--- title: "I Received an Email Telling Me I Don\'t Exist. I Stopped to Think" date: 2026-05-22 description: "A promotional email I received made me wonder what it truly means to exist in today\'s digital world — and how much of the fear others sell us is actually real." author: "Petru Cojocaru" categories: - name: "Online promotion" url: "https://cojocarupetru.info/online-promotion.md" --- # I Received an Email Telling Me I Don't Exist. I Stopped to Think It was a short message. The kind written by someone who knows they have one chance — a single read before deletion — and so compresses everything into a tight, salutation-free opening, with a small, precise blow in the first line. "Most Romanian brands DO NOT appear at all in AI responses." The DO NOT in capitals. Deliberate. [![I Received an Email Telling Me I Don't Exist. I Stopped to Think](https://cojocarupetru.info/images/marketing/original-asumat-sau-continut-generic.jpg "I Received an Email Telling Me I Don't Exist. I Stopped to Think")](https://cojocarupetru.info/index.php?option=com_content&view=article&id=358:don-t-exist&catid=27:online-promotion&lang=en-GB&Itemid=504) I read it twice. Not because it had convinced me, but because it had stopped me — and that is a rare thing, amid the daily stream of messages I scroll through with the half-awake eyes of a man who has seen enough pitches in eight years of building websites for public institutions. Something in the phrasing had caught on something. Perhaps it was "do not exist" — not "appears less often", not "has reduced visibility", but simply: *does not exist*. As though absence from a chatbot's generated response is equivalent to metaphysical non-being. I stood up from my desk. I reached for the cup of coffee that had already gone cold, because I'd stood up half an hour earlier to answer something my son had asked and had forgotten all about it — which is perfectly normal in this house, where cold coffee has somehow become a symbol of how I live, with my attention perpetually fragmented between work and family and thoughts that can never quite find their peace. I drank it cold. And I began to think. --- **What Fear Looks Like When It's Professionally Packaged** There is a kind of fear you don't feel suddenly, like an alarm — you absorb it slowly, through repetition. The message from Marius — or from whoever signed it as Marius, because I have no way of knowing if that is a person or a campaign alias — operates on precisely this logic. It doesn't tell you "you are in danger now". It tells you "you may already be falling behind, and you don't even know it." It is a more insidious fear, because it has no precise moment of arrival. It enters through the door of uncertainty that anyone building something — a business, a blog, an online presence — carries with them wherever they go. I know this uncertainty from the inside. I have built websites for town halls and hospitals and schools for eight years, and throughout all of that time I have lived with the question of whether I am doing things correctly, whether I am visible enough, whether I am being overtaken by changes I hadn't anticipated. SEO has evolved several times under my hands. Mobile-first arrived, then core web vitals, then AI overviews — and each time, someone was there to transform the change into a sales emergency before the market had properly understood what was actually happening. Marius's message does exactly the same thing. It takes a real phenomenon — and the phenomenon is real, I shall get to that — and packages it in the language of existential urgency. "Does not exist." "Someone else appears in your place." "Reply with one word and I'll tell you directly." That last part gave me the longest pause. Reply with a single word and I'll give you a diagnosis. As though eight years of work, four active websites, and dozens of clients in the public sector can be assessed on the basis of a word sent by email. As though my digital visibility — everything I have built slowly, through mistakes and corrections, through nights spent in configuration files and articles written at hours when my son has long since gone to sleep — can be summarised in a second of analysis by someone I've never met. I deleted the email. Then I opened a new document. --- ### What Is Real Behind the Alarm I need to be honest here, because I dislike building arguments on half-truths. The phenomenon Marius invokes does exist. Generative AI systems — ChatGPT, Gemini, Perplexity, and the others still to come — are changing the way people search for information and, through that, the way brands are or aren't discovered. A study carried out by the agency Marketez in the autumn of 2025 showed that Romanian brands have significantly lower visibility in ChatGPT than on Google, even in sectors with otherwise solid digital performance. In tourism, well-known agencies go unmentioned — ChatGPT cites Tripadvisor and Lonely Planet instead. In fashion, international brands dominate entirely. The only Romanian entities that appear consistently are those with a massive editorial presence: large clinics, national pharmacy chains, educational platforms with vast quantities of content indexed over many years. The mechanism is straightforward once you understand it, and for that reason it should prompt understanding rather than fear. A language system — I prefer this formulation to "artificial intelligence", which personifies