GEO: visibility in AI assistants and AI search
GEO (generative engine optimization) is work aimed at getting ChatGPT, Perplexity, Gemini, DeepSeek, Yandex Neuro and Google AI Overviews to name your company when answering customer questions. We start with measurement: which questions are asked, whom assistants name today and why. Then we prepare content that models can cite and the site's technical base for AI crawlers.
Measuring brand visibility
We compile the questions customers bring to assistants: choosing a contractor, comparing solutions, asking about prices and terms. We run them through ChatGPT, Perplexity, Gemini, DeepSeek and Yandex Neuro in several series and record whom the models name, which sources they cite and how they describe your company. That is what we did for Neoved: 29 prompts, two runs, a map of the competitors the assistants name. The plan is built on this baseline and progress is measured against it.
Content that models cite
Assistants rely on pages with concrete answers: definitions, comparisons, figures, terms, step-by-step instructions, answers to questions. We restructure key pages so that every important question has a short verifiable answer, add FAQ sections, tables and structured data. We also work on mentions beyond the site: directories, industry platforms, publications and reviews, because models assemble the picture from several sources.
Technical base for AI crawlers
We check access for GPTBot, ClaudeBot, PerplexityBot, YandexBot and other crawlers in robots.txt and on the server, content delivery without JavaScript, speed and correctness of structured data. We add an llms.txt file describing the company and key pages and submission of new URLs through IndexNow. For Free Lines Company we wrote a Bitrix module with an IndexNow queue and an llms.txt generator and checked access for nine AI crawlers.
What the work includes
GEO is run as a measurable programme alongside SEO, not as a one-off report.
- Question list and baseline visibility measurement in assistants
- Map of competitors and sources cited by the models
- Change plan for key pages: answers, FAQ, tables, structured data
- AI crawler access check and setup, llms.txt, IndexNow
- Work on external mentions: directories, publications, reviews
- Repeat measurements and a report on mention dynamics
How the work is organised
-
Step 1
You send a link to the website, the list of services and competitors, access to Yandex Webmaster and Search Console.
-
Step 2
We compile the questions, run the baseline measurement in assistants and a technical audit of the site for AI crawlers.
-
Step 3
We prepare a plan: which pages to change and how, which external mentions are needed, what to add to structured data and llms.txt.
-
Step 4
We implement changes with your team or as part of support, repeat the measurements and compare with the baseline.
Cost of GEO
The cost depends on the number of questions and assistants in the measurement, the volume of pages to improve, the state of the site's technical base and the need to work on external mentions. The baseline measurement and audit can be ordered separately to understand the starting position before a change programme. For an estimate, send a link to the website and the list of services where visibility matters.
Frequently asked questions
How does GEO differ from SEO?
SEO works with rankings and traffic from search results. GEO works with whether assistants name your company in answers and which sources they cite. The foundation is shared: a technically clean site and useful content, but GEO adds measurement in assistants and work on external mentions.
Can a mention in ChatGPT be guaranteed?
No. Model answers vary from request to request and from version to version. We measure the share of answers mentioning the brand across a series of prompts and work to increase it, but a fixed result cannot be promised.
How soon are changes visible?
Models with search access reflect site changes within weeks after recrawl; models without search access update less often. That is why repeat measurements are planned at one- to two-month intervals.
Should the site be open to AI crawlers?
That is the owner's decision: access gives a chance to appear in answers, but content is used for training. We usually open crawlers that build answers with sources and record the rules in robots.txt.
Does GEO work for Yandex?
Yes. Yandex Neuro and AI answers in search rely on indexed pages and Webmaster data. The same content and structured data changes affect visibility there as well.