First I tried the old approach: many city pages
Recently I launched websites with different structures and algorithms. In one experiment I created dynamic pages that automatically substituted city names and search phrases.
The logic seemed reasonable. Take about 20 service pages, multiply them by 10 cities and you have 200 pages. Add an English version and the number doubles. Astro builds the site quickly, the pages load quickly and search engines discover them quickly. It looked perfect.
The result was less attractive. The pages initially entered the index, but a few days later a large share disappeared. This happened particularly quickly in Yandex.
I cannot prove that city substitution alone caused the removal. Similar content, actual query demand, canonical URLs, internal structure and other factors all affect indexing. Yandex itself says that similar or low-demand pages can remain outside search even when they are technically sound.
My conclusion is simple for now: page count by itself produces no value. If a city page differs only by a city name, it has little reason to exist. It needs real local data, examples, terms, prices or another genuine purpose.
The second experiment: one name and one clear context
After that I went in the opposite direction. Instead of multiplying pages on my website, I started building a coherent digital footprint around one person.
The rules were simple:
- use the name “Roman Mikhailov” everywhere;
- use the same photograph or a recognisable visual identity;
- keep one specialisation: websites, AI agents, bots and CRM integrations;
- write clear descriptions without contradictions;
- link to the RM Systems website wherever a platform allows it;
- use no paid promotion.
The experiment is not finished. Some pages are very new and search engines have not fully crawled them. However, Yandex Alice has already begun to show me occasionally in answers and recommend me as a developer.
I cannot honestly claim that the new profiles alone caused this. The website, older pages, articles, business listings and other signals may all have contributed. For now I only see a correlation in time and continue to observe it, but the correlation was interesting enough to document.
Where I created profiles
The pages below are real. I do not claim that every one directly improves rankings. The point is different: several independent sources now connect the same name, photograph, expertise and examples of work.
Freelance platforms
On FL.ru I published individual services and portfolio items. Moderation can take time and direct enquiries are still limited, but the public profile contains my name and clearly shows what I do.
Moderation is faster on Kwork. It also makes it easy to present separate, understandable services instead of one vague “I do everything in IT” offer.
I also created a profile on Freelance.ru. At this stage it is another new point in the experiment rather than a proven source of traffic. I will evaluate it after indexing.
Yandex Services and Yandex Business
For Yandex I completed both a Yandex Services profile and the RM Systems listing in Yandex Business.
I would not stop at a company name and phone number. I added services, prices, photographs, descriptions and a website link. Even without considering AI, a complete listing is clearer both to a person and to the search engine.
After completing these pages I began to notice better answers about me in Yandex Alice more often. This is an observation, not proof of a formula such as “complete a listing and enter AI answers.”
Google Business Profile
I also completed the RM Systems Google Business Profile, although the result is less clear so far. Information about me sometimes appears in Google answers, but one listing and a short observation period are not enough for a confident conclusion.
The practical priority is to keep the profile complete and collect genuine reviews after real work. Google officially states that complete information, prominence, links and reviews may affect local search. It makes no equivalent guarantee about a direct effect on AI answers.
GitHub, Habr Career and Behance
GitHub matters for a developer, but search engines cannot see private repositories. A public footprint needs a complete profile, description, website link, README files and public repositories or code examples that can actually be crawled.
I completed Habr Career as a professional profile. It has no dedicated case-study section, so I use the available fields: specialisation, experience, skills and description.
On Behance I published visual presentations of projects. It is not my primary sales channel; it is another place where my name is connected to specific work such as an AI landing page and a bot.
I am not rushing to publish on Habr itself. That platform needs a genuinely useful technical article, not an advertisement or an empty profile created for a link. I will add it when I have the right material.
Workspace, Profi and Avito
On Workspace I added a profile and case studies. The platform is useful because it lets me show not only a list of services but also the task, process and outcome of a project.
On Profi I added services, prices and explanations. It is another public specialist profile with the same name and expertise.