what is, in essence, a statistical tool in a rather misleading way — does not search the internet in real time as a classical search engine does. It generates responses on the basis of statistical patterns extracted from the billions of public texts it "read" during training. If your name or your company's name appears frequently in authoritative sources — specialist press, Wikipedia, professional forums, reviews — then you exist within the model's statistical representation. If you appear only on your own website, you are, from the system's perspective, background noise too faint to reproduce. This is not a condemnation. It is a description. And like any technical description, it is more useful than any sales pitch built upon it. --- ### Father, Person, Website Builder — and All the Questions That Come With It I think sometimes — particularly in the evenings, when my son is in his room and I'm left with the laptop and thoughts that didn't finish when the working day did — about what it means to exist in this parallel, digital world that has grown so large it sometimes seems more real than the street outside. My father was a mechanic. He repaired engines. His existence was concrete, verifiable by hand: the car ran or it didn't. If it ran, his work had had meaning. There was no algorithm to tell him he wasn't visible enough, that someone else would appear in his place when someone asked "who fixes engines in this town". His existence was certified by the sound of an engine turning over. I build websites. Things that exist on servers I have never seen, in physical locations I don't know, accessible via protocols I understand technically but cannot touch. And now someone arrives to tell me that perhaps I don't exist there either — that in the eyes of a prediction system built on statistical patterns, eight years of effort are invisible. I know this is an exaggerated reading of Marius's message. I know he was speaking about visibility in ChatGPT, not about the value of my work as a human being. But there is a temptation — one I recognise in myself, and which I must name in order to defuse it — to allow digital metrics to become metrics of personal worth. It is a temptation our culture feeds constantly, through likes, through engagement, through SEO rankings, through how often you are cited. And now through how often you are mentioned by a chatbot. I don't want to live by that metric. And I don't want my son, who sees me working at hours when most people are already asleep, to believe that the value of what I build is measured in the responses of an AI system. --- ### What Hides Behind the Absent Methodology I returned to the email after a few hours. Not because I had changed my mind, but because I needed to read it again with the eyes of someone who had already passed through his first emotional reaction. "I have tested Romanian brands." The more I read this sentence, the emptier it seems. Which brands? Which sector? What questions were put to the AI system? Which version — with internet access or without, in Romanian or in English — because the results differ significantly depending on the language of the query? What analytical tools were used? The serious people working in AI visibility — and they do exist, building monitoring platforms and publishing studies with transparent methodologies — speak of tools such as Profound, GEOflux or Conductor, of tens or hundreds of systematic queries, of comparisons across platforms and versions. They speak of months of work before results become visible, not of diagnoses delivered in response to a single word sent by email. The absence of methodology from Marius's message is not a casual omission. It is structural. If he explained the methodology, he would become evaluable — and therefore open to challenge. Instead, he remains in the realm of vague promises, where the only concrete step required is mine: to reply, to open the door, to enter the sales funnel. This is not a fraud in the classical sense. It is a sales technique. But it is a technique that feeds on ignorance, and for that reason it troubles me more than an outright fraud would. An outright fraud requires you to be naive. This one merely requires that you haven't studied a sufficiently technical subject in sufficient depth — which is a perfectly normal situation for any busy person. --- ## What Can Be Done, Concretely and Without Panic And yet, if the phenomenon is real — if visibility within AI systems matters and will matter more and more — the practical question remains: what do you actually do? I like to think through things in small, concrete steps that I can take alone, or with limited resources. It is the way I have always worked — not with large budgets and marketing agencies, but with attention, methodically, without skipping stages. The first step is to test your own visibility before paying anyone to do it. Open ChatGPT or Perplexity and ask the questions a potential client of yours might ask. "Which web service providers exist for public institutions in Romania?" "Who builds websites for town halls in Moldova?" Note what appears. It is real information, free, obtained in a matter of minutes. The second step is to understand why visibility in AI works differently from classical SEO. The system doesn't index your pages directly — it reproduces patterns from authoritative sources in which you are mentioned. That means the work is done not on