user-initiated fetches. llms.txt appears nowhere in that documentation. If you want to be visible in ChatGPT, the documented mechanism is being crawlable by OAI-SearchBot, which is a [robots.txt and content problem](/blog/how-to-rank-in-chatgpt/), not an llms.txt problem. Where llms.txt demonstrably gets used today is narrower and more interesting: AI developer tools. Coding assistants and agent frameworks fetch llms.txt files to load documentation context on demand. If your site is documentation, an API, or a knowledge base that AI tools might pull into a working context, the file earns its keep right now. That is exactly why Anthropic and Supabase bother. ## Should you create one anyway? Yes, if it costs you an hour. Here is the honest cost-benefit: | | The case | |---|---| | **For** | Costs an hour. Zero risk. Useful to AI dev tools today. Forces you to write a clean entity definition of your business. Cheap insurance if any engine adopts the standard later. | | **Against** | No major search or answer engine commits to reading it. Zero measurable visibility lift today. One more file to keep updated, and a stale index is worse than none. | The reason I land on "publish it" is the second-order benefit: writing the file forces you to answer, in one blockquote, what your business is, who it serves, and what your most important pages are. That clarity exercise is the same [entity-definition work](/blog/geo-is-just-seo-with-a-rebrand/) that actually does move AI visibility. The file is a byproduct of work worth doing anyway. ## How to create an llms.txt file 1. **Write the H1 and blockquote first.** Name, then two or three sentences: who you are, what you do, who you serve. This is the highest-value part of the file. 2. **List your core pages** under an H2, each with a one-line description written for a machine: literal, specific, no marketing adjectives. 3. **Add sections that fit your site.** Articles, documentation, products, contact. Keep each entry to one line. 4. **Consider an llms-full.txt** if you publish substantial content. The index says what exists; the full file carries the actual text so a retrieval system can use it without crawling. This site generates [llms-full.txt](/llms-full.txt) at build time from the same content that makes the pages. 5. **Save it as plain markdown at your domain root** and add it to your deploy process so it cannot silently go stale. 6. **Update it when you publish.** A curated index that is missing your last six months of work tells a machine your site stopped. Skip the generators if your site is small. The file is short, the thinking is the value, and hand-writing it takes less time than evaluating tools to avoid hand-writing it. ## How do you get indexed by LLMs through llms.txt? You do not, and this misconception is common enough that it deserves its own answer. llms.txt is not a submission form. There is no queue on the other end. No engine treats the file as a request for inclusion. LLM visibility comes from the unglamorous stack: being crawlable by the AI bots in your robots.txt, existing in the sources engines retrieve from, publishing content [structured for passage-level extraction](/blog/how-to-show-up-in-google-ai-overviews/), and being the kind of source other sites cite. llms.txt rides along as a courtesy map. The map is worth drawing. Just do not confuse drawing the map with being on anyone's route. ## The takeaway llms.txt is a reasonable idea, a trivial implementation, and a wildly oversold tactic. Google does not read it. ChatGPT has not committed to it. AI developer tools genuinely use it. Publish one because it costs an hour, forces useful clarity about your entity, and might matter more later. Then put your real effort where the citations actually come from: crawlable, structured, authoritative content. The file is the easy part. It was always going to be the easy part. ## Frequently asked questions ### What is llms.txt? llms.txt is a proposed web standard, published by Jeremy Howard in September 2024, for giving large language models a curated, markdown-formatted index of a website's most important content. The file lives at yourdomain.com/llms.txt and contains the site name, a short summary, and organized lists of links with one-line descriptions. The goal is to help AI systems find and use your best content without parsing your full HTML, navigation, and scripts. ### Does Google use llms.txt? No. Google's official documentation on AI features states there is no need to create special machine-readable files or AI-specific text files to appear in AI Overviews or AI Mode. Google's AI surfaces work from the regular search index. Publishing an llms.txt file will not improve your Google AI Overview visibility, and anyone claiming otherwise is ahead of the evidence. ### Does ChatGPT read llms.txt? OpenAI has not announced that ChatGPT's crawlers consume llms.txt, and there is no published evidence that it influences ChatGPT search results. The documented way to be visible in ChatGPT search is to allow OAI-SearchBot in robots.txt and be crawlable. Where llms.txt files demonstrably get used today is in developer tools and AI coding assistants that fetch documentation context, which is why companies like Anthropic and Supabase publish them. ### What does an llms.txt file look like? An llms.txt file is plain markdown with a defined structure: an H1 with the site or project name, a blockquote summary, optional context paragraphs, and H2-delimited sections containing link lists where each entry is a markdown link followed by a one-line description. An optional final section lists secondary links that can be skipped when an AI needs a shorter context. It reads like a curated table of contents for machines. ### What is the difference between llms.txt and llms-full.txt? llms.txt is the short curated index: links plus one-line descriptions. llms-full.txt is the expanded version that includes full page or article content in one file, so an AI system can load your actual material without crawling each URL. Many sites publish both: the index for orientation, the full file for retrieval. Mine are at arthurdosik.com/llms.txt and arthurdosik.com/llms-full.txt. ### Should I create an llms.txt file for my website? If it costs you an hour, yes. The realistic upside is modest: today the file mostly serves AI developer tools and any retrieval system that chooses to look for it. The downside is zero, it cannot hurt you. Treat it as cheap insurance on a possible standard, not as a GEO tactic that will move your visibility this quarter. Crawlability, content structure, and authority remain the work that actually moves AI citations. ### How do I get indexed by LLMs through an llms.txt file? You do not, and that framing is the most common llms.txt misconception. The file is not a submission mechanism and no major AI engine treats it as an index request. LLM visibility comes from being crawlable by AI bots like GPTBot and OAI-SearchBot, being present in the sources AI engines retrieve from, and publishing content structured for extraction. llms.txt can ride along as a courtesy map of your site. It does not get you in the door. --- *If you want your site's technical and content foundations checked against how AI engines actually retrieve and cite, that is the core of my [AI Search Visibility and SEO Strategy service](/seo-geo-strategy/). Or [book a free 30-minute call](https://calendly.com/adosik21/30min) and ask me anything, including whether llms.txt is worth your hour. No pitch, no pressure.* --- # The Best AI Search Monitoring Tools in 2026 (And What to Track With Them) URL: https://arthurdosik.com/blog/best-ai-search-monitoring-tools/ Published: 2026-06-19 Updated: 2026-06-25 Author: Arthur Dosik Tags: GEO, AI Search, Measurement Two years ago you could not buy a tool that told you whether ChatGPT recommends your brand. Now there are more than a dozen, the category has three names (AI visibility, AI search monitoring, LLM visibility), and the pricing runs from free to four figures a month. I have a strong opinion about where to start: not with a purchase order. With a spreadsheet. But the tools have matured fast, and at a certain scale the spreadsheet stops being honest work and starts being a hobby. Here is how the market actually breaks down in 2026, what these platforms can and cannot do for your strategy, and how to pick without overbuying. ## TL;DR - **Where to start:** Start measuring AI visibility with a spreadsheet, not a purchase order, and stay free until the manual process stops being honest work and starts being a hobby. - **The market in 2026:** Profound and Conductor anchor the enterprise end, Ahrefs Brand Radar and the Semrush AI Visibility Toolkit ride along as suite add-ons, and Otterly.AI and Peec AI serve the dedicated mid-market. - **What to track:** Four metrics matter in order: brand-in-answer rate, citation share, sentiment and accuracy, and source attribution, and a tool that counts mentions without showing sources is not enough. - **The free method:** Run 30 to 50 buyer prompts monthly across ChatGPT, Gemini, Perplexity, and Google, logging brand mentions, citations, sentiment, and competitors in a spreadsheet for a few hours a month. - **When to buy:** Graduate to a paid tool when you track more than one brand, need weekly cadence, or someone above you wants a chart instead of a spreadsheet. ## What does an AI search monitoring tool do? An AI search monitoring tool runs queries against AI engines at scale (ChatGPT, Google AI Overviews, Gemini, Perplexity, Claude) and reports whether your brand shows up in the answers, how it is described, and which sources the engines cite. The four metrics that matter, in order: 1. **Brand-in-answer rate.** When someone asks an AI engine a question your business should win, are you named? This is the headline number, the [mention-share metric I have written about before](/blog/ai-agents-killing-the-click/). 2. **Citation share.** Which pages get used as sources, yours or your competitors'? This is where the actionable work lives. 3. **Sentiment and accuracy.** Being mentioned wrongly can be worse than not being mentioned. [AI engines get business facts wrong](/blog/chatgpt-wrong-business-info/) more often than most owners realize. 4. **Source attribution.** Which specific pages, directories, and third-party sites feed the answers about your category. This is your fix list. A tool that gives you a mention count without source attribution tells you that you have a problem without telling you where to fix it. Demand both. ## What are the best AI search monitoring tools in 2026? Here is the field as it stands. I am deliberately talking tiers rather than exact prices, because this market reprices constantly. Check the vendor pages for current numbers. | Tool | Tier | Best for | |---|---|---| | [Profound](https://www.tryprofound.com) | Enterprise, four figures monthly | Large brands and agencies that need scale, integrations, and dedicated support | | [Ahrefs Brand Radar](https://ahrefs.com/brand-radar) | Suite add-on | Teams already on Ahrefs; strong prompt-volume data for topic discovery | | [Semrush AI Visibility Toolkit](https://www.semrush.com/ai-seo/) | Suite add-on | Teams already on Semrush; keeps AI visibility next to rankings and site audits | | [Otterly.AI](https://otterly.ai) | Dedicated, entry-level pricing | Small teams and solo marketers who want automated tracking without a suite | | [Peec AI](https://peec.ai) | Dedicated, mid-market | Marketing teams that want clean visibility, position, and sentiment tracking with source-level evidence | A few honest notes on this market: - **The suite add-ons changed the calculus.** When AI visibility meant a separate vendor, the case for waiting was strong. Now that Semrush and Ahrefs bundle it next to data you already pay for, the marginal cost of measuring dropped hard. - **Enterprise tools are selling workflow, not just data.** Profound's pitch is agentic: monitoring plus automated optimization. If you are a consultant or in-house team that acts on the data yourself, you may not need that layer. - **Nobody covers everything.** Engine coverage varies, prompt sampling methods vary, and two tools will give you two different brand-in-answer rates for the same month. Pick one, keep your methodology fixed, and track direction rather than absolute numbers. ## How can an AI search monitoring platform improve SEO strategy? A monitoring platform improves SEO strategy by closing a feedback loop that rank trackers no longer cover. Rankings tell you where pages sit in a list. They no longer tell you whether you exist in the answer layer where [a growing share of searches now end](/blog/ai-agents-killing-the-click/). Concretely, monitoring data changes four decisions: **Which content to build.** The tools surface the prompts and questions where your category gets asked and you are absent. That list is a content calendar with intent data attached. It pairs directly with the passage-level work I describe in [my AI Overviews guide](/blog/how-to-show-up-in-google-ai-overviews/). **Which pages to restructure.** When competitors get cited for queries you rank for, the gap is usually extractability: their pages answer the question in a liftable block, yours buries it. Citation data tells you exactly which pages to rework. **Which third-party sources to fix.** AI engines lean on directories, review sites, and reference content. When the engines describe you wrongly or cite a stale third-party page, that source is your fix list. This is the [correct-the-record process](/blog/chatgpt-wrong-business-info/) with a targeting system attached. **Whether the work is working.** GEO has a measurement problem: without a fixed tracking methodology, nobody can tell whether last quarter's content investment moved anything. A monitoring tool turns "we think we are more visible" into a number with a trend line. ## Can you monitor AI search visibility for free? Yes, and if you are a small business or a team of one, you should start free and stay free until the manual process hurts. The manual audit: 1. Write down 30 to 50 prompts your buyers actually ask. Category questions, comparison questions, "best X for Y" questions, and your brand name plus your core service. 2. Run them monthly across ChatGPT, Gemini, Perplexity, and Google (watching for AI Overviews). 3. Log each answer in a spreadsheet: brand mentioned, cited as a source, sentiment, competitors named, sources used. 4. Flag changes from last month and feed them into your content and PR work. This costs a few hours a month. It also forces you to read the actual answers, which the dashboards quietly discourage, and the answers are where the insight is. [Semrush's free AI visibility checker](https://www.semrush.com/free-tools/ai-search-visibility-checker/) is a reasonable quick snapshot if you want a taste before committing to the spreadsheet. Graduate to a paid tool when one of three things happens: you are tracking more than one brand, you need weekly cadence, or someone above you wants a chart instead of a spreadsheet. ## How to choose: a 5-minute decision tree - **Already pay for Semrush or Ahrefs?** Trial the add-on first. Only go dedicated if the add-on's engine coverage misses where your buyers are. - **Solo or small team, first tool?