article · November 9, 2026 · Gregory Shevchenko

Kurenkova.Agency case study: 24 AI citations from zero in the first content month

How a consultancy with zero AI citations earned 24 citations on 18 queries in one content month and moved ahead of its closest competitor by number of cited URLs.


Cited across

  • ChatGPT
  • Gemini
  • Google AIO
  • Google AI Mode
  • Perplexity
  • Copilot

Cover for the case study «24 AI citations from zero in one month»: Kurenkova.Agency

On September 7, 2026, we ran a baseline measurement for Kurenkova.Agency, a Russian consultancy that runs strategy sessions and helps companies with crisis management. The result was zero citations. AI assistants knew nothing about the brand, the website or any of the client's publications.

By November 5, after the first content month, that had changed. AI assistants now answer 18 of 200 client queries with a link to Kurenkova materials. That is 24 citations and 9% coverage. By number of cited URLs, the client already leads its closest competitor, zhdanov.bz: 24 versus 13.

Below we walk through what we did, what worked and what is still missing.

In this article

Section 01

Where we started

The client had three starting problems.

  • A zero baseline. No AI assistant linked to kurenkova-agency.ru.
  • A young domain with no authority. Models do not yet treat it as a reliable source.
  • Expert competitors already in the answers. zhdanov.bz and ecopsy.ru are cited on the topic's queries.

The client's goal was to launch an LLM visibility channel from scratch, so that AI assistants answer niche queries with Kurenkova materials. The path from there is simple. A person asks a question, gets an answer that cites the client's article, visits the site and leaves a request.

Section 02

How the program was built

The program started with analysis and then moved to content.

Analysis. We collected 200 client queries across the stages of choosing a consultant and took the baseline on September 7. The plan followed one logic: first teach AI assistants to read the client's texts, and only then push them to name the brand and link to the site.

Content. We combined two formats. Articles went onto the client's site, and rewrites of them went onto vc.ru, a Russian business media platform, because user-generated platforms reach AI answers faster than anything else. Between October 6 and 20 we published 16 articles on the site, and from October 23 to November 2 we published rewrites on vc.ru. The tracker showed 24 materials against a plan of 18.

Measurement. Each measurement cycle in AI Visibility (Hermes) covers 200 queries across five AI assistants. That makes 800 answers, and we check each one for citations of the client and of competitors. In parallel, publication tracking shows which materials the assistants pick up and which they ignore.

Section 03

What the measurement showed

On 18 of 200 queries, AI assistants answered with a link to the client's materials. The rewrites were indexed within 1 to 4 days, so the goal of reaching the answers within 1 to 2 weeks of publication was met.

Four materials produced the citations, and all four were vc.ru rewrites.

  • "Preparing for a strategy session," published November 2: 8 citations on 6 queries.
  • "How to implement decisions after a strategy session," published October 30: 7 citations on 6 queries.
  • "How to bring strategy session decisions to results," published October 30: 7 citations on 7 queries.
  • "Strategy session: choosing a format for growth," published October 30: 2 citations on 2 queries.

All four texts went live 1 to 4 days before the measurement. The articles on the client's own site are not cited yet.

Section 04

Citations where the buyer chooses a provider

The most useful result of the first wave is the distribution across the funnel. We are cited closest to the purchase decision.

  • Decision, the buyer picks a provider: 8 of 35 queries, 22.9%.
  • Consideration, the buyer compares options: 6 of 35, 17.1%.
  • Evaluation, the buyer assesses the approach: 4 of 35, 11.4%.
  • Adjacent questions: 0 of 39.
  • General top-of-funnel questions: 0 of 56.

The explanation is straightforward. The first articles answered questions like "how to prepare for a strategy session," "how to bring decisions to results" and "which format to choose." That is the language of someone who has already decided a strategy session is needed. The top of the funnel ("what is," "why," "how much does it cost") and adjacent topics will be covered by the reserve topics in the plan.

Section 05

Channels and competitors

Here is how the assistants compare.