your own website, but on your presence in the wider information ecosystem: specialist press, professional directories, public case studies, genuine reviews on platforms that carry weight. The third step, which is both the hardest and the most important, is to produce content that answers questions rather than describing services. The difference seems small and is in fact enormous. An AI system looks for concrete answers to concrete questions. "What must a town hall website contain under Law 544/2001?" is an answer I can provide. "Professional web services for public institutions" is a brochure the system doesn't know what to do with. I have written a great deal over the past few years — articles about web accessibility, about GDPR as applied to public institutions, about cybersecurity for Joomla, about everything pertaining to my daily work. Not because anyone told me it would help with AI visibility — that phrase didn't even exist when I wrote the earliest pieces. I wrote because it seemed to me that I knew useful things others didn't, and that it was worth putting them somewhere they could be found. That there was a kind of responsibility to one's field in doing so. It may well be that precisely this logic — producing something useful rather than optimising for a system — turns out to be the best long-term strategy. Not because it is romantic, or because I am nostalgic for the web before SEO. But because AI systems, at their core, reproduce what people have found useful and worth citing. If you build useful things and make them public, you grow organically in their statistical representation. --- ### What I Would Tell My Son, If He Asked I think sometimes about the conversations I haven't yet had with my son, but probably will — about work, about building things, about what it means to make something that matters in this loud and hurried world. If he asked me about Marius's email — if I showed him that compact, anxiety-laden message and said "look, someone is telling me I don't exist" — I think he would laugh. He is at the age when he holds fast to things that are concrete and verifiable, at which the absurdity of a system that certifies your existence through mentions would seem to him obviously ridiculous. And he would be right. My digital existence is not certified by how many times ChatGPT cites me. It is certified by the websites that work, by the town halls that have a better online presence than they did five years ago, by the people who rang me to say that the article about web accessibility had helped them pass an audit. This doesn't mean I'm ignoring the AEO phenomenon. It means I'm integrating it into my work without allowing it to become a source of anxiety. It means I will continue writing useful articles, building clear and semantically structured websites, being present in places where my presence has meaning — not because a Marius told me I wouldn't otherwise exist, but because it is simply the way I have always worked. --- ### The Anatomy of a Pitch That Sells Fear I want to be explicit about the structure of this kind of message, because I see it circulating in various forms and believe it merits public deconstruction. Step one: a dramatic statement presented as verified fact. "Most Romanian brands DO NOT appear." No data, no methodology, no context. Purpose: to activate insecurity. Step two: the personalisation of fear. "That means when someone asks, someone else appears in your place." You don't know who. You don't know in what context. You don't know how often. But the image is planted: an imaginary competitor displacing you in the mind of an imaginary client. Step three: the offer of a free diagnosis with a minimal barrier. "Reply with one word and I'll tell you directly." A single word. What have you got to lose? But that word is all the sender needs: a signal that you are present, that you read it, that the subject touched you enough to prompt a response. From there, you are in the funnel. There is no analysis prior to your reply. The diagnosis being promised is prepared to be delivered regardless of who you are and what your business does — a script adapted post-reply, not a genuine assessment. I recognise this structure because I have seen it in other forms across the years. In 2015 it arrived as "your website isn't optimised for mobile" — sent when mobile-first indexing was new and frightening. It arrived as "you're not on the first page of Google for your important keywords" — without anyone knowing which keywords mattered for me. Now it arrives as AI visibility. The form changes. The mechanism stays the same. --- ### What It Actually Means to Exist I return to the question that stopped me on my first reading of the email, because it refuses to leave entirely. What does it mean to exist in the digital world? The simple answer — the one Marius is selling — is: to be mentioned by AI systems when someone asks a relevant question. The more complicated answer, the one I have been building across eight years of work, is: to leave behind things that function, that help real people, that endure after you've finished the project and moved on. The website of a town hall I built four years ago that still works — with information accessible to citizens, with the correct structure for people with disabilities, with contact details kept up to date — that exists, regardless of whether ChatGPT mentions it. The article I wrote about