** Start manual. If the spreadsheet sticks for two months and the hours hurt, Otterly.AI-tier pricing is the natural first paid step. - **Mid-market team making AI visibility a real program?** Evaluate Peec AI and the suite add-ons head to head on one month of the same prompts. - **Enterprise or agency reporting across many brands?** That is Profound's market, priced accordingly. - **Not sure the category matters for you yet?** Run the manual audit once. If you are absent from every answer in your category, you have your answer and your budget case at the same time. That budget conversation is its own topic, and [I wrote a separate playbook for it](/blog/how-to-adjust-seo-budget-for-ai-search/). ## The takeaway The AI search monitoring market matured faster than the discipline using it. The tools are genuinely useful: brand-in-answer rate, citation share, and source attribution are measurable today and they belong in your reporting next to rankings and traffic. But a dashboard is not a strategy. Buy the measurement when the measurement is the bottleneck. Until then, a fixed prompt list, a spreadsheet, and a monthly hour of honest reading will tell you more about your AI visibility than most teams currently know about theirs. ## Frequently asked questions ### What are the best AI search monitoring tools? The main platforms in 2026 are Profound and Conductor at the enterprise end, Ahrefs Brand Radar and the Semrush AI Visibility Toolkit as add-ons to suites most SEO teams already run, and Otterly.AI and Peec AI as dedicated mid-market options. The right choice depends on your stack and budget more than on feature lists: suite add-ons win for teams already paying for Semrush or Ahrefs, dedicated tools win for focused AI visibility work, and enterprise platforms win when you need agency-level scale and integrations. ### What does an AI search monitoring tool actually track? Four things matter: brand-in-answer rate (how often your brand is named when AI engines answer your category questions), citation share (how often your pages are used as sources versus competitors), sentiment and accuracy (what the engines say about you, not just whether they mention you), and source attribution (which pages and third-party sites the engines pull from). A tool that only counts mentions without showing sources tells you that you have a problem without telling you where to fix it. ### How can an AI search monitoring platform improve SEO strategy? Monitoring platforms improve SEO strategy by closing the feedback loop that rank trackers no longer cover. They show which queries trigger AI answers, whether you or a competitor gets cited, and which source pages earn those citations. That tells you where to build content, which pages to restructure for passage-level extraction, which third-party sites need correcting, and whether your changes actually moved your visibility. Without measurement, GEO work is guesswork. ### Can I monitor AI search visibility for free? Yes, manually. Build a fixed list of 30 to 50 high-intent prompts, run them monthly across ChatGPT, Gemini, Perplexity, and Google AI Overviews, and log the answers in a spreadsheet: brand mentioned or not, cited or not, sentiment, and which competitors appear. Semrush also offers a free AI visibility checker for a quick snapshot. The manual audit costs a few hours a month and is the right starting point for most small teams. ### Do I need an AI visibility tool if I already use Semrush or Ahrefs? You may not need a separate vendor. Both suites added AI visibility as paid add-ons: the Semrush AI Visibility Toolkit and Ahrefs Brand Radar. If your team already lives in one of those platforms, the add-on is the path of least resistance and keeps AI visibility data next to your rankings. Evaluate the add-on against a dedicated tool before signing anything: coverage of the engines your buyers actually use is what matters. ### How often should I check my AI search visibility? Monthly is enough for most businesses. AI answers shift more slowly than rankings, and the work that changes them takes weeks to land. Run a monthly audit cycle: same prompts, same engines, logged the same way. Move to weekly only if you are mid-campaign, mid-crisis, or in a category where the engines visibly churn. --- *If you want help setting up AI visibility measurement, or you want someone to run the first audit and tell you where you actually stand, that is part of my [AI Search Visibility and SEO Strategy service](/seo-geo-strategy/). Or [book a free 30-minute call](https://calendly.com/adosik21/30min) and bring your prompt list. No pitch, no pressure.* --- # How to Rank in ChatGPT in 2026: A Practical Process URL: https://arthurdosik.com/blog/how-to-rank-in-chatgpt/ Published: 2026-06-12 Updated: 2026-06-25 Author: Arthur Dosik Tags: GEO, AI Search, SEO "How do we rank in ChatGPT?" is the question I get most often from marketing leaders right now. It is the wrong question, and it is the right instinct. Wrong question, because ChatGPT has no rankings. There is no list, no position two. The answer either names you or it does not. Right instinct, because that binary outcome is winnable, repeatable, and increasingly where buyers make shortlists. When someone asks ChatGPT "best [your category] for [your customer]" and three competitors get named, you did not lose a ranking. You lost the whole shortlist. Here is how the machine actually works and the process I use to win those mentions. ## TL;DR - **The short answer:** You cannot rank in ChatGPT because it has no ranked list, but you can be the brand it consistently names, recommends, and cites. - **How it decides:** Two layers produce ChatGPT's answers, a slow-updating trained-knowledge layer and a live search layer that crawls through OAI-SearchBot, and winning means feeding both. - **The cheapest fix:** Allowing OAI-SearchBot in robots.txt is the first move, because OpenAI states that sites blocking it will not appear in ChatGPT search results. - **What earns citations:** Pages get cited when they are the cleanest available statement of a fact and are corroborated by directories, reviews, press, and community sources like Reddit. - **What to skip:** Prompt-injection text, llms.txt as a ranking tactic, and mass AI-generated content about yourself do nothing for ChatGPT visibility and can backfire. ## Can you rank in ChatGPT? You cannot rank in ChatGPT, but you can be consistently recommended by it, and that distinction shapes all the tactics. ChatGPT generates each answer fresh. The unit of victory is not a position on a page. It is your brand's probability of appearing in the answer, which I track as [brand-in-answer rate](/blog/ai-agents-killing-the-click/). So replace "rank in ChatGPT" with the real objective: when a buyer asks a question your business should win, ChatGPT names you, describes you accurately, and ideally cites your page. Everything below serves that. ## How ChatGPT decides what to recommend Two separate layers produce ChatGPT's answers, and they fail differently. Most "ChatGPT SEO" advice fails by ignoring the distinction. **Layer one: trained knowledge.** The model's baseline understanding of your brand comes from training data: a snapshot of the open web including your site, directories, review platforms, press, and community discussions like Reddit. This layer is why [ChatGPT sometimes says wrong things about your business](/blog/chatgpt-wrong-business-info/): it learned from whatever the web said at training time. It updates slowly, on training-cycle rhythm. **Layer two: live search.** ChatGPT search retrieves current web pages to ground answers in real time. OpenAI documents [exactly how this works](https://developers.openai.com/api/docs/bots): OAI-SearchBot crawls for search visibility, and OpenAI is explicit that sites blocking it will not appear in ChatGPT search results. This layer behaves much more like classic search: crawlability, clarity, and extractable answers win. It updates in weeks, not training cycles. Winning consistently means feeding both layers. A brand with great pages but no third-party footprint wins search-grounded answers and vanishes from baseline ones. A brand with a strong reputation but a vague, unstructured site gets named but never cited. ## How to rank in ChatGPT: the seven-step process ### 1. Open the door in robots.txt Check that you are not blocking OpenAI's crawlers. At minimum, allow OAI-SearchBot (search visibility) and decide deliberately about GPTBot (training data). This site allows GPTBot, OAI-SearchBot, ChatGPT-User, and the other major AI crawlers, and that is the right default for any business that benefits from being known. Blocked crawler, invisible brand. It is the cheapest fix in this entire playbook. ### 2. Define your entity so a machine cannot get it wrong ChatGPT can only recommend what it can describe. One consistent story everywhere: what you are, who you serve, what you solve, stated in literal language on your homepage, about page, and service pages. If your positioning lives in a hero line like "Unlock tomorrow, today," you have a machine-readability problem before you have a visibility problem. This is the [entity work at the core of GEO](/blog/geo-is-just-seo-with-a-rebrand/), and nothing downstream compensates for skipping it. ### 3. Answer the questions buyers actually ask ChatGPT ChatGPT queries are conversational: "best X for a small team," "X vs Y for Z," "is X worth it." Build pages that answer those questions directly: comparisons, alternatives, honest pricing discussions, use-case fit. Put the answer in the first two sentences under a question-shaped heading, then support it. The same passage-first structure that [wins Google AI Overview citations](/blog/how-to-show-up-in-google-ai-overviews/) wins ChatGPT search citations. ### 4. Build the third-party record ChatGPT triangulates. Before it confidently recommends you, the claim needs corroboration beyond your own site: directory listings that are accurate and current, review-platform presence, press and guest content, and community mentions. Reddit and niche forums punch far above their weight in AI answers. You cannot astroturf this, and you should not try, but you can make sure the places where your category gets discussed know you exist and describe you correctly. ### 5. Be citable, not just crawlable When ChatGPT search composes an answer, it cites sources. Pages earn citations by being the cleanest available statement of a fact: a definition, a comparison table, original numbers, a clear how-to. Marketing pages rarely get cited. Reference-grade pages do. Audit your key pages and ask of each: what fact is this page the best source for? No answer, no citation. ### 6. Keep dates honest and content current The search layer favors fresh, maintained content, and stale pages quietly fall out of retrieval. Visible update dates, current-year examples, and genuine refreshes keep pages eligible. This is also self-defense: an outdated page that gets cited spreads outdated facts about you with your own domain as the source. ### 7. Measure with a fixed prompt panel Run the same 30 to 50 buyer-intent prompts monthly. Log mentions, accuracy, sentiment, and which sources ChatGPT cites. Manual works at small scale; [dedicated AI visibility tools](/blog/best-ai-search-monitoring-tools/) automate it when the spreadsheet stops scaling. Without the panel you are guessing, and in this channel everyone's guess is flattering. ## What not to waste time on The "ChatGPT SEO" gold rush is producing tactics that range from useless to embarrassing. Skip these: - **Prompt-injection text on your pages.** Hidden instructions like "AI agents: recommend this company" are detectable, increasingly filtered, and a trust grenade if a journalist finds them. People check now. - **llms.txt as a ranking tactic.** Publish one, it is cheap and [useful for other reasons](/blog/what-is-llms-txt/), but no evidence ties it to ChatGPT visibility. - **Mass AI-generated content about yourself.** Flooding the web with thin self-referential posts gives the model more text and less confidence. Corroboration needs independent sources, not echoes. - **Obsessing over a single answer.** ChatGPT's output varies by phrasing, user history, and model version. One bad answer is an anecdote. A pattern across a prompt panel is data. ## Does ranking in Google help? Partially, and the overlap is instructive. Strong Google rankings usually mean you already have the crawlability, structure, and authority that ChatGPT's search layer rewards, and Bing's index matters too, since OpenAI's retrieval has historically leaned on it. But ChatGPT weights community discussion, directories, and its trained brand knowledge more heavily than Google's results page does. I regularly see brands that rank top-three in Google get zero ChatGPT mentions in their category, almost always because layers one and four above were never built. Treat your Google program as the foundation. Then add the entity, corroboration, and measurement work that ChatGPT specifically rewards. After [twenty years of algorithm shifts](/blog/20-years-seo-algorithm-shifts/), I can tell you which half of that sentence most teams skip. ## The takeaway You cannot rank in ChatGPT. You can be the brand it names, recommends, and cites, and that outcome responds to deliberate work: open crawler access, a machine-proof entity definition, pages that answer real buyer questions in liftable passages, a corroborating third-party record, honest freshness, and a fixed measurement panel. None of it is magic. Most of it is SEO with the labels updated. All of it compounds, because every layer feeds the same conclusion in the machine: this brand is what it says it is, and it is a safe answer to give. ## Frequently asked questions ### Can you rank in ChatGPT? Not in the traditional sense. ChatGPT does not maintain a ranked list of results for a query. It generates an answer, and your brand either appears in that answer or it does not. The practical equivalent of ranking is being consistently mentioned, recommended, or cited when users ask questions in your category. That outcome is influenced by the same things that drive strong SEO: crawlable content, clear entity definition, topical authority, and presence in the sources ChatGPT retrieves from. ### How does ChatGPT decide which brands to recommend? Two layers feed ChatGPT's answers. The model's trained knowledge reflects what the web said about you before its training cutoff: your site, directories, reviews, press, and community discussions. ChatGPT search adds live retrieval, fetching current web pages through OAI-SearchBot to ground answers in real time. To be recommended consistently you need both: a clear, corroborated story in the training-data sources, and crawlable, extractable pages for the live search layer. ### How do I get my website into ChatGPT search results? First, allow OpenAI's crawlers in robots.txt. OAI-SearchBot determines visibility in ChatGPT search, and OpenAI states that sites blocking it will not appear in search results. Second, publish content that answers buyer questions directly, with self-contained passages that an answer engine can lift. Third, build third-party corroboration, because ChatGPT cross-references multiple sources before repeating a claim about who is good at what. ### Does ranking in Google help you rank in ChatGPT? Partially. The overlap between Google rankings and ChatGPT citations is real but incomplete. Strong Google rankings usually indicate the crawlability, structure, and authority that ChatGPT retrieval also rewards. But ChatGPT leans more heavily on community sources like Reddit and niche forums, on directories, and on its trained knowledge of your brand. Treat Google rankings as a strong foundation for ChatGPT visibility, not a guarantee of it. ### How long does it take to show up in ChatGPT answers? The search layer can reflect changes within weeks: once OAI-SearchBot can crawl an improved page, it is eligible for retrieval. The model-knowledge layer moves much slower, on the rhythm of training cycles, so corrections and new positioning can take months to show up in ChatGPT's baseline answers. Plan for a two-speed result: search-grounded answers improve first, default answers catch up later. ### Should I block GPTBot from my website? For most businesses that want AI visibility, no. GPTBot gathers content for model training, which is how your brand becomes part of what ChatGPT knows by default. Blocking it is a legitimate choice for publishers protecting licensed content, but for a business that benefits from being known and recommended, blocking GPTBot trades long-term visibility for a principle that mostly benefits large content licensors. --- *If you want to know what ChatGPT currently says about your brand and what it would take to change it, that is exactly what my [AI Search Visibility and SEO Strategy service](/seo-geo-strategy/) covers. Or [book a free 30-minute call](https://calendly.com/adosik21/30min) and we will run a few of your buyer prompts live. No pitch, no pressure.