March 23, 202617 min read

5 Niches, 716 Brands, 8 AI Providers: Key Findings from the GEO Scout Study

A comprehensive study of AI visibility for Russian brands across 5 niches: EdTech, E-commerce, FinTech, tourism, and hosting. 716 brands, 8 neural networks, 10 key findings — from Yandex.Market records to ChatGPT paradoxes.

Vladislav Puchkov

Founder of GEO Scout, GEO optimization expert

Neural networks are already making decisions for users. Which bank to choose. Where to order hosting. Which platform to study on. Previously, a user would open 10 browser tabs, compare reviews, and read articles — today they ask ChatGPT or Alice a single question and receive a ready-made shortlist of 3-5 brands.

If your brand is not on that shortlist, you've lost the battle for the customer before they even visited your website.

At GEO Scout, we conducted a large-scale study: 5 niches, 716 brands, 8 AI providers, thousands of neural network responses. This is the first public summary of the results — with concrete numbers, paradoxes, and practical takeaways.

About the Study

Scale

The GEO Scout study covers five major niches of the Russian digital market:

• EdTech — online education (Netology, Skillbox, Yandex Practicum, and others)
• E-commerce — marketplaces and online stores (Yandex.Market, Ozon, Wildberries, and others)
• FinTech — banks and financial services (Alfa-Bank, Tinkoff, Sberbank, and others)
• Tourism — travel services (Yandex Travel, Aviasales, Ostrovok, and others)
• Hosting — hosting providers (Timeweb, Beget, REG.RU, and others)

Total volume: 716 brands, monitored daily across 8 AI providers.

AI Providers

Data was collected from 8 platforms: ChatGPT, Claude, DeepSeek, Gemini, Google AI Mode, Grok, Perplexity, YandexGPT (Yandex with Alice).

Each provider responds to commercial queries in Russian — "which bank to choose," "best hosting for an online store," "where to learn Python." We track: which brands are mentioned, at what position, with what tone, whether the domain is cited, and whether a direct recommendation is given.

Key Metrics

Data snapshot — March 23, 2026. Full rankings with daily updates — on the GEO Scout ratings page.

Leaders of Each Niche

Each leader is the brand AI mentions most frequently in its niche. But "mentions most often" and "recommends most often" are, as we will see, entirely different things.

Detailed breakdown for each niche:

• EdTech brand rankings
• E-commerce rankings
• FinTech rankings
• Tourism brand rankings
• Hosting rankings

Records Across All Niches

When you compare 716 brands from 5 niches, unexpected leaders emerge — those who stand out not within their industry, but against everyone.

Yandex.Market is mentioned in virtually every e-commerce response — 97 out of 100. This is a record across all niches and providers.

Wildberries with position 1.31 shows that when AI lists marketplaces, Wildberries almost always comes first. This is stronger than any brand in any other niche.

Aviasales is the absolute record holder for recommendations. 36.6% — that is 18 times higher than the fintech leader (2.07%). AI readily recommends a flight search service but won't take responsibility for choosing a bank.

Timeweb with a 94.09% positive rate is a brand that AI almost never says anything negative about. 94 out of 100 mentions are in a positive context.

HostingHUB — a paradox: mention rate is only 22.45%, but domain citation is 49.39%. AI cites its content as a source without mentioning the brand itself. This is pure content-driven GEO in action.

Netology — the EdTech leader, but with the worst positive rate among all niche leaders: 63.11%. AI frequently mentions Netology, but not always in a positive context.

10 Key Findings of the Study

1. Business Size Does Not Equal AI Position

This is perhaps the most counterintuitive finding of the entire study. The largest companies — with billions in revenue, thousands of employees, federal advertising — don't necessarily rank first in neural network responses.

Sberbank — Russia's largest bank with assets exceeding 40 trillion rubles — ranks 4th in the FinTech AI visibility rating. It is surpassed by Alfa-Bank, Tinkoff, and VTB. Selectel — a major provider with its own data centers — is only 7th among hosting companies. Amazon — the global e-commerce giant — is 14th in our ranking.

Neural networks don't know about your revenue. They know about your content. AI builds its worldview from texts: blogs, expert articles, reviews, mentions on authoritative platforms. If a brand generates quality content that makes it into AI training data, it gains visibility. If a brand relies on advertising and offline presence, AI may simply not know about it.

This fundamentally changes the competitive landscape. A mid-sized brand with a strong content strategy can outrank an industry giant in AI responses. And this is already happening.

2. ChatGPT Doesn't Know the Russian Market

ChatGPT is the world's most popular AI assistant. But when it comes to Russian brands, it shows a systematic gap.

In EdTech: SkillFactory gets 6.7% mentions in ChatGPT, Skypro — 0%, ProductStar — 0%. Meanwhile, Coursera — 86.7%. In hosting: AdminVPS — 0%, SprintHost — 3.3%. While DigitalOcean — 56.7%.

The pattern is obvious: ChatGPT favors international brands. It is trained primarily on English-language data, and Russian companies make it into its worldview only if they actively produce English content or have international positioning.

For Russian businesses, this means: you cannot evaluate AI visibility by ChatGPT alone. Your brand may be invisible in ChatGPT but lead in Gemini, Google AI Mode, or YandexGPT. Monitoring a single provider gives a distorted picture. You need a multi-provider approach — and that is exactly what GEO Scout provides.

3. AI Only Recommends Where Risk Is Low

The highest recommendation rate in the entire study is 36.6% for Aviasales in tourism. At the same time, the FinTech leader Alfa-Bank gets only 2.07%, and the EdTech leader Netology — 1.26%.

Why such a difference? AI is aware of responsibility. Recommending a flight search service is low risk: the user will still choose the flight and airline themselves. Recommending a bank means taking responsibility for a financial decision. Recommending an online school means taking responsibility for an educational choice.