  • ChatGPT: a citation on 13 of 200 queries. This is the main channel.
  • Gemini: 8 queries.
  • GigaChat and Google AI Mode: citations are present, with exact counts due in the next measurement.
  • Yandex's Alice: no data. We received 196 answers from a lightweight model and left them out of the result, because that is "no data," not zero.

By number of cited URLs, the topic looks like this: Kurenkova.Agency 24, zhdanov.bz 13, ecopsy.ru 12, strategessa.ru 3. The leading domain in the niche, consultant.ru, collected 156. The platform vc.ru is in the top 8 domains of the topic with 44 citations, 24 of which belong to the client. It is also the only platform where Kurenkova has any materials so far.

Section 06

What has not worked yet

We report the weak spots as well, because they define the next stage.

  • The domain is not cited. kurenkova-agency.ru has zero direct citations. All reach comes through vc.ru. That is normal for a young site, but visits to the site and leads depend on direct citations.
  • The brand is not named. The word "Kurenkova" did not appear in any of the 800 answers. AI assistants read the texts but do not know the author.
  • Only 4 of 24 materials are cited. The other 20 need work, mainly a direct answer moved into the first paragraphs.
  • Alice is not measured. It needs a re-measurement on the full model.

Section 07

What we do next

  1. Add the brand name to author signatures and bios on vc.ru, and place a short "based on the experience of Kurenkova.Agency" line inside the texts. The goal is for AI assistants to start naming the company.
  2. Write four articles for the client's site aimed at the hottest Decision queries. They should deliver the first direct citations of the domain.
  3. Take top-of-funnel topics from the reserve, for example "what is a strategy session," "why a facilitator" and "how much it costs." This widens coverage beyond 9%.
  4. Re-measure Alice and get exact counts for GigaChat and Google AI Mode.
  5. Continue with the next content month. 14 site articles are in production: 8 are already with the client and 6 are approved.

Section 08

What you can take from this

  • Rewrites on user-generated platforms deliver the first citations faster than a young domain. If your site has no authority yet, start with platforms like vc.ru or their local equivalents, but do not stop there.
  • Build the content plan around the questions people ask before choosing a provider. In our measurement, exactly those texts earned citations.
  • Zero brand mentions alongside citations is a separate problem. Solve it with author signatures and a branded line inside the text, not with more articles.
  • Count only what was actually measured. We did not count the lightweight-model data for Alice, and that principle should hold in every report.

Section 09

Want the same measurement for your niche

We will show how AI assistants answer the queries in your niche today and which of them you can take first. Contact us and we will run a baseline measurement.

Section 10

FAQ

How quickly did the first citations appear?

The vc.ru rewrites reached AI answers 1 to 4 days after publication. All 24 citations came from four materials that went live 1 to 4 days before the measurement.

Why are the articles on the client's site not cited?

The domain is young and has not yet built authority as a source. 16 articles were published on the site and none has a direct citation yet. That is why the plan includes four articles for Decision queries and work on the domain's authority.

What does 9% coverage mean?

AI assistants answered with a link to the client's materials on 18 of 200 queries. Five assistants take part in each measurement cycle, for 800 answers in total.

Why was Alice left out of the result?

We received 196 answers from a lightweight model and do not count them. That means a lack of data, not a lack of citations. A re-measurement on the full model is planned.

How do we know the citations lead to leads?

So far we have confirmed only the first step of the chain: question, answer with the article, visit, request. The next milestone is direct citations of the domain and a mention of the brand by name.

For your team

Stop hiring agencies and freelancers

Hire not agencies and freelancers — but Marketing AI Agents for the AI Search.

  • Per-engine citation map across 6 AI engines
  • Content + schema work that earns the citation
  • Honest 30-min strategy call before you commit

Cited across

  • ChatGPT
  • Gemini
  • Google AIO
  • Google AI Mode
  • Perplexity
  • Copilot


Want to talk?

Book the strategy call. Thirty minutes, free.

An engineer from the team runs your brand through Hermes before the call.

You arrive to a per-engine citation map of your category, the closeable gaps, and an honest read on whether any tier fits.