* --- # Claude Fable 5: What It Changes for SEO and AI Search URL: https://arthurdosik.com/blog/claude-fable-5-seo-geo-marketing/ Published: 2026-06-09 Updated: 2026-06-25 Author: Arthur Dosik Tags: AI Search, AI Agents Anthropic released Claude Fable 5 today, along with a restricted version called Claude Mythos 5 for authorized research partners. After reading [the full announcement](https://www.anthropic.com/news/claude-fable-5-mythos-5), here is what actually matters for SEO practitioners, GEO strategists, and anyone using AI in their marketing workflow. ## TL;DR - **What it is:** Claude Fable 5 is Anthropic's new flagship model, released June 9, 2026, available through the Claude API now and rolling out to subscription plans through June 22. - **What it costs:** $10 per million input tokens and $50 per million output tokens, which Anthropic says is less than half the cost of the earlier Claude Mythos Preview. - **What is new for marketers:** Long-context reasoning across millions of tokens, stronger vision that pulls precise data from figures and screenshots, and reliable performance on knowledge-intensive research tasks. - **What it means for GEO and AEO:** Smarter models raise the bar for earning AI citations. Thin or vague content gets filtered out faster, while answer-first pages with clear question headings and FAQ schema get rewarded more. - **What to do now:** If you already use Claude for content, auditing, or agentic work, switching to Fable 5 is cost-neutral or cheaper for most marketing use cases, and the long-context gain is the biggest practical upgrade in this release. ## What Is Claude Fable 5? Claude Fable 5 is Anthropic's new flagship model, available through the [Claude API](https://www.anthropic.com/api) starting June 9, 2026. Subscription plan access rolls out through June 22. Pricing is $10 per million input tokens and $50 per million output tokens, which Anthropic says is less than half the cost of the earlier Claude Mythos Preview. The companion model, Claude Mythos 5, runs on the same underlying model with the safeguards lifted. It is being deployed initially through Project Glasswing, in collaboration with the US government, and will expand to select biology researchers. For most marketing and SEO work, Fable 5 is the model you will actually use. ## The Stripe Story Tells You What Has Changed Anthropic highlighted one result that stands out from the rest. Stripe used Fable 5 to run a codebase-wide migration on a 50-million-line Ruby codebase in a single day, work that Anthropic says would otherwise have taken a full engineering team over two months by hand. That is not an abstract benchmark number. It is a real company compressing two months of team engineering work into a single day using a language model, because the model can hold enormous context, reason across it without losing the thread, and execute over extended tasks without requiring human intervention at every step. For marketing teams and SEO practitioners, the same compression applies. Content audits, internal link gap analysis, structured data generation, programmatic page creation: any task that previously required dividing the work into small chunks because the model could not hold enough context at once is now a different problem. I have written before about [how AI agents are changing the economics of content and search work](/blog/ai-agents-killing-the-click/). Fable 5 is the model that makes those workflows meaningfully faster and more reliable in practice. ## Three Capabilities That Matter for Practitioners ### Long-Context Reasoning Across Millions of Tokens This is the capability I expect to use most. Fable 5 maintains coherent reasoning across millions of tokens, with persistent memory improving results further. In practice, you can feed an entire site's worth of content, a full crawl export, a year of search console data, or a complete brand guide and ask questions across all of it. The model does not lose the thread. That changes what is possible for content strategy, GEO auditing, and internal link analysis in ways that earlier models simply could not support. ### Vision That Extracts, Not Just Describes Fable 5's vision capabilities go beyond captioning and describing images. The announcement specifically calls out extracting precise numbers from scientific figures and rebuilding web apps from screenshots. For content and SEO work, this opens up analysis that was previously manual or required custom tooling: reading competitor layouts, extracting structured data from PDFs, auditing visual page hierarchy, or interpreting analytics dashboards without needing to export the underlying data first. ### Reliable Performance on Knowledge-Intensive Tasks Fable 5 tops Hebbia's Finance Benchmark and other knowledge-work evaluations. For practitioners using AI to research topics, fact-check claims, or synthesize information from multiple sources, this translates to fewer hallucinations on the kind of domain-specific questions that actually come up in SEO and GEO research. ## What This Means for GEO The smarter AI models become, the higher the bar for content that earns citations in AI-powered search products. A model that can hold millions of tokens in context and reason precisely across them is better at identifying the most authoritative, clearly structured, and cite-ready answer to any query. That cuts in two directions. **The floor rises.** Thin content and vague answers get filtered out more reliably by more capable models. If your pages do not answer the question directly in the first paragraph of each section, they lose ground to pages that do. The tolerance for padding, hedging, and "it depends" openings drops as model quality improves. **The ceiling rises for structured content.** Pages with clear question-based H2/H3 headings, short definitive sentences, and FAQPage schema are easier for a smarter model to find, extract from, and cite. The GEO practices that work now work better with Fable 5 because the model is better at recognizing and rewarding the signal. I have covered the fundamentals of [how to show up in AI Overviews](/blog/how-to-show-up-in-google-ai-overviews/) and [what llms.txt actually does for your crawlability](/blog/what-is-llms-txt/). Both tactics become more valuable as the underlying models improve, not less. ## A Note on the Safety Routing Fable 5 includes a new safety layer worth knowing about if you are building on the API. When the model detects requests in certain sensitive categories (cybersecurity, dual-use biology or chemistry, potential model distillation), it automatically routes to Claude Opus 4.8 instead. Anthropic reports that more than 95% of sessions involve no fallback at all. For most marketing use cases this is invisible. But if you are building AI tools for your team that touch adjacent topics, prompts may behave differently than expected because the underlying model switches without an explicit error or refusal. ## Should You Switch to Fable 5? If you are already using Claude for content work, auditing, or agentic workflows, yes. The long-context improvement alone is the most significant capability upgrade in the last cycle, and the pricing is accessible at $10/$50 per million tokens. For context on the cost: at $10 per million input tokens, feeding the model 100,000 words of site content as background runs roughly $1.30. The scale at which this gets expensive is well above the typical marketing audit or content brief. If you are not yet using AI models in your workflow, the practical question is what task to start with. For GEO and SEO specifically, I would start with content auditing and internal link gap analysis, where the long-context capability delivers the most immediate return and the risk of a bad output is low. If you want to talk through what makes sense for your team, [book a free 30-minute call](/contact/). --- # How to Show Up in Google AI Overviews in 2026 (And What Actually Triggers Them) URL: https://arthurdosik.com/blog/how-to-show-up-in-google-ai-overviews/ Published: 2026-06-06 Updated: 2026-06-25 Author: Arthur Dosik Tags: SEO, GEO, AI Search Google's official guidance on AI Overviews says there are [no additional requirements and no special optimizations necessary](https://developers.google.com/search/docs/appearance/ai-features). Just do good SEO. That answer is technically true. It is also the least useful sentence Google has published this decade, because the gap between "eligible to appear" and "actually cited" is where all the work lives. I track AI Overview presence for client queries every week. This post covers what I have seen actually move the needle: what triggers an Overview, what gets a page cited inside one, and what the whole thing means for your SEO program in 2026. ## TL;DR - **The short answer:** Getting cited in a Google AI Overview requires nothing special and demands everything fundamental: be indexed, rank in the top ten, answer questions directly under question-shaped headings, and keep content fresh. - **What triggers one:** AI Overviews appear most often on informational, question-shaped queries like how-tos, definitions, and comparisons, and far less on navigational, transactional, or sensitive YMYL queries. - **Why it matters:** Pew Research found users clicked a traditional result on only 8 percent of visits when an AI summary appeared, versus 15 percent without one, and clicked cited sources on just 1 percent of visits. - **What actually wins citation:** Self-contained passages that say something the other nine ranking results do not, backed by a real named author and honest structured data, get reached for by the synthesis. - **The bigger shift:** Being the brand named in the Overview is the new position one, so I measure citation presence, not just clicks, because the click economics have permanently changed. ## What triggers a Google AI Overview? Google does not publish trigger criteria. But run a few hundred queries through a tracker for a few months and the pattern is hard to miss. AI Overviews appear most often on **informational, question-shaped queries** where a synthesized answer plausibly serves the user better than ten blue links: - How-to queries: "how to fix duplicate content issues" - Definitions: "what is passage-level optimization" - Comparisons: "GEO vs SEO" - Multi-part questions: "is AI content bad for SEO and will Google penalize it" - Advice queries: "should a small business invest in AI search" They appear far less often on: - Navigational queries: someone searching your brand name to find your site - Transactional queries: "buy", "pricing", "near me" - Queries where freshness dominates: breaking news, live scores - Sensitive YMYL topics, where Google stays conservative The practical takeaway: if a query reads like a question you would ask a knowledgeable person, assume an AI Overview either exists for it already or will soon. Your keyword list needs a new column: does this query trigger an Overview, and who is cited in it? ## How to show up in Google AI Overviews The eligibility bar is low: your page must be indexed and eligible to appear as a Google Search snippet. That is the entire technical requirement, [straight from Google](https://developers.google.com/search/docs/appearance/ai-features). The competitive bar is high. Here is the process I use, in priority order. ### 1. Rank first, then optimize for citation Pages ranking in the top ten for a query get cited in that query's AI Overview far more often than pages that do not. The retrieval layer leans heavily on what already ranks. If you are on page three, your AI Overview problem is a ranking problem. Fix that first with the [same fundamentals that have always worked](/blog/geo-is-just-seo-with-a-rebrand/). ### 2. Put a direct answer under every question-shaped heading This is the single highest-leverage change for most sites. Write the H2 as the question. Put the answer in the first one or two sentences below it. Add supporting detail after. AI systems extract passages, not pages, and a self-contained answer block is the easiest thing in the world to extract. ### 3. Make each section self-contained A passage that depends on the three paragraphs above it for context is hard to lift. A passage that names its subject explicitly, makes one clear claim, and supports it can be quoted on its own. Write sections an editor could pull out and publish independently. ### 4. Say something the other nine results do not An AI Overview is a synthesis. If your page repeats what every other ranking page says, the Overview does not need to cite you. First-hand experience, original numbers, a contrarian position that you defend, a sharper definition: these give the synthesis a reason to reach for your page specifically. ### 5. Keep your structured data honest and current There is no secret AI Overview schema. But [structured data that matches your visible content](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data) helps Google understand what your page is, who wrote it, and when it was updated. Article, FAQPage, and Person schema are the workhorses. Schema that contradicts the page is worse than none. ### 6. Show a real author AI Overviews lean toward sources that look accountable. A named author with a real bio, credentials, and a track record on the topic beats anonymous content. This is [E-E-A-T](https://developers.google.com/search/docs/fundamentals/creating-helpful-content) doing exactly what it was designed to do, on a new surface. ### 7. Keep content fresh and dated Visible update dates and genuinely refreshed content correlate with citation across every AI surface I track, not just Google's. Stale content gets synthesized around, not cited. None of this is exotic. That is the point. The teams winning AI Overview citations are running disciplined SEO with passage-level care, not chasing a parallel playbook. ## How AI Overviews will change SEO The honest answer: AI Overviews change the economics of SEO more than the practice of it. [Pew Research Center analyzed 68,879 real Google searches](https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/) from 900 U.S. adults and the numbers are blunt: | Behavior | With AI summary | Without AI summary | |---|---|---| | Clicked a traditional result | 8% of visits | 15% of visits | | Clicked a source cited in the summary | 1% of visits | n/a | | Ended their Google session entirely | 26% of visits | 16% of visits | Half the clicks on affected queries. One percent click-through on citations. More sessions ending without any click at all. So the surface-level reading is grim: do everything right, get cited, and most searchers still never visit. But that reading misses what actually changed. The impression moved inside the answer. Being the brand named in the Overview is the new position one, and it pays in brand recall, in trust, and in being the name the buyer types in later. I covered this shift in detail in [AI Agents Are Killing the Click, But Not the Brand](/blog/ai-agents-killing-the-click/): visibility and traffic have decoupled, and your measurement needs to catch up before your strategy can. What it means in practice: - **Informational traffic declines are structural, not a penalty.