On Avito I published a separate listing and completed a brand page. In this experiment Avito is both a classified platform and another public brand page.
LinkedIn, TenChat and social networks
A completed profile alone is not enough on LinkedIn or TenChat. They need at least a modest active trail: posts, projects and observations. I chose a simple rhythm of occasional useful posts rather than a daily content race.
I also use Instagram, a Telegram channel, X and VK. I do not expect every network to bring clients independently. Their simpler role is to show that the same name and work exist beyond one young website.
Links are not a magic button
It is easy to draw the wrong conclusion: add twenty links to a website and wait for search engines to merge everything. I do not believe that works.
A link from my site helps a crawler discover a profile and understand its context. A two-way and consistent picture is stronger: the site lists the profile, the profile links back, the name matches, the description does not change from “developer” to “marketer,” and real work and publications appear nearby.
I therefore use links as navigation through my genuine digital footprint, not as purchased SEO. They lead to real pages and help a reader verify my experience.
The technical layer on the website
The central page of this footprint is Roman Mikhailov, RM Systems developer.
It should bring together:
- full name and photograph;
- expertise and description of the work;
- projects and articles;
- contact details;
- links to external profiles.
For crawlers I also use structured data. An article is marked as Article; its author is a Person linked to /en/about. The author page uses Person or ProfilePage, while sameAs lists verified external profiles.
This is not a command to a search engine and does not guarantee that every page will be merged. The markup simply describes who wrote the article and which page identifies that person. Google specifically recommends an author URL or sameAs value.
No separate “secret AI file” is required. Google says that its AI search features need no special markup or new machine-readable file. They need an indexable page, useful visible text and structured data that matches that text.
One more important point: an article date must not be changed merely to look fresh. dateModified should represent a real update, not decoration.
What the GEO research says
The term GEO, generative engine optimization, is now used for improving how content appears in generative answers. In a widely discussed paper by researchers from Princeton, Georgia Tech and other institutions, some presentation methods increased content visibility in experimental answers by up to 40 percent.
That number is interesting but easy to misuse. The research used a dedicated query benchmark and measured content visibility. It does not prove that ten profiles will automatically lift a specific person in Yandex Alice by 40 percent.
My more restrained conclusion is that specific facts, clear language, sources and original experience give a generative system material that is easier to use in an answer.
My ten rules for a digital footprint
- Start with a sound website. Technical SEO, indexing, speed and useful content still matter. AI search must first be able to find the page.
- Use one name across platforms. Do not change the spelling, specialisation or core description without a reason.
- Maintain one main author page. Articles and profiles should point to a clear /en/about page.
- Link directly to profiles. Use the actual public pages, not only registration or review forms.
- Link both ways where allowed. The website links to the profile and the profile links back.
- Complete the profile. An empty account with no work, text or photograph proves very little.
- Publish first-hand experience. One real experiment is more useful than ten copies of the same SEO article.
- Keep reviews genuine. They help people decide and may strengthen local search. Buying or fabricating reviews is not acceptable.
- Record the results. Track indexing in Yandex Webmaster and Google Search Console, write down the query and date, and save screenshots of AI answers.
- Do not mass-produce pages with no value. If only a city name and a few keywords change, a search engine may retain one page or exclude the rest as low demand.
The result so far
I did not find a secret AI-SEO button, and I do not believe promises to place a company in AI answers within a week.
The broader principle still seems sound. Search engines and AI services can understand a person more easily when information is not isolated on one young website. There is a central author page, several complete profiles, consistent context, real projects, articles and reviews.
The experiment continues. Some platforms have only begun to be indexed, genuine reviews still need to accumulate and results must be evaluated over time. If the picture changes, I will update this article and show what happened.
For now, my experience, projects and contacts are collected on the Roman Mikhailov developer page. This was done without paid promotion — just a website, public platforms and a great deal of careful manual work.