How Intellms outranked every vendor competitor in AI answers except itself

Intellms builds enterprise services: ERP systems, LLM assistants, and AI solutions for business. At the start of the project the brand was almost invisible to AI engines: 18 site citations, coverage of 8.8% (only 14 of 160 queries returned an answer with a link to Intellms), and 11th place in the competitive field.

Worse, the Decision cluster, queries like "order development," "how much does an ERP system cost," and "implement an LLM solution," delivered only 7.5%. When the brand did show up in AI answers, it appeared as a source of expertise, not as a supplier worth contacting.

The client's task came down to three points:

• appear in ChatGPT, Perplexity, Yandex, and Google AI Overviews answers on enterprise development and LLM topics;
• overtake direct vendor competitors in AI results;
• build a three-month content strategy with a measurable result.

We used 160 LLM queries split into four funnel clusters (Awareness, Consideration, Evaluation, Decision), 40 queries per cluster, and five measurement platforms: Perplexity, Yandex, ChatGPT, Google AI Overviews, and Gemini. Over the period we ran three control measurements: July 17 (point A), and August 18 and 21 (points B and C). Each was a full 800-step scan.

First we looked at the competitive field in the niche and saw what became the basis of the strategy. 56% of all citations in the niche go not to vendors but to UGC media, habr.com and vc.ru. So the plan was built both around the client's own site and around publications on sites that AI engines already cite willingly.

Out of 78 topics collected in the content backlog, we identified a winning topic: ERP for large companies. A single page on this topic collected 8 citations out of 44 at the time of analysis, which is 18% of all citations at the start of the program.

During the publication period, 13 pieces that received citations were published on the site. Besides the ERP leader, the top included articles on SaaS in the corporate environment, choosing an LLM solution, ERP comparison, building an internal LLM model, and business process automation.

Next we built a three-month content plan: 35 topics out of 78 in the backlog, approved by the client. The plan moves the funnel from top to bottom instead of growing everything at once: September closes Consideration, October is the transition to Decision, and November is Decision topics proper.

In 35 days site citations grew 2.9 times, coverage rose from 8.8% to 15%, and the place among the niche's domains went from 11th to 3rd. The picture by platform is uneven, and we discuss it separately below, because some of the zeros are measurement failures.

Awareness, Consideration, and Evaluation grew together and noticeably, 3 to 3.75 times each. One in three queries like "what is ERP" or "how to choose an LLM solution" now gets Intellms in the answer. But queries like "how much does it cost," "how to order," and "development timeline" are exactly where they were in July. AI engines recognized Intellms as an expert. They are not ready to treat it as a supplier yet.

Intellms overtook all direct vendor competitors, except the UGC media, which are still out of reach in this niche.

At the same time habr.com and vc.ru work differently in this niche. Habr is cited twice as often and covers 28% more queries thanks to technical depth (LLMs, development, architecture). VC.ru is cited less often but is stronger on commercial topics, vendor comparisons, and price breakdowns. The conclusion for our own content strategy on these sites: publishing the same material on both is inefficient, and specialization is needed. Habr for technical deep dives, VC.ru for commercial cases.

There is another layer the figure of 53 citations does not show. Perplexity, the channel that delivered almost all of this growth, links to intellms.ru pages as a source but almost never names the brand in the answer text. All 160 of 160 queries were measured, but brand mentions are close to zero.

The user sees the Intellms URL in the list of sources. They do not see the phrase "Intellms recommends" or "according to Intellms" in the answer itself. That is better than being absent, but it is not the same as the model recognizing the brand.

The plan for the next three months is built around two gaps that are already visible in the numbers.

• Decision. The 35 approved topics in the content plan are deliberately shifted to the bottom of the funnel. In November all 10 pieces cover ordering, cost, and timelines, which Intellms content has almost none of today. The goal is to raise Decision from 7.5% to 12%+ by the end of November.
• ERP as a scaling point. Since one page delivered 15% of all citations, it makes sense to build a cluster around it: comparisons by company size, ERP versus custom development, a market overview, a cost breakdown, and implementation cases. Each such page is potentially another 5 to 8 citations.

If the plan is carried out, the forecast for the next measurement in three months is: citations rise from 53 to 80-100, coverage grows from 15% to 22-25%, and the Decision cluster rises from 7.5% to 12-15%, with the goal of securing a top-2 place among the niche's domains and passing vc.ru.

• Citations grew 2.9 times in 35 days, the brand rose from 11th to 3rd place in AI results, and it passed all direct vendor competitors.
• Growth is uneven across the funnel: the top and middle (Awareness, Consideration, Evaluation) grew 3 to 3.75 times, while Decision did not move. That is the main priority of the next stage.
• Citation and mention are different metrics. 53 citations on Perplexity barely convert into brand recognition by name, which becomes a separate task.
• Some platforms (ChatGPT, Google AI Overviews, Gemini) are not yet measured correctly. That is a blind spot in the data, not confirmed absence of the brand.