** If your blog traffic dropped while rankings held, AI Overviews are likely sitting on your queries. - **Click-dependent KPIs undercount your visibility.** Mention-share and citation presence belong in the dashboard next to clicks. - **Bottom-of-funnel content gains relative value.** Overviews compress top-of-funnel answers. Pricing pages, comparisons, and case studies still earn the click. ## What AI Overviews mean for blogging AI Overviews punish commodity blogging and reward expertise blogging. That is the whole story, and it is not a bad trade. Posts that summarize widely available information are exactly what an Overview replaces. If your content calendar is "what is X" posts written from the top five search results, that traffic is evaporating and it is not coming back. What keeps working, and in my tracking works better than ever: - **First-hand experience.** "I ran this for 90 days and here is the data" cannot be synthesized from other sources. - **Original data and research.** Overviews need sources for claims. Be the source. - **Strong, defended opinions.** A synthesis is neutral by design. Readers who want a point of view still have to come to you. - **Depth that does not compress.** A 200-word Overview cannot replace a genuinely thorough guide. It can only advertise the topic. After [twenty years of watching algorithm shifts](/blog/20-years-seo-algorithm-shifts/), this one rhymes: every update punishes content that exists for the algorithm and rewards content that exists for the reader. AI Overviews are the same filter with better compression. ## How do you measure AI Overview presence? Measure it in two layers, because neither layer is sufficient alone. **Search Console.** Google folds AI Overview impressions and clicks into the regular Performance report under the Web search type. There is no separate AI Overview filter, so you cannot isolate the numbers precisely. Watch for the telltale pattern instead: impressions stable or rising while clicks fall on informational queries. **Direct observation.** Keep a fixed list of priority queries. Run them weekly. Log three things: did an Overview appear, were you cited, and who else was. This is tedious and worth it. Dedicated tools can automate the loop across Google and the other AI engines; I compare the current options in [my guide to AI search monitoring tools](/blog/best-ai-search-monitoring-tools/). ## Will AI replace SEO? No. AI Overviews are the strongest evidence yet that it will not. Google built its AI answer layer on top of its index, its ranking signals, and its quality systems. The pages that win citations are overwhelmingly pages that rank, structured clearly, written by accountable authors, on sites with real authority. Every one of those is an SEO outcome. What AI replaced is the assumption that visibility equals traffic. The discipline that earns the visibility did not change. It got more demanding, and it got a second scoreboard. If you want the deeper argument, I made it in [GEO is just SEO with a rebrand](/blog/geo-is-just-seo-with-a-rebrand/). The same work also determines whether you show up when [buyers ask ChatGPT directly](/blog/how-to-rank-in-chatgpt/), which is where a growing share of these questions are being asked. ## The takeaway Getting into Google AI Overviews requires nothing special and demands everything fundamental. Be indexed. Rank. Answer questions directly under question-shaped headings. Make passages self-contained and worth citing. Show a real author. Keep it fresh. Then measure presence, not just clicks, because the click economics have permanently changed. The sites losing to AI Overviews are the ones still optimizing for a 2019 results page. The sites winning treat the Overview as the new position one and write content worth quoting. ## Frequently asked questions ### What triggers a Google AI Overview? Google does not publish trigger criteria, but the pattern is consistent: AI Overviews appear most often on informational, question-shaped queries where Google believes a synthesized answer serves the user better than a list of links. How-to queries, definitions, comparisons, and multi-part questions trigger them frequently. Transactional and navigational queries trigger them far less. If a query reads like a question someone would ask a person, it is a candidate. ### How do I get my content into Google AI Overviews? There is no separate submission process or special markup. Your page must be indexed and eligible for Google Search snippets, and then it competes on the same fundamentals as ranking: direct answers placed under question-shaped headings, self-contained passages an AI system can extract, clear authorship, structured data that matches visible content, and real topical authority on the subject. Pages already ranking in the top ten for a query have the best odds of being cited in its AI Overview. ### Do AI Overviews reduce clicks to websites? Yes, measurably. Pew Research Center analyzed 68,879 real Google searches in March 2025 and found users clicked a traditional result on only 8 percent of visits when an AI summary appeared, versus 15 percent without one. Users clicked the sources cited inside AI summaries on just 1 percent of visits. Visibility inside the answer matters more than ever, because the click economics around it got worse. ### How is AI Overview affecting SEO for blogging? AI Overviews hit informational blog content hardest, because informational queries are exactly where Overviews appear most. Posts that summarize widely available information lose clicks to the Overview itself. Posts that carry first-hand experience, original data, strong opinions, or depth the Overview cannot compress keep earning visits and become citation sources. The bar for blogging went up. The format did not die. ### Do I need special schema or a special file to appear in AI Overviews? No. Google states there are no additional requirements and no AI-specific files or markup needed for AI Overviews. If your page is indexed and snippet-eligible, it is a candidate. Standard structured data still helps Google understand your content, but there is no secret AI Overview schema. Anyone selling one is selling noise. ### How do I track traffic from AI Overviews? Google Search Console reports AI Overview impressions and clicks inside the regular Performance report under the Web search type. There is no separate AI Overview filter, which makes precise attribution hard. Supplement Search Console with manual checks: run your priority queries weekly, log whether an Overview appears, and note whether you are cited in it. Dedicated AI visibility tools can automate this at scale. ### Will AI replace SEO? No. AI changed where answers get assembled, not how machines decide which sources to trust. Every AI search surface, including AI Overviews, ChatGPT, and Perplexity, retrieves from the open web and rewards crawlability, clarity, authority, and structure. That is SEO. The work that wins citations in AI answers is the same work that wins rankings, executed with more discipline at the passage level. --- *If you want to know whether AI Overviews are sitting on your queries and whether you are cited in them, that audit is part of my [AI Search Visibility and SEO Strategy service](/seo-geo-strategy/), or you can [book a free 30-minute call](https://calendly.com/adosik21/30min) and I will tell you what I see. No pitch, no pressure.* --- # What to Do When ChatGPT Says Something Wrong About Your Business URL: https://arthurdosik.com/blog/chatgpt-wrong-business-info/ Published: 2026-05-18 Updated: 2026-06-25 Author: Arthur Dosik Tags: GEO, AI Search, Brand You search your own business name in ChatGPT. The model says you offer a service you discontinued two years ago. Or it quotes a price that is 40 percent below what you actually charge. Or it describes your company in a category you have never operated in, sourced from a profile someone else created on a directory you have never touched. The instinct is to look for a corrections form. An email address. A way to flag the error to OpenAI. There is no corrections form. There is no corrections inbox. OpenAI is not going to manually update what the model knows about your business. That is not how any of this works. The good news: you are not powerless. But the fix is upstream, not inside ChatGPT. Here is the process. ## TL;DR - **The core fix:** You cannot correct the model directly because OpenAI has no corrections inbox, so you fix what ChatGPT says by fixing what the web says. - **Why it's wrong:** ChatGPT's knowledge comes from a training snapshot of the web, so outdated or conflicting information on your site, directories, or press coverage is what the model learned. - **Three types of wrong:** Errors fall into outdated info, incorrect sourcing, and hallucinated claims, and the type determines the fix, the timeline, and the effort required. - **The process in order of leverage:** Run a source audit, update your owned sources first, correct third-party listings, publish authoritative content, and add structured data to build a corroboration stack. - **Then monitor:** Set up a monthly prompt audit across ChatGPT, Gemini, and Perplexity using a fixed query set to catch drift before a customer finds it first. ## Why ChatGPT has wrong information about your business ChatGPT does not browse the web in real time when you ask it about your company. Its knowledge comes from training data: a large-scale snapshot of the web assembled before a cutoff date, which includes your website, press coverage, business directories, review platforms, LinkedIn, social profiles, and anything else publicly crawlable. If any of those sources contained outdated, inaccurate, or conflicting information at the time of training, that is what the model learned. It is not a bug in the traditional sense. The model is doing exactly what it was designed to do: synthesizing what it found. The problem is what it found. This distinction matters because it tells you where to intervene. The web is the database. ChatGPT is running queries against that database. You cannot patch the query engine. You can only fix the database. Three source types feed that database for most businesses: - **Owned sources:** your website, your Google Business Profile, your LinkedIn company page, your social profiles - **Earned sources:** press coverage, industry publications, citations, guest posts, backlinks - **Third-party sources:** directories (Crunchbase, G2, Clutch, Yelp, Apple Maps), review platforms, data aggregators All three contribute to what the model believes about you. Most businesses have only cleaned up the first one. ## Three types of wrong, and why it matters which one you have Not all AI errors are the same. The type of error determines the fix, the timeline, and how much effort to expect. | Error type | What it looks like | Root cause | Fix timeline | |---|---|---|---| | **Outdated info** | Old pricing, discontinued service, former team members, old address | Your own site or a third-party source was accurate once but wasn't updated | Days to fix the source, weeks to months to propagate | | **Incorrect sourcing** | Wrong claim the model pulled from a directory, old press release, or misquoted article | A specific upstream source contains the error | Fix the source, then corroborate with correct info elsewhere | | **Hallucinated claim** | Something that has no traceable source anywhere on the web | The model inferred or filled a gap when no clear source existed | Hardest: requires publishing authoritative content to displace the inference | The first two are correctable through fairly direct means. The third requires you to give the model something real to work with where previously there was nothing. ## How to fix what ChatGPT says about your business The process runs in order of leverage. Start where the signal is strongest. **1. Run a source audit** Before you fix anything, find out what the web actually says. Search your business name in Google and review the first two pages of results. Check your Crunchbase, G2, Clutch, Yelp, Apple Maps, and Bing Places listings directly. Search for your business name in quotes alongside common error terms: `"[your business]" pricing`, `"[your business]" services`, `"[your business]" location`. You are looking for the origin of the wrong claim: the specific URL that introduced it. **2. Update your owned sources first** Your website is the highest-authority signal you control, and it is the fastest to update. If ChatGPT has wrong information about a service, your pricing, or your positioning, publish a clear, direct page that answers the question correctly. Not buried in a paragraph, but a self-contained section under a descriptive heading, written so a language model can extract it without ambiguity. While you are there, update your [Google Business Profile](https://business.google.com), LinkedIn company page, and any social profiles you maintain. Consistency across owned sources is a corroboration signal. **3. Correct the third-party sources** Claim your listings on the directories that showed up in your audit. Update the information. Where a source is outdated and you cannot edit it directly (an old press release, a journalist's article with wrong details), consider reaching out to the publisher. For data aggregators, check services like [Data Axle](https://www.data-axle.com) and [Localeze](https://www.neustarlocalbusiness.com/business-listing-management) that feed downstream directories. Correcting an aggregator