implementing GDPR for public institutions, which was passed from one town hall to the next by email — that exists. The conversation last night with my son, in which he showed me something he was building and I stopped my work to look at it properly — that exists too, even though it isn't indexed anywhere and no AI system will ever reproduce it. There is a difference between visibility and existence. Marius is selling one of them. The other is built differently: more slowly, with less noise. --- ### What to Do Before Replying to Any Email of This Kind If you have received, or will receive, a similar message — and you will, because AEO has become the new terrain for anxiety-driven pitches — here is what I suggest, from my own experience and from everything I have read and verified about the phenomenon. Test your own visibility in AI systems yourself, using the questions relevant to your field, free of charge, in a few minutes. Don't pay anyone to do this before you have done it yourself. Check whether the sender has a verifiable portfolio, documented results, and a publicly explained methodology. A serious AEO specialist does not begin the conversation with a data-free alarm. Think about your actual audience: does your typical client use ChatGPT in their decision-making process? If so, at what stage? If not, AI visibility is considerably less urgent than the email implies. And if you decide you do want to invest in AI visibility, look for demonstrated competence, not induced fear. There are serious people working in this field — it's just that they don't generally send cold emails with capital letters demanding a one-word reply. --- ## A Final Note I finished writing this piece after midnight, with Gri on the desk — my cat, who has a habit of settling directly on the laptop keyboard whenever she decides I've given her too little attention today. She may well be right. Tomorrow I'll send a client an article about GDPR compliance. The day after, I'll check some server logs for a hospital that has been having loading-speed problems. Next week I'll write something for my son, if I can find the right words for the subject I've been turning over in my mind for the past few months. None of these things will make me visible in ChatGPT unless I accompany them with a broader editorial strategy. And yet all of them exist — with or without a mention from a prediction system built on statistical patterns. That is my real reply to Marius's email. I didn't send him a single word back. --- --- # 20 Questions and Answers **What is AEO and why is it emerging now?** AEO — Answer Engine Optimisation — is the discipline of building an online presence that generative AI systems will cite when answering questions. It is emerging now because increasing numbers of people use ChatGPT, Gemini or Perplexity in place of classical Google searches for informational and exploratory queries. The phenomenon is real; the manufactured urgency with which it is being sold, rather less so. **Why don't Romanian brands appear in ChatGPT?** AI systems reproduce patterns from the authoritative sources they "read" during training. Romanian brands are underrepresented in high-weight international sources — global press, English-language Wikipedia, major review platforms. This is not a question of quality; it is a question of information ecosystem. **Is an email telling me I don't exist in AI a scam?** Not necessarily in any criminal sense. It is a sales pitch that uses anxiety as a mechanism for opening a conversation — one that lacks methodology, lacks verifiable credentials, and makes a diagnostic promise it could not possibly deliver on the basis of a single-word reply. **Should I invest in AI visibility right now?** It depends on your audience. If your clients make decisions after broadly informational searches — "which type of X firm should I choose" — then AI visibility makes sense as a medium-term investment. If decisions come through direct recommendations, tenders, or existing relationships, local SEO and presence on sector-specific platforms remains the priority. **What can I do right now, for free?** Test your own visibility using the questions a potential client would ask in ChatGPT or Perplexity. Audit the consistency of your message across all pages of your website. Write an article that answers a frequently asked question in your field — not a service brochure, but a genuinely useful and verifiable response. **Does AI visibility replace classical SEO?** No, at least not completely and not across all sectors. The two coexist and complement each other. The authority built through SEO — links, quality content, presence in authoritative sources — also contributes, indirectly, to AI visibility. **How do AI systems decide what to mention?** They reproduce patterns from the texts on which they were trained. The more frequently a brand or name appears in high-authority sources — specialist press, encyclopaedias, professional directories, credible review platforms — the more likely it is to be reproduced in generated responses. **Does the language I use on my website matter for AI visibility?** Significantly. Romanian-language content has much lower representation in global AI training data than English-language content. For audiences that search in Romanian, visibility in locally trained or language-specific models matters. For international reach, English content reaching authoritative international sources is essential. **What is the difference between AEO and GEO?