GEO: Visibility in ChatGPT and AI Search

People increasingly choose products on the advice of ChatGPT, Perplexity and other AI assistants. GEO (Generative Engine Optimization) is the work of making assistants know your product and name it. Two things about this field are worth saying plainly before you buy. First, the result is measurable, and I measure it. Second, much of what is sold under this name does not affect the result. On a sample of about 300,000 domains, no measurable link was found between having an llms.txt file and being cited by AI. I still create the file, because it is cheap and some crawlers read it, but it is not the argument. What the research does support is different: structured entity markup, content shaped as answers to real questions, and sources cited inside your own claims. Two verifiable examples. My own service Media4Estate: ChatGPT names it first in its niche, real clients arrive from there, and there is no ad budget. Case study: mykhailo.koblychenko.com/chatgpt-case. And this site: per-page entity markup, question blocks on commercial pages, real 404s, published research with 29 sources. Everything I sell here is running here, and you can open the page and check it. Packages: 1. GEO Audit, EUR 890. How AI crawlers actually see your site: access (GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot), whether a page returns full HTML or an empty shell, whether the markup is valid, whether non-existent addresses return 404 or quietly return 200, machine readability of key pages. You get the issue list, the order in which to fix them, and a baseline measurement across a fixed set of control prompts in four assistants, so there is a point to compare against. 2. GEO Setup, EUR 2,400. The audit plus implementation. Per-page entity markup rather than one global block: Organization, Person, Service with offers, FAQPage, Article, BreadcrumbList, linked by @id into a single graph. Question blocks on commercial pages. Crawler access and real 404s instead of soft ones. Canonical and hreflang for multilingual sites. IndexNow into the Bing index that ChatGPT Search reads. Measurement wired up: an AI traffic channel, a "how did you hear about us" field in your form, and a repeat of the control prompts. llms.txt is included as hygiene. 3. GEO Turnkey with Content, EUR 4,900. Everything in Setup plus the content assistants actually cite: pages answering the real questions of your niche, with sources linked inside your own claims. That last part is the one technique with a directly measured effect, up to +115% visibility for pages sitting fifth in search. Plus a before-and-after comparison across ChatGPT, Perplexity, Claude and Gemini. 4. Ongoing visibility, EUR 990/mo. Weekly control-prompt checks with a written log, fresh content, markup kept in step with changes to the site, adaptation to shifts in AI search, and a monthly report: what changed and what it cost. The honest limit: nobody controls what an assistant answers, and no position can be promised. What can be done is to remove every reason not to name you, and to show the movement in numbers rather than in assurances. The data this is built on: https://www.omnibound.ai/blog/generative-engine-optimization-statistics Before the call: the technical part of the audit runs free at mykhailo.koblychenko.com/geo-check, twelve checks in thirty seconds. The paid audit adds what automation cannot: a baseline across control prompts in four assistants, a review of the content and entities, and a prioritized plan for your team.