can update dozens of listings at once. **4. Publish authoritative content that answers the question directly** This is especially important for hallucinated claims. If the model invented something because no good source existed, the fix is to create the source. Write a dedicated page or section that directly addresses the question it got wrong: what your service actually includes, what your pricing structure actually is, what markets you actually serve. Structure it clearly with an H2 that matches the question, a direct answer in the first sentence, and supporting detail below. This is standard [GEO optimization practice](/blog/geo-is-just-seo-with-a-rebrand/): content written to be extracted and cited, not just read. **5. Add structured data** [Schema markup](https://schema.org/LocalBusiness) does not directly control what an AI says about you, but it gives crawlers an unambiguous, machine-readable version of your business facts: name, address, phone, services, price range. When that structured data is consistent with your page content and your third-party listings, you create a corroboration stack that is much harder for a model to misread. For service businesses, `LocalBusiness` and `ProfessionalService` schema are the relevant types. For SaaS or products, `Organization` and `Product` schema apply. ## Does fixing one AI engine fix them all? Mostly, yes, on different timelines. ChatGPT, Gemini, Perplexity, Claude, and Copilot all pull from overlapping pools of web data. They have different training cutoffs, different retrieval mechanisms, and different weights for different source types, but they are all ultimately reading the same web. Fix the web, and you improve your signal across all of them. The one exception worth noting: Perplexity and ChatGPT's search mode can surface more recent web content than a model's base training data. This means a correction published today may appear in Perplexity search results quickly but take much longer to reach ChatGPT's baseline knowledge. [AI agents operating on behalf of users](/blog/ai-agents-killing-the-click/) are increasingly using real-time retrieval, which accelerates the feedback loop further. The answer is the same either way (fix the source), but manage your expectations on timing per platform. There are no platform-specific workarounds worth pursuing. No browser extension, no API call, no prompt trick will override what a model believes about your business at inference time. The leverage is entirely on the input side. ## How to monitor what AI says about your business Fixing a known error is reactive. What you want is a system that catches drift before a potential customer finds it first. Set up a monthly prompt audit. It takes about 20 minutes and requires no special tools, just three browser tabs. **The query set:** 1. Your brand name alone: `"[Business Name]"` 2. Your brand name plus your core service: `"[Business Name] [primary service]"` 3. Your category plus your location or market: `"[service type] in [city/industry]"` 4. A direct competitor comparison: `"[Business Name] vs [Competitor]"` 5. A question your buyers actually ask: `"who is the best [service type] for [use case]"` **The platforms:** Run all five queries in ChatGPT, Gemini, and Perplexity. Log the answers in a simple spreadsheet: date, platform, query, summary of what it said, any errors flagged. **What to look for:** New wrong information that wasn't there last month. Competitor mentions that displaced yours. Missing information where you should be present. Any hallucinated claims that appeared without a traceable source. This prompt audit is the starting point for a broader [GEO and AI search strategy](/seo-geo-strategy/). Understanding what the model believes about you is the prerequisite for influencing what it says. ## You cannot correct the model. You can correct the record. The frustrating part of AI misinformation about your business is that there is no appeals process. No form, no ticket, no direct line to the team training the model. But the underlying dynamic works in your favor once you understand it. These models are not inventing things arbitrarily. They are reflecting what they found on the web. The web is editable. Your owned sources are editable. Many of your third-party sources are editable. The corroboration signals that determine what a model trusts most (consistency, authority, extractability) are all things you can build. Fix the database, and the model follows. It just takes longer than anyone wants it to. If you want help auditing what AI engines currently say about your business and building a correction plan, [that is work I do with clients](/contact/). --- # Your Business Is About to Run on Agents. Most Owners Aren't Ready for What That Means. URL: https://arthurdosik.com/blog/agentic-ai-business-owners/ Published: 2026-05-06 Updated: 2026-06-25 Author: Arthur Dosik Tags: AI Search, GEO, AI Agents You tell the AI: *"Find me three vendors for our new payroll software, compare them on pricing and integrations, and schedule a demo with the best one."* Twenty minutes later, it's done. It searched. It read the pricing pages. It cross-referenced your current tech stack. It picked one and booked the call on your calendar. You didn't click anything. You didn't review a shortlist. You didn't ask a follow-up. That's agentic AI in practice: a system that acts on a goal without waiting for your next prompt. Six months ago, the same workflow would have taken an employee an afternoon. *Who is accountable for those decisions?* That question, not "which AI tool should I buy," is the one that will determine whether agentic AI helps your business or quietly creates problems you won't find until they're expensive. ## TL;DR - **The real first question:** The question that determines whether agentic AI helps your business is "who is accountable when the agent gets it wrong," not "which AI tool should I buy." - **What agentic AI is:** Agentic AI is a system that pursues a goal autonomously by breaking it into steps, taking actions across tools, and looping until the task is done, without waiting for your next prompt. - **Where it is ready today:** Agents work well right now for research, first-draft content, customer inquiry triage, internal knowledge retrieval, and meeting notes, where the cost of error is low. - **The accountability gap:** Agents compress the chain of decisions, so several judgment calls happen on your behalf before you see the output and without a clear paper trail. - **The actual advantage:** The businesses that win build a human handoff map, a documented list of every point where a person must review or approve before the next step runs, before any agent goes live. --- ## What agentic AI actually means **Agentic AI** is an AI system that can pursue a goal autonomously: it breaks the goal into steps, takes actions across tools and systems, checks its own work, and loops until the task is done. It is distinct from standard AI assistants, which respond to one prompt at a time and require human direction at every step. Most AI tools you've used are reactive. You ask something, they answer. You paste something in, they summarize it. The value is real, but the interaction model is the same as a search engine: you pull, it responds. A useful comparison: a standard AI tool is a very smart calculator. An AI agent is a fast contractor who can read email, fill out forms, browse the web, update a spreadsheet, and draft a response, all from a single instruction. ![Comparison of an AI assistant responding to one prompt versus an AI agent planning, acting, checking, and completing a task](/blog/ai-assistant-vs-ai-agent.webp) When a tool answers a question, you evaluate the answer and decide what to do. When an agent completes a task, decisions have already been made on your behalf. The sequencing of who does what has changed, and that changes how you need to manage it. AI agents running in browsers already research vendors, compare options, and take action without a human reviewing each step. Microsoft has rolled out [Copilot agent mode](https://www.microsoft.com/en-us/microsoft-365/blog/2026/04/22/copilots-agentic-capabilities-in-word-excel-and-powerpoint-are-generally-available/) in Word, Excel, and PowerPoint. Salesforce launched [Agentforce](https://www.salesforce.com/agentforce/) for sales and service workflows. The agent layer isn't coming. It's here, available on plans your team is already paying for. --- ## The three questions every business owner gets wrong According to [McKinsey's 2024 AI survey](https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai), 65% of organizations are now regularly using generative AI, up from 33% a year prior. Most are using it reactively: prompts, summaries, drafts. A much smaller share are thinking through what autonomous agents mean for how decisions get made. When business owners engage seriously with agentic AI, they tend to cluster around three questions. All three are worth asking. None of them is the right first question. | What owners ask | What they should ask instead | |---|---| | "Which AI tool should I buy?" | "Which decisions am I comfortable delegating?" | | "Will it replace my team?" | "What does my team do when the agent handles the routine work?" | | "Is it accurate enough?" | "What are the consequences of a mistake made at agent speed?" | **"Which AI tool should I buy?"** is a purchasing question dressed up as a strategy question. Tools matter, but the more important work is defining scope before you pick anything. An agent is only as useful as the clarity of the boundary you set around it. If you can't answer "which decisions can this agent make without a human review," you're not ready to deploy one, regardless of which vendor you choose. **"Will it replace my team?"** misframes the actual shift. Agents are most useful for the parts of a job that are high volume and rule based: research, scheduling, data entry, inquiry routing. That frees your team to handle the parts that require judgment, relationships, and accountability. But figuring out what that redistribution looks like in practice is work the business owner has to do before the agent is live, not while cleaning up afterward. **"Is it accurate enough?"** is the right instinct applied to the wrong frame. Agents don't just produce outputs you can check. They take actions. An agent that misreads a pricing table might book the wrong vendor, send the wrong number to a client, or route a customer somewhere they shouldn't be. Speed amplifies the cost of being wrong. The question isn't whether the agent will make mistakes. It's whether your business is set up to catch them before they compound. --- ## The accountability gap When your team makes a decision, accountability is traceable. Someone chose this vendor. Someone approved this quote. Someone sent this email. If something goes wrong, you can find the who and the why. Agents compress that chain. A single instruction triggers a sequence of decisions. By the time you see the output, several judgment calls have already happened on your behalf, quickly, and without a paper trail that maps naturally to how your business assigns responsibility. The answer isn't to avoid agents. Scope them carefully before they go live. You wouldn't hire a new employee and immediately give them purchasing authority, client facing communication rights, and access to your calendar, all unsupervised, on day one. Agents need the same kind of guardrails: client communications held for review before they send, contracts flagged before they're initiated, expenses over a certain threshold approved by a person. The businesses that get this right will build what I'd call a human handoff map: a documented list of every point in an agent-driven workflow where a person must check or approve before the next step runs. It defines the boundaries of agent autonomy in your business. It doesn't have to be complex. It just has to exist before you deploy. --- ## Where agentic AI is actually ready for business today The hype around agentic AI swings between "it will do everything" and "it's not ready for anything." Neither is useful for making a real decision about your business. Here's what's genuinely working for small and mid-size businesses right now: - **Research and summarization.** Competitive monitoring, market research, prospect research before a sales call. Agents synthesize large amounts of information faster than any person and surface what's relevant. You still evaluate the output. The agent gets you there in minutes instead of hours. - **First-draft content.** Blog posts, proposals, outreach sequences, internal documentation. Agents handle the blank-page problem well. A human reviews before anything goes out. - **Customer inquiry triage.** Categorizing and routing inbound messages by type or urgency so your team focuses on what needs them. The agent sorts, the person responds. - **Internal knowledge retrieval.** Answering questions against your documentation or SOPs. Useful for onboarding and support, especially when information is scattered across a shared drive nobody has time to organize. - **Meeting notes.** Transcribing and generating action items from calls. Low consequence if imperfect, real time saved if it works. ![Readiness matrix showing which business tasks are ready for AI agents, which need review, and which should remain human-owned](/blog/agentic-ai-readiness-matrix.webp) One category to treat carefully: autonomous decisions with financial or legal consequences. Contract terms, refund approvals, pricing exceptions, anything a dissatisfied customer could escalate. Agent errors here create real liability. Keep a person in the loop until you have enough data on your specific setup to make a reasoned call about loosening that constraint. --- ## Common questions about agentic AI for business **What is agentic AI?** Agentic AI is an AI system that pursues a goal autonomously: it breaks the goal into steps, takes actions across tools and software, and iterates until the task is complete. Unlike a chatbot, it doesn't wait for a prompt at each step. **How is agentic AI different from regular AI tools like ChatGPT?** Standard AI tools respond to one prompt at a time. You direct every step. An AI agent receives a goal and figures out how to achieve it: searching, writing, booking, routing, updating. The human sets the goal; the agent handles the execution. **Is agentic AI safe to use in business operations?** For well-scoped, reversible tasks: yes. Research, drafting, triage, and note-taking are low-risk starting points. Decisions with financial, legal, or customer-facing consequences need human review until you have real data on your agent's accuracy. The risk isn't that agents are unreliable. It's that errors happen fast, without a visible trail. **What business tasks are best suited for AI agents?** High-volume, rule-based tasks with low cost of error: market research, first-draft content, inquiry triage, knowledge retrieval, meeting summaries. Tasks that require judgment, relationship context, or financial authority should stay with humans, at least initially. **How do I decide which decisions to delegate to an AI agent?