** AEO (Answer Engine Optimisation) focuses on being cited in direct answers generated by AI systems. GEO (Generative Engine Optimisation) is a broader term sometimes used to describe the same field, or to indicate optimisation across generative search systems more generally. In practice, the two terms are used interchangeably by most practitioners. **Does having a Wikipedia page help with AI visibility?** Yes, substantially. Wikipedia is one of the most heavily represented sources in AI training data. A well-maintained, properly cited Wikipedia page is one of the most reliable signals of authority that a language model can draw upon. **Is it worth hiring a specialist for AEO?** If you have already tested your own visibility, understand the field's fundamentals, and have identified a real need for improvement, then a specialist with a demonstrable portfolio and transparent methodology can add genuine value. If you are responding to an alarm email with no methodology attached, the answer is: not yet. **What kind of content builds AI visibility?** Content that answers specific, frequently asked questions. Comprehensive guides, practical comparisons, case studies with documented outcomes, structured FAQ sections. The common thread is utility: content that answers a question someone is actually asking, clearly and verifiably. **Does social media presence affect AI visibility?** Indirectly, and to a limited extent. Social media content is not systematically integrated into the training data of most large language models. What matters more is whether social media generates secondary effects — press coverage, citations on authoritative platforms, reviews. **How long does it take to become visible in AI systems?** Months to years, depending on sector, existing authority, and publishing consistency. AI models are retrained periodically, not in real time. Progress is gradual, not sudden. Anyone promising rapid results should be regarded with caution. **Can a small business or a public institution appear in ChatGPT?** Yes, but generally through indirect means rather than direct indexing. Being mentioned in specialist press, in professional directories, in studies published by recognised bodies, or in widely shared citizen reviews — these all contribute to building the statistical weight that makes citation possible. **What should I do if I receive a cold email about AEO?** Before replying, check: does the sender have a verifiable portfolio? Do they explain their methodology? Are there documented case studies? If none of these are present, the most sensible response is to conduct your own visibility test — free, in a few minutes — and decide from there. **Does internal site structure affect AI visibility?** Yes. Clear, semantic HTML structure, properly used headings, structured FAQ markup, and coherent internal linking all make it easier for AI systems to interpret and reproduce information from your pages. Good structure serves both classical SEO and AI visibility simultaneously. **What is the biggest mistake in approaching AEO?** Treating it as an emergency rather than as a gradual strategic adjustment. The brands that will be most visible in AI systems in three years are those that have been publishing consistent, useful, authoritative content since now — not those that purchased a last-minute AEO package out of anxiety. **Is fear a legitimate marketing tool?** Fear is a human response to genuine threat. Using fear as a sales technique — by amplifying uncertainty, removing context, and promising rapid resolution — is legitimate in a narrow commercial sense, but ethically corrosive in a field that depends on trust between specialist and client. It works in the short term; it undermines professional relationships in the long. **What is the most honest answer to the question of digital existence?** To exist digitally means to leave behind things that function, that help real people, and that endure after the project is complete. Visibility in AI systems is one dimension of that existence — an increasingly relevant one, and worth tending to. But it is not the whole of it, and it is certainly not a substitute for the substance that makes visibility meaningful in the first place. --- ## Selected Bibliography Marketez — *"Which Romanian Brands Are Most Mentioned by ChatGPT?"*, November 2025. [marketez.ro](https://www.marketez.ro/care-sunt-cele-mai-mentionate-branduri-din-romania-de-chatgpt) Capital.ro — *"Local Brands Are Almost Invisible in AI Recommendations"*, 2025. [capital.ro](https://www.capital.ro/interesul-pentru-chatgpt-explodeaza-in-romania-brandurile-locale-sunt-totusi-aproape-invizibile-in-recomandarile-ai.html) Wall-Street.ro — *"The Mistakes That Cost You Money Because You Don't Exist for ChatGPT"*, 2025. [wall-street.ro](https://www.wall-street.ro/articol/ecommerce/greselile-care-te-fac-sa-pierzi-bani-pentru-ca-nu-existi-pentru-chatgpt.html) Amsive — *"Answer Engine Optimization: Evolving Your SEO Strategy in the Age of AI"*, June 2025. [amsive.com](https://www.amsive.com/insights/seo/answer-engine-optimization-aeo-evolving-your-seo-strategy-in-the-age-of-ai-search/) Evergreen Media — *"Answer Engine Optimization: AI Visibility in 2026"*, February 2026. [evergreen.media](https://www.evergreen.media/en/guide/answer-engine-optimization/) --- ## Author's Note: Limits, Tensions, and Acknowledged Errors **The essay's structural limitations** This essay claims to dismantle an anxiety-producing rhetorical mechanism, yet does so using a narrative structure that itself demands considerable emotional investment from the reader. The length and confessional density are aesthetically justified; functionally, rather less so. A reader who simply wants to know whether or not to invest in AEO would be better served by a direct comparative analysis of six hundred words. The choice of the long form is not neutral — it filters the audience towards readers of literary essays and distances precisely those most vulnerable to pitches of the kind described here: people who lack the time or the habit to read three thousand words before making a decision. ### Unresolved Internal Tensions There is a contradiction the text handles gracefully without resolving: the author describes in some detail what correct AEO strategy looks like — useful content, presence in the wider editorial ecosystem, semantic structure — and acknowledges that he already practises it. This means that, in substance, he is not so far from what Marius is selling. The difference is one of tone and methodology, not of essence. The text benefits from this ambiguity without naming it directly: it positions itself simultaneously as a critique of commercialised AEO and as a competent practitioner of authentic AEO. It is a rhetorically comfortable position; structurally, it is less than fully honest. A second tension: the author criticises the absence of methodology in Marius's email, yet the Marketez study invoked as the primary source is cited from memory, without verifiable access to its methodology. The reader is offered a counter-argument based on authority rather than on raw data. This is precisely the charge levelled at Marius — applied in reverse. ### The Rhetorical Tools Used and Their Effects **Diminishment through naming** — Marius receives no professional qualification. He is "Marius — or whoever signed it as Marius". This move de-individualises the adversary, turning a person into a type, which makes criticism both easier and less falsifiable. You cannot rebut a critique aimed at a type. **Authority built through humility** — "eight years of work with town halls and hospitals" is not an academic credential, but in context it functions more powerfully than one: it is moral legitimacy, hard to contest, deployed precisely when the author is under attack. This move is effective, but it does not demonstrate specific AEO competence. Building technically sound websites for public institutions does not automatically qualify someone to evaluate AI visibility methodologies. **The affective anchor** — the mechanic father, the son, the cold coffee, the cat on the keyboard. These elements are not ornamental; they construct a narrative self carrying moral weight, against which any counter-argument risks appearing cynical. It is a technique Knausgård refined to an art form, but in an argumentative context it creates an emotional shield around intellectual positions as well as around the person. A reader who wished to challenge a claim about AEO must first pass through the image of a man drinking cold coffee and thinking of his dead father. **Reversal of the burden of proof** — "A serious AEO specialist does not begin a conversation with a data-free alarm." This is true, but it is a criterion that can be deployed to dismiss any commercial approach, including legitimate ones. The text offers no examples of serious specialists with a different approach; it leaves the space empty, which implicitly validates the rhetoric: if serious practitioners don't send cold emails, then the only valid contact is inbound — which is precisely the kind of authority the author is constructing through this very article. ### Acknowledged Subjective Errors **Generalisation from a single case.** One Marius's email becomes representative of an entire category of AEO pitches. It is possible that Marius has a genuine methodology, a serious portfolio, and a sales approach that sounds poor but delivers correctly. The author doesn't know this — he deleted the email before replying. The critique is built on a recognised pattern, not on evidence specific to this case. **Conflating existence and visibility** — the essay's most memorable moment is also its most intellectually exposed. The distinction "there is a difference between visibility and existence" is philosophically true, but contextually evasive: a town hall website that functions but is undiscoverable through a relevant channel does not reach the citizen who searches for it via AI. Existence without visibility is a real problem for the author's clients, not only for him. The text resolves this aesthetically, not functionally. **The privilege of competence.** The final advice — "test it yourself, check the portfolio, think about your audience" — is useful for someone who already knows what testing, evaluating a portfolio, or mapping an audience actually involves. For the mayor of a small town who receives the same email and lacks eight years of web experience, this advice is insufficient. The text is written, in effect, for a reader with competencies similar to the author's own. Everyone else — the audience most vulnerable to Marius's pitch — is left unserved. --- These are not reasons not to publish it. They are reasons to read it with open eyes. ---