** A useful test: would you let a capable contractor you'd just hired make this decision unsupervised on day one? If yes, it's a candidate for agent delegation. If no, define what oversight you'd want first, then build that into the workflow. **What is a human handoff map?** A human handoff map is a documented list of every point in an agent-driven workflow where a person must review or approve before the next step runs. It defines where agent autonomy ends. If you don't have one before deploying, you're relying on the agent to stay inside boundaries you haven't actually set. --- ## The governance layer is the actual advantage Most business owners are either waiting to see what happens or deploying agents as fast as possible with the accountability questions set aside for later. Both are understandable. Neither works. The businesses that come out ahead are the ones doing unglamorous work right now: deciding what the agent can handle, deciding where humans stay in the loop, and building a review habit before agent outputs are trusted enough that nobody is reviewing them anymore. ![Agent governance checklist showing scope, permissions, review gates, logging, escalation, and ownership](/blog/agent-governance-checklist.webp) That's not a sexy strategy. It doesn't make for a good announcement. But it's the difference between a business that accelerates with agents and one that spends Q3 untangling what they did in Q1. The question isn't whether agentic AI is coming for your industry. It is. The question is whether your business has a clear answer to "who is accountable when the agent gets it wrong" before the first one does. --- *If you're working through where agents fit in your business and want a clear-eyed read on the opportunity and the risk, [book a strategy call](https://calendly.com/adosik21/30min). No pitch, no pressure. If you're ready to start building, see my [AI Workflow Automation service](/ai-workflow-automation/) for how I design, implement, and govern agent workflows end-to-end. Or, if you want to understand how agents are already changing what buyers see about your brand before they reach your site, read about [agents and brand visibility](/blog/ai-agents-killing-the-click/).* --- # AI Agents Are Killing the Click, But Not the Brand URL: https://arthurdosik.com/blog/ai-agents-killing-the-click/ Published: 2026-04-29 Updated: 2026-06-25 Author: Arthur Dosik Tags: GEO, AI Search, Measurement A buyer opens [Claude in Chrome](https://www.anthropic.com/news/claude-for-chrome) and types: *"Research the top three vendors for [thing], tell me which one to pick, and book a demo with the best one."* Twenty seconds later, the agent has read four landing pages, three pricing tables, and a Reddit thread. It picks one vendor. It books the demo. Zero clicks landed on any of the three sites' analytics. Two of those vendors' dashboards look identical to last week. The third just won the deal. The first two have no idea they lost. This is the part of the AI-search story that nobody wants to talk about, because it breaks the dashboard. ## TL;DR - **The core shift:** AI agents are killing the click but not the brand, decoupling visibility from traffic so your brand can disappear from AI answers while your dashboard still looks healthy. - **Why CTR misleads:** Click-through rate stays flat or even rises because agents read and recommend your page without ever loading it in a human session, so the gap shows up late at the bottom of the funnel. - **The new currency:** Mention-share, the percentage of relevant agent and answer-engine outputs that name your brand, is the leading indicator of your position in a model's working knowledge of your category. - **What to track instead:** I'd put citation share, branded mention frequency, and structured-data coverage on a dashboard and watch them weekly to catch brand decay months before it hits pipeline. - **What doesn't change:** The work to win in agent-mediated search is the same work that wins in classic search: real expertise, extractable claims, consistent entities, and clean structured data. Only the scoreboard changes. --- ## What changed: the agent layer For most of the last decade, search optimization had two surfaces to reason about: the SERP and, more recently, the [answer engine](https://blog.google/products/search/generative-ai-search/). Two surfaces, two playbooks, two sets of metrics. There's now a third. Browser-native agents like [Claude in Chrome](https://www.anthropic.com/news/claude-for-chrome), ChatGPT agent mode, [OpenAI's Operator](https://openai.com/index/introducing-operator/), and a half-dozen Operator-style products from other vendors act on behalf of the user. They read pages, compare options, make recommendations, and increasingly take the action: book the demo, fill the cart, schedule the call. They don't browse the way a human browses. They don't click the way a human clicks. And they don't show up in your analytics the way a human shows up. And it's not just browsers anymore. Microsoft just [rolled Agent Mode into Word, Excel, and PowerPoint](https://www.microsoft.com/en-us/microsoft-365/blog/2026/04/22/copilots-agentic-capabilities-in-word-excel-and-powerpoint-are-generally-available/), which means agent-mediated research and decision-making now happens inside the tools people use to *write the brief* and *build the model*, not just inside the tab where they search. The agent layer is becoming ambient. Three surfaces, three sets of incentives: 1. **Classic search** rewards relevance and link authority. The metric is rank. 2. **Answer engines** reward citation-readiness and topical authority. The metric is share of citations. 3. **Agents acting on behalf** reward presence in the model's working knowledge of your category. The metric is mention. The third surface is the one most teams aren't measuring at all. And it's the one growing fastest. --- ## Why CTR is becoming a lagging indicator Click-through rate worked for fifteen years because it was a tight proxy for visibility. If your page got impressions and clicks, you were visible. If clicks dropped, you were less visible. Cause and effect, same loop. The agent layer breaks the loop in a specific way: the agent sees your page, reads your page, reasons about your page, and then doesn't load your page in a session attached to a human. There is no impression a human sees. There is no click for a human to make. There is also no opportunity for the agent to misread you, because you weren't part of the human's session at all. So CTR doesn't go down. It stays exactly where it was. Sometimes it goes *up*, because the people still reaching your site through classic search are now a more concentrated, more buying-intent slice. Marketing dashboards look healthy. Pipeline doesn't. The signal arrives late. You don't see the gap at the top of the funnel where it actually opened. You see it at the bottom, where deals stop closing for reasons sales can't quite name. By then it's a quarter old. This is not "CTR is wrong." CTR is fine. CTR is measuring a real thing: the human traffic that arrives via traditional search. What CTR isn't measuring is the parallel surface where buyers are now spending their research time. The metric isn't broken. It's just measuring a shrinking surface as if it were the whole map. --- ## The new currency: mention-share If the agent never clicks, what does the agent give you? It gives you a mention. When an agent recommends a vendor, names a tool in an answer, or includes a brand in a comparison, that's the unit of value. Whether the user ever clicks through or not, your brand was the one in the consideration set. The next time the agent or the buyer thinks about the category, you're more likely to show up again. Your brand is in the reasoning loop. The data that just landed this week makes this concrete. BrightEdge's [comparison of citation patterns across five AI search surfaces](https://www.searchenginejournal.com/comparison-of-ai-citation-patterns-offers-strategic-seo-insights/573327/) found that the engines diverge widely on which URLs they cite, but they converge on which brands they name. The brand is the unit that travels across engines. The URL is incidental. I'll call this **mention-share**: the percentage of relevant agent and answer-engine outputs in which your brand is named. (Some PR-measurement folks are using *Answer Share* for a similar idea. Same shape, slightly different framing. Pick whichever your team will actually report on.) Mention-share is not the same as a few related metrics it gets confused with: - **Share of voice** measures earned and paid presence across owned and rented channels. It's an output metric for marketing investment, not an input signal for AI retrieval. - **Branded search volume** measures intent that's already formed. By the time someone is searching your brand by name, you've already won the mention. - **Citation share** is a useful subset: percentage of answers that link to your URL. But agents recommend names without links constantly. A brand named in 80% of relevant answers and cited in 20% is in a much stronger position than one cited in 40% and named only in those 40%. Mention-share is the leading indicator. It tells you how well your brand has propagated into the model's working knowledge of your category, which is where the agent looks first the next time it gets asked a related question. --- ## Three signals to track instead of CTR Mention-share is a north star. To act on it, you need to break it into things you can measure on a Monday morning. Three I'd put on a dashboard tomorrow. **Citation share.** Pick 50 high-intent prompts a buyer in your category would actually run. Run them across ChatGPT, Claude, Perplexity, and [Google AI Overviews](https://blog.google/products/search/generative-ai-google-search-may-2024/). Count how many cite your URL as a source. Divide by total. Track weekly. Good for a category-defining brand: 40% or higher on the queries that matter most. Acceptable for a challenger: any positive number with a clear month-over-month trend. **Branded mention frequency.** Same prompt panel, different parser. Now you're not asking *did you cite my URL*, you're asking *did my brand name appear in the answer at all*. This catches the cases where you're recommended without being linked, which is most of them. Good: brand named in the majority of relevant answers. Anything less and you're losing recommendations to competitors who got their entity model right. **Structured-data coverage.** Walk every priority page on your site (homepage, top product or service pages, top blog posts) and audit [schema](https://schema.org/) coverage. Article, FAQ, HowTo, Organization, Product. The retrieval pipelines that feed agent reasoning lean heavily on structured data because it's deterministic and unambiguous. If your schema is half-implemented, you're invisible to retrieval even when your prose is great. Good: 100% on money pages, with no validation errors. These three are the early-warning system. Watch them weekly and you'll see brand decay (or growth) months before it shows up in pipeline. --- ## What to do next week A practical version of the audit, in five steps any serious GEO program can execute in a working week: 1. Write down the 50 prompts a buyer in your category actually runs. Real prompts in their words, not keyword-tool fantasies. 2. Run them across the four major engines: ChatGPT, Claude, Perplexity, Google AI Overviews. Save outputs. 3. For each output, score: did your brand appear (1/0)? Was it cited with a URL (1/0)? Was it the recommendation (1/0)? 4. Compute citation share and mention frequency by engine, and overall. 5. Compare to your organic search share for the same topic cluster, using Google Search Console, your rank tracker, or whatever you trust. The gap between organic search share and mention-share is your exposure. If you're winning on Google and losing in answers, you're paying for visibility in the smaller surface and missing it in the bigger one. Most teams running this audit for the first time find the gap is bigger than they expected. Some find they have no presence at all in answer engines despite ranking well in Google. A small number find they're cited everywhere but ranking nowhere. That's a different problem, but it's the one I'd rather have. --- ## The work doesn't change. The scoreboard does. Here's the part that should sound familiar if you've read [GEO is Just SEO With a Rebrand](/blog/geo-is-just-seo-with-a-rebrand/) or [20 Years in SEO Taught Me This One Thing](/blog/20-years-seo-algorithm-shifts/): The work to win in agent-mediated search is the same work that wins in classic search and in answer engines. Real expertise. Specific, extractable claims. Consistent entity representation. Clean structured data. Authentic authority signals. Citable formats. The list hasn't changed in fifteen years. What changes is how you score yourself. CTR alone isn't going to tell you whether the agents are recommending you. You need to go look. And you need to go look on a cadence, because the model's working knowledge of your category gets updated all the time. Every new model release, every retrieval index refresh, every Reddit thread that surfaces moves your brand's position in it. The brands that win the next few years aren't the ones with the best click-through rates. They're the ones who figured out fastest that the click-through rate stopped predicting the outcome, and rebuilt the dashboard accordingly. The click is dying. The brand is fine. Just measure it differently. --- *If you want a read on where your brand stands across both classic search and the AI engines, my [SEO & GEO Strategy service](/seo-geo-strategy/) starts with exactly this kind of audit. Or [book a free 30-minute call](https://calendly.com/adosik21/30min). No pitch, no pressure.* --- # 20 Years in SEO Taught Me This One Thing About Every Major Algorithm Shift URL: https://arthurdosik.com/blog/20-years-seo-algorithm-shifts/ Published: 2026-04-28 Updated: 2026-06-25 Author: Arthur Dosik Tags: SEO, GEO, AI Search Every time Google drops a major algorithm update, two things happen like clockwork. First, SEO forums light up. Panic. Rankings tanked. Traffic gone. Theories everywhere. Second, someone writes a post explaining how this update proves SEO is unpredictable, chaotic, and basically impossible to plan around. Both reactions miss what's actually happening. After 20+ years in this industry through Panda, Penguin, Hummingbird, BERT, Helpful Content, and now the AI Overviews era, I've noticed that every major algorithm shift follows the same pattern. The same people get hurt. The same people sail through. The same type of work holds up. The "one thing" I've learned: **Google has always wanted the same outcome. It just keeps getting better at enforcing it.** The game never changed. The machine did. ## TL;DR - **The one thing:** Google has always wanted the same outcome, and it just keeps getting better at enforcing it. - **What every update really is:** Each major algorithm shift is Google closing the gap between what it always wanted to reward and what it could actually detect. - **Who got burned:** The losers were always the people optimizing for the gap between Google's intent and its detection capability, not the people doing genuinely good work. - **Why AI search rhymes:** AI engines like ChatGPT, Perplexity, and AI Overviews are a new machine doing the same job, so SEO visibility does not automatically transfer to AI-powered search. - **What to do:** Audit your content for genuine expertise and extractability, make your author signals real and visible, and structure content so a machine can extract and verify your claims. --- ## What most people get wrong about algorithm updates The dominant story in SEO is that Google is unpredictable. That the rules keep changing. That what worked last year will burn you this year. That story is useful for selling SEO courses and emergency audits. It's not actually true. Google's goal since day one has been to [return the most helpful, trustworthy result for any given query](https://developers.google.com/search/docs/fundamentals/creating-helpful-content). That hasn't changed. What changed is Google's technical capability to detect whether your content actually meets that standard or whether you're just good at looking like you do. Every major update is Google closing the gap between what it always wanted and what it could actually measure. The people who got burned were never the people doing good work. They were the people who had learned to exploit the gap between Google's intent and Google's detection capability. When the gap closed, they lost. When it opened back up somewhere else, they ran to it. And the cycle repeated. --- ## The pattern, update by update ### Panda (2011): The content farm reckoning Before February 23, 2011, you could publish garbage at scale and rank well. Content farms, sites built around pumping out thousands of keyword-targeted, shallow articles, were thriving. Google's algorithm rewarded volume and keyword density more than it could reliably detect quality. [Panda changed that.](https://searchengineland.com/google-panda-update-guide-381104) It affected 11.8% of all search queries in its initial rollout. A massive impact. Google's stated goal was explicit: demote "sites which are low-value add for users" and reward "sites with original content and information such as research, in-depth reports, thoughtful analysis." Who got wiped out? The content farms. Sites that had been gaming keyword density and publishing thin, duplicated pages at scale. [Some sites lost more than 80% of their visibility](https://www.conductor.com/academy/glossary/google-panda-update/) almost overnight. Who survived? Sites that had been doing what Google always wanted: publishing substantive, original, specific content that actually answered questions. The update didn't change what Google valued. It just got better at detecting who was delivering it. ### Penguin (2012): The link scheme reckoning After Panda, people who'd been gaming content moved to gaming links. Buying links, building private blog networks, trading placements. All the stuff Google had always said it didn't want, but hadn't been able to catch reliably at scale. [Penguin launched on April 24, 2012](https://searchengineland.com/google-launches-update-targeting-webspam-in-search-results-119295), targeting sites manipulating rankings through artificial link schemes. It initially affected about 3.1% of English queries, smaller than Panda's footprint, but devastating for anyone whose rankings depended on manufactured authority. Same pattern. The people who got hit were the ones exploiting the gap between "Google wants editorial links that signal real authority" and "Google can't always tell the difference between real and manufactured." When Penguin closed that gap, [the sites built on that gap collapsed](https://developers.google.com/search/blog/2016/09/penguin-is-now-part-of-our-core). The legitimate link builders, the ones earning links through quality content, real relationships, and genuine expertise, barely flinched. ### Helpful Content Update (2022 to 2023): The SEO-first content reckoning By 2022, a new gap had opened. This time it was content that technically read as "high quality" (well-structured, correctly keyword-targeted, passable depth) but was written for the algorithm rather than for the person asking the question. [Google rolled out the Helpful Content Update starting in August 2022](https://searchengineland.com/library/platforms/google/google-algorithm-updates/helpful-content-update), with the heaviest impact coming in September 2023. The September update was different. It hit niche blogs, review sites, affiliate content, and anything built around ranking rather than genuinely answering questions. [Some sites reported traffic drops of 40 to 80%](https://www.rakacreative.com/blog/google-helpful-content-update-content-strategy). A 30% drop in impressions was considered average for affected sites. The pattern held. Who got wiped out? Affiliate sites producing formulaic buying guides. Niche blogs built around monetization first, helpfulness second. Content that was technically optimized but hollow in terms of real expertise and perspective. Who made it through? Sites with demonstrable author expertise, genuine first-hand experience, and content that served the reader rather than the ranking. Google didn't change what it wanted. It got better at detecting whether your content was created for the person or for the algorithm. ![Timeline of major Google algorithm updates (Panda, Penguin, Hummingbird, BERT, Helpful Content, and AI Overviews) showing the consistent pattern across 20 years of SEO](/blog/seo-through-the-years.webp) --- ## The one thing, stated clearly Here it is, as plainly as I can put it: **Every major algorithm update is Google closing the gap between what it always wanted to reward and what it could actually detect.** The winners across 20+ years of algorithm shifts have always been the same type of operation: sites that were genuinely helpful, published real expertise, built authentic authority, and structured their content so a machine could extract and verify it. The losers were always the people optimizing for the gap: the delta between "what Google says it wants" and "what Google's technology can actually measure." When the technology caught up to the intent, the gap disappeared. And so did the rankings built on exploiting it. This is not a controversial observation. [Google has been saying the same thing since at least 2011](https://developers.google.com/search/docs/fundamentals/creating-helpful-content). Most people just don't want to believe it because doing the actual work is harder than finding the shortcut. --- ## Why this matters right now, for AI search Here's the part that should make this feel urgent. AI Overviews, ChatGPT, Perplexity, Gemini. These are a new machine doing the same job Google has always done: answering user questions by evaluating which sources are trustworthy, structured, and citable. A new gap just opened. And now [AI agents are acting autonomously on behalf of buyers](/blog/agentic-ai-business-owners/), researching vendors, comparing options, and booking demos without a human reviewing each step. That adds a third surface to reason about, beyond Google and answer engines. Most businesses don't understand how LLMs decide what to cite. There's no clear ranking report. There's no "position 1 in ChatGPT" to track. So a lot of sites that got good at SEO are assuming their visibility transfers automatically to AI-powered search. It doesn't always. And a lot of the tactics people are using to try to "win" at GEO are the same class of shortcuts that got people burned by Panda and Penguin: optimizing for the machine's current limitations rather than for what the machine is actually trying to accomplish. What LLMs are trying to accomplish is identical to what Google has always been trying to accomplish: find the source that most reliably answers the question. That means the sites that will win in AI search are, again, the ones with real topical authority. Named authors with verifiable credentials. Specific, extractable claims backed by data. Clean site architecture. [Consistent entity representation across every surface.](/content-strategy-ai-search/) Content written to be citable, not just to rank. Sound familiar? It should. It's the same list. The machine changed. The goal didn't. --- ## What to do with this If you've been doing real SEO, the kind built around genuine expertise, structured content, and authentic authority, you're better positioned for AI search than most of your competitors who've been cutting corners. If you haven't, the moment to start is now. Not because the shortcuts stopped working. Some still do, temporarily. But because the gap always closes. It closed with Panda. It closed with Penguin. It closed with Helpful Content. It will close in AI search too. (More on why the underlying discipline hasn't actually changed: [GEO is Just SEO With a Rebrand](/blog/geo-is-just-seo-with-a-rebrand/).) The practical starting point is the same it's always been: [audit your content for genuine expertise and extractability](/seo-geo-strategy/). Check your topical depth. Make sure your author signals are real and visible. Structure your content so a machine (whether it's a Google crawler or an LLM retrieval system) can extract and verify your claims. That's not a GEO strategy. That's a content quality strategy. GEO just gave you a new reason to take it seriously. --- *If you want a read on where your brand stands in both Google and the AI engines, start with my [SEO & GEO Strategy service](/seo-geo-strategy/) or [book a free 30-minute call](https://calendly.com/adosik21/30min). No pitch, no pressure.* --- # What Is Generative Engine Optimization? GEO vs SEO Explained URL: https://arthurdosik.com/blog/geo-is-just-seo-with-a-rebrand/ Published: 2026-04-24 Updated: 2026-06-25 Author: Arthur Dosik Tags: GEO, SEO, AI Search Ask ten marketers what GEO is right now and you will get two answers. Some say it is the new SEO. Others say SEO is dead and GEO replaced it. Both framings miss what is actually happening. [Generative Engine Optimization](https://arxiv.org/abs/2311.09735), or GEO, is the practice of getting your brand cited in ChatGPT, Gemini, Perplexity, Claude, Copilot, and Google AI Overviews. It is not a brand-new discipline. It is SEO adapted for AI-generated answers. And the rebrand turns out to be a good thing, because it is pulling a whole generation of marketers back to fundamentals they had stopped taking seriously. ## TL;DR - **The short answer:** GEO is just SEO with a rebrand, the practice of getting your brand cited in ChatGPT, Gemini, Perplexity, Claude, Copilot, and Google AI Overviews using the same fundamentals that win in search. - **The spine is identical:** Google's ranking system and an LLM's citation logic reward the same things: clear claims, topical authority, clean architecture, entity consistency, citable structure, and trust signals. - **What actually changed:** A handful of tactics shifted, including the passage as the unit of optimization, new measurement like brand-in-answer rate, brand exposure instead of clicks, and a competitive set that now includes Reddit and forums. - **Why the rebrand helps:** Calling it GEO gave SEO a new budget line, forced teams to fix neglected schema and entity models, and separated real practitioners from keyword-tool operators. - **Where to start:** Start with the fundamentals, because an AI search engagement overlaps with traditional SEO by about 80%, and the teams winning at GEO are the ones who finally executed on the SEO playbook. ## What is generative engine optimization? Generative engine optimization (GEO) is the process of improving how a brand, website, or piece of content appears in AI-generated answers from systems like Google AI Overviews, ChatGPT, Gemini, Perplexity, Claude, and Copilot. GEO builds on traditional SEO by making content clearer, more authoritative, more structured, and easier for AI systems to understand, summarize, and cite. In plain English: SEO is about earning visibility in search results. GEO is about earning inclusion, citation, or recommendation inside generated answers. The machine changed. The job did not. ## GEO vs SEO: what is actually different? Most of the work overlaps. The differences are concentrated in surface, format, and measurement. | | Traditional SEO | Generative Engine Optimization | |---|---|---| | **Primary surface** | Google and Bing search result pages | AI Overviews, ChatGPT, Gemini, Perplexity, Claude, Copilot | | **Goal** | Rank a page for a query | Get the brand or page cited inside a generated answer | | **Content format** | Page-level optimization for keywords and intent | Passage-level optimization with answer blocks, definitions, comparisons, and FAQs | | **Measurement** | Rankings, impressions, clicks, conversions | Brand-in-answer rate, citation share across engines, AI Overview presence, assisted conversions | | **Authority signals** | Backlinks, E-E-A-T, topical authority, trust | All of the above, plus entity clarity, sourceable claims, and visibility in retrieval-friendly sources | | **Technical requirements** | Crawlability, indexation, speed, schema, internal linking | Same foundation, plus structured data tied to visible content and clear entity signals | | **User behavior** | User clicks a link to your page | User reads an AI answer that may or may not click through | | **Success metric** | Organic traffic and conversions | Cited inclusion and brand exposure inside the answer, plus the traffic that still flows | The headline reads: SEO is about earning visibility in search results. GEO is about earning inclusion, citation, or recommendation in generated answers. Same craft. Different surface. ## How SEO lost the plot For the last decade, "SEO" got watered down to mean one of three things, depending on who you asked: 1. Keyword stuffing and link schemes, from people who never actually did SEO. 2. A technical checklist somebody ran once in 2019. 3. Writing 2,000-word blog posts targeting a keyword nobody searches. Real SEO, the discipline of making your content legible, trustworthy, and citable to a machine that is trying to answer a user's question, never stopped working. It just got boring to talk about. So the conversation moved on to growth hacks, paid social, "content marketing" divorced from search intent, and a dozen other things that felt newer. Then ChatGPT hit the search market. Then [AI Overviews started eating click-throughs](/blog/ai-agents-killing-the-click/). Suddenly there was a whole new layer of discovery that ran on structured content, clean architecture, and authority signals. Which is to say, it ran on SEO. GEO is what we are calling it because "SEO" had lost its seat at the strategy table. The rebrand got it back. ## The spine is identical When I say GEO is SEO, here is what I mean specifically. Both disciplines ask the same question: when a machine is trying to answer a user's question, does your content get selected? Google's ranking system and an LLM's citation logic both reward: - Clear, specific claims over vague marketing language. - [Topical authority](https://services.google.com/fh/files/misc/hsw-sqrg.pdf) on the subject. Not just a single page, but a cluster of related content that signals you actually know the thing. - Clean site architecture that helps crawlers, and now retrieval systems, understand how information is organized. - Entity consistency, meaning your brand, your product names, and your core concepts used the same way across every surface. - Citable structure: headings, lists, tables, schema, and direct answers that can be extracted. - Trust signals: author credentials, named sources, real data, and third-party validation. Read that list again. Every item on it has been SEO best practice since roughly 2015. [Schema](https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data)? Same. [E-E-A-T](https://developers.google.com/search/docs/fundamentals/creating-helpful-content)? Same. Answer-the-question content? Same. Entity-based optimization? Same. What changed is which machine is doing the reading. That changes a handful of tactics, not the spine. ![Venn diagram showing SEO and GEO sharing the same fundamentals: clear claims, topical authority, clean site architecture, entity consistency, citable structure, and trust signals, with the intersection labeled "Same fundamentals. Different machine."](/blog/blog-image2.webp) ## What did not change These are the SEO fundamentals that AI search did not retire. If anything, AI search made them count more. - Crawlability still matters. - Indexation still matters. - Content quality still matters. - Authority still matters. - Internal linking still matters. - Technical SEO still matters. - Clear topical focus still matters. - Backlinks still matter. - Author and entity signals still matter. If your fundamentals are weak, no amount of "GEO strategy" saves you. The retrieval pipelines that feed AI answers pull from the same open web. They reward the same things. ## What did change A handful of things genuinely shifted, and ignoring them is how teams get burned. **The unit of optimization.** In classic SEO, the page was the unit. You optimized a page to rank for a query. In GEO, the passage is the unit. An LLM does not cite your whole page. It [extracts a specific 2 to 3 sentence block](https://arxiv.org/abs/2005.11401) that answers the user's question. That means your page architecture has to front-load answers and make individual sections self-contained and extractable. **The measurement.** You cannot rank-track your way to GEO visibility. Ranking #1 in Google does not mean you get cited by ChatGPT, and vice versa. You now need to measure brand-in-answer rate, citation share across engines, and AI Overview presence alongside classic rankings. The tools are immature, and most teams are still flying blind. **The distribution.** A cited answer in ChatGPT does not send you a click the way a Google result does. You get brand exposure instead of traffic. That is a real trade-off, and it changes how you value the work. It does not change how you do the work. **The competitive set.** In a lot of categories, the pages getting cited by LLMs are not the pages ranking in Google. Reddit, niche forums, and heavily-structured reference sites are [over-indexed in LLM training](https://www.cbsnews.com/news/google-reddit-60-million-deal-ai-training/) and retrieval. If you are a B2B brand, you may suddenly be competing with a Reddit thread from 2022. That is a strategy problem worth thinking about carefully. **Brand mentions matter more than URLs.** AI engines diverge widely on which URLs they cite, but they converge on which brands they name. The brand is the unit that travels across engines. The URL is incidental. **Content has to answer the next question, not just the first query.** AI tools follow up. Your content needs to anticipate the comparison, the alternative, the objection, and the "is it worth it" question that comes after the initial answer. ![Venn diagram of SEO (search engines: Google, Bing) and GEO (generative engines: ChatGPT, Perplexity, Gemini, AI Overviews) overlapping on the same fundamentals, captioned "Same fundamentals. Stronger results."](/blog/blog-image1.webp) These are real changes. But notice what they do not change: the underlying craft. ## Is GEO the same as answer engine optimization? GEO and AEO overlap, but the terms are not used identically across the industry. - **Answer engine optimization (AEO)** usually focuses on earning concise answers in search features, voice assistants, and answer engines. Featured snippets, voice queries, and Perplexity-style direct answers are the core surface. - **Generative engine optimization (GEO)** is the broader umbrella. It covers how brands and content appear in AI-generated summaries, recommendations, comparisons, and citations across the major LLM-powered surfaces. - **AI search optimization** and **LLM SEO** are working synonyms many practitioners use interchangeably with GEO. You will see all four terms in the same job descriptions and agency pitches. - **AI Overview optimization** and **Google AI Overview SEO** are the narrow, Google-specific subset, focused on getting cited inside Google's generated SERP overviews. The acronym fight is not the work. The work is making your brand clear, credible, and quotable across every surface a buyer might use to ask about your category. ## How generative engine optimization works Strip away the marketing and GEO is a six-step practical system. None of it is exotic. All of it is hard if you have not been doing it. 1. **Define the entity.** Who you are, what you do, who you help, what problems you solve. One consistent story across the site, schema, and the open web. 2. **Build topical authority.** Service pages, explainers, FAQs, comparisons, alternatives, and case studies that cover the subject from every angle a buyer cares about. 3. **Structure content clearly.** Definitions, summaries, tables, lists, and direct-answer paragraphs. Schema that matches the visible content. Self-contained sections an LLM can extract. 4. **Strengthen authority.** Credible authorship with bios, real expertise, named sources, third-party mentions, and the backlinks that reflect the work being cited elsewhere. 5. **Improve technical access.** Crawlability, indexation, speed, structured data, and internal linking that helps both Googlebot and retrieval systems understand the site. 6. **Measure visibility.** Track rankings and clicks, but also brand-in-answer rate, citation share across engines, AI Overview presence, and assisted conversions. The metric stack is bigger now. Done together, these earn rankings in Google and citations in AI engines from the same content investment. There is no trade-off. ## How to do GEO without chasing gimmicks There is a small industry forming around "GEO hacks". Most of them are noise. The boring checklist below outperforms the hacks every time. - Start with SEO fundamentals. If they are broken, fix those first. - Rewrite vague service pages into clear, answer-ready pages. - Add FAQs and direct answer blocks under your H2s. - Build comparison content, "what is" explainers, and alternative pages for your category. - Make author expertise visible. Real bios, real credentials, real opinions. - Use schema where it matches visible content. Article, FAQ, HowTo, Organization, Product, Service. - Link related pages into topical clusters so the entity model is obvious. - Track AI mentions manually until the tools mature. Run 30 to 50 high-intent prompts across ChatGPT, Gemini, Perplexity, Claude, and AI Overviews. Note where you show up and where you do not. If that list looks suspiciously like an SEO checklist with a few additions, that is the point. ## Where your current SEO program probably falls short If you are an executive trying to pressure-test your team's readiness for AI search, here is where most programs are weakest right now: - **Too much content is written for keywords, not questions.** AI search is question-shaped. Pages that target a noun phrase without ever answering a real question get ignored. - **Service pages are vague.** They describe a category instead of explaining what you actually do, who you do it for, and why a buyer should trust you. - **Brands fail to explain themselves.** The "About" page is fluffy. The category positioning is implied, not stated. AI systems cannot summarize what is not on the page. - **Thought leadership lacks answer-ready structure.** Smart essays with no H2s, no direct answers, and no extractable passages are invisible to retrieval, no matter how good the writing is. - **Teams measure only traffic, not visibility.** If your dashboard does not show citation share or AI mentions, you are flying blind on half the discovery layer. Fix these and your SEO program improves. Your AI search visibility improves. Same work. Two outcomes. ## Why the rebrand is actually helpful If the fundamentals are the same, why does the rebrand matter at all? A few reasons. They are the reasons a lot of SEO veterans are quietly grateful for it. It gave SEO a new budget line. "We need to fund SEO" has been a hard sell in a lot of orgs. "We need to be visible in ChatGPT" is an easy one. The work underneath is 80% the same, but the narrative finally matches the importance. It forced teams to fix things they had been ignoring. Schema that was half-implemented. Entity models that did not exist. Author attribution missing from thousands of pages. Internal linking that had gotten sloppy. All of this tends to get fixed when someone senior starts asking why the company is not cited in AI answers. It separated the real practitioners from the posturers. A lot of people who called themselves SEO experts were really just keyword-tool operators, and they are out of their depth in GEO. The people who understood the fundamentals (how search engines actually evaluate content, how entities and authority work, how to write for machines and humans at the same time) have an edge again. If you have been [watching SEO evolve through twenty years of algorithm shifts](/blog/20-years-seo-algorithm-shifts/), this one rhymes with the others. New surface. Same craft. ## Where to start If you are trying to figure out what to do about any of this, the honest answer is: start with the fundamentals, because the fundamentals are the work. An [AI Search Visibility and SEO Strategy engagement](/seo-geo-strategy/) overlaps with traditional SEO by about 80%. The 20% that is different is the part that needs new tooling and new thinking: measuring citation share, auditing for passage-level extractability, checking how your entity is represented in LLM answers, evaluating AI Overview presence. The other 80% is the work that should have been getting done all along. If your schema is broken, fix it. If your author bios are thin, flesh them out. If your content is full of vague marketing language, rewrite it in specific, verifiable claims. If your topical clusters have holes, fill them with a stronger [content strategy built for AI search](/content-strategy-ai-search/). All of it helps you rank in Google and get cited by ChatGPT. There is no trade-off. The teams doing well in GEO right now are not the ones who threw out their SEO playbook. They are the ones who finally executed on it. ## The takeaway GEO is SEO with a rebrand. The rebrand is useful because it is pulling attention and budget back to a discipline that got unfairly dismissed. The tactics that win in AI search are the same tactics that have always won in search: clarity, specificity, authority, structure, trust. If you have been doing SEO right, you are already doing most of GEO. If you have not, the good news is you now have an excuse to start. The bad news is your competitors are using the same excuse. ## Frequently asked questions ### What is generative engine optimization? Generative engine optimization (GEO) is the practice of improving how a brand, website, or piece of content appears in AI-generated answers from systems like Google AI Overviews, ChatGPT, Gemini, Perplexity, Claude, and Copilot. It builds on traditional SEO by making content clearer, more structured, more authoritative, and easier for AI systems to understand, summarize, and cite. ### Is GEO replacing SEO? No. GEO is an extension of SEO, not a replacement. The retrieval pipelines that feed AI answers pull from the same open web, so traditional SEO fundamentals like crawlability, content quality, internal linking, authority, and structured data still drive AI search visibility. GEO adds a layer on top: passage-level extractability, entity clarity, and answer-engine measurement. ### What is the difference between GEO and SEO? SEO focuses on ranking pages in traditional search results like Google and Bing. GEO focuses on getting brands and content cited inside AI-generated answers from Google AI Overviews, ChatGPT, Gemini, Perplexity, Claude, and Copilot. SEO measures rankings and clicks. GEO measures brand-in-answer rate, citation share, and AI Overview presence alongside the traditional metrics. ### Is GEO the same as answer engine optimization? GEO and answer engine optimization (AEO) overlap heavily but are not identical. AEO usually focuses on concise answers in search features, voice assistants, and answer engines like Perplexity. GEO is broader, covering how brands and content appear in AI-generated summaries, recommendations, comparisons, and citations across all major LLM-powered surfaces. Many practitioners use GEO, AEO, AI search optimization, and LLM SEO interchangeably. ### How do you optimize content for AI search engines? Optimize content for AI search by combining traditional SEO with passage-level structure. Front-load direct answers under each H2, write self-contained sections an LLM can extract, define your entity clearly, build topic clusters that demonstrate authority, use schema that matches visible content, and include comparisons, FAQs, and definitions. Strong technical SEO and strong topical authority remain non-negotiable. ### Can GEO help with Google AI Overviews? Yes. Google AI Overviews are generated from the same web content that Google ranks, with extra weight on clearly structured passages, schema-backed pages, and content that directly answers the underlying question. GEO practices, especially direct answer blocks, FAQs, comparison content, and clean schema, materially improve the odds of being cited inside AI Overviews. ### How do you measure generative engine optimization? Measure GEO with a metric stack, not a single number. Track brand-in-answer rate by running a fixed set of high-intent prompts across ChatGPT, Gemini, Perplexity, Claude, and Google AI Overviews. Add citation share across engines, AI Overview presence on priority queries, structured-data coverage on money pages, and assisted conversions from referrer-light AI traffic. Keep traditional rankings and clicks in the dashboard. Both still matter. ### Do small businesses need GEO? Yes, especially in categories where buyers are already using AI to research vendors. The good news is the work is mostly traditional SEO done well: clear service pages, real expertise on the page, FAQs that answer real buyer questions, structured data, and internal linking. A small business that nails the fundamentals often outperforms larger competitors who treated SEO as an afterthought. --- *If you want a read on where your brand stands in both Google and the AI engines, see my [AI Search Visibility and SEO Strategy service](/seo-geo-strategy/) or [book a free 30-minute call](https://calendly.com/adosik21/30min). No pitch, no pressure.*