Neural networks use three strategies instead of direct recommendations: listing options ("consider the following"), comparative analysis ("depends on your needs"), and conditional recommendations ("if price matters — then X, if quality — then Y").

For brands, this means: don't chase recommendation rate in high-risk niches. Focus on mention rate and position — these determine whether your brand makes it onto the user's shortlist.

4. Praising Doesn't Mean Recommending

One of the study's most striking paradoxes. AI can speak about a brand exclusively in positive terms — while still not giving a direct recommendation.

Ozon: 92.24% positive rate, but only 4.22% recommendation. Timeweb: 94.09% positive — a record across all niches, but recommendation is just 2.45%. The neural network praises the brand, describes its advantages, highlights its strengths — but doesn't say "I recommend it."

This is a fundamental difference between "AI knows the brand is good" and "AI is willing to vouch for it." Positive tone is a necessary but insufficient condition for a recommendation. AI must not only know about the brand's quality but also consider it objectively the best choice for the specific query.

For GEO strategy, the practical takeaway: work on both metrics in parallel. Positive tone is built through expert content and reviews. Recommendations come through unique positioning and evidence base (comparisons, rankings, case studies).

5. Google AI Mode — the Best Friend of Russian Brands

Among the 8 providers, Google AI Mode shows the deepest understanding of the Russian market. Virtually all top brands in every niche receive 90-100% Share of Voice in Google AI Mode.

The specific numbers are impressive. Best positions: Timeweb — 1.43 (almost always first), Skillbox — 1.57. Recommendations: Aviasales — 50% (every other response is a direct recommendation). In EdTech: Netology — 96.7%, Skillbox — 100%.

Google AI Mode relies on its search index — and here Russian brands have an advantage. Years of SEO work, Google optimization, quality Russian-language content — all of this is taken into account. If your brand ranks well in Google search, there's a high probability that Google AI Mode will actively mention it.

This is a key strategic insight: SEO and GEO are allies, not competitors. Brands with strong SEO gain an advantage in Google AI Mode, which is one of the fastest-growing AI providers in Russia.

6. Domain Citation — a New Metric That Everyone Ignores

Most brands focus on mention rate — how often AI mentions the brand. But there's a metric that may be even more important: domain citation — how often AI cites the domain as an information source.

HostingHUB: mention rate 22.45%, domain citation 49.39%. AI mentions HostingHUB as a hosting brand in only one in five responses. But it references its content as a source in every other one. Similarly, HowTrip in tourism: mention 13.62%, citation 20.85%.

This is a fundamentally different model of AI visibility. Content sites — blogs, reviews, comparison rankings — receive citations even if AI doesn't mention them as a brand. The neural network uses their content to form responses and references them as sources.

For businesses, this opens a new channel: content-driven GEO. You can get traffic from AI not through brand mentions, but through citation of your content. Create expert reviews, guides, comparisons — and AI will cite your domain, driving users to your website.

More on domain citation — in the article who AI cites directly.

7. Sanctions and Blocks Kill AI Visibility

AI providers know about blocks and sanctions — and this directly affects how they mention brands.

Booking.com: positive rate only 8.86%, negative — 5.06%. For comparison, the niche leader Yandex Travel has a positive rate of 84.15%. Airbnb: negative rate 12.24%. In YandexGPT, the situation is even harsher: Booking — 0% mentions, Google Flights — 0%.

AI doesn't just know about blocks — it takes them into account when forming responses. The neural network understands that recommending a blocked service to a Russian user is pointless or harmful. So even if a brand is globally well-known, in the Russian context AI may completely ignore it.

For services blocked in Russia, this effectively means losing the AI channel in the Russian market. And for their competitors, it means open opportunities: when AI doesn't mention Booking, it redistributes attention to Ostrovok, Yandex Travel, and Aviasales.

8. Gemini — the Most Complete Coverage of Russian Brands

If Google AI Mode is the best friend of Russian brands in terms of depth of knowledge, then Gemini is the champion in coverage. Gemini demonstrates 100% mention rate for leaders of virtually every niche.

Netology — 100% in Gemini. Skillbox — 100%. Yandex Practicum — 100% with position 1.5 (almost always first or second). Yandex.Market — 100%. Ozon — 100%. Timeweb — 100%.

Gemini knows the Russian market and knows it well. Unlike ChatGPT, which skips entire segments of Russian brands, Gemini includes them in every relevant response. Moreover, Russian brand positions in Gemini are consistently high — they're not just mentioned, they're placed first.

For brands with good AI visibility in Gemini, there's a practical takeaway: your content is working. Gemini confirms that AI systems can deeply understand the Russian market — it's all about the training data. If ChatGPT ignores you, the problem is not your content, but ChatGPT's training data priorities.

9. Perplexity — the Citation King

Perplexity holds a special place among AI providers. Its model — responses with sources — makes domain citation the key metric. And it's in Perplexity where we see record citation numbers.

HostingHUB — 90% domain citation in Perplexity. Yandex Practicum — 56.7%. Trip.com — 40%. Perplexity more frequently than other providers links to specific domains in its responses.

This makes Perplexity strategically important for content marketing. If your site generates expert content that Perplexity considers authoritative, you get direct links from AI responses. This isn't just a "brand mention" — it's real traffic.

For Perplexity optimization, the same principles as SEO apply: structured content, expert depth, source authority. But there's a nuance: Perplexity especially values fresh and up-to-date content. Regular updates, current comparisons, and guides — that's what gets cited by Perplexity.

10. The Top Three Capture 35-43% of Share of Voice

In each of the five studied niches, we observe the same pattern: the top 3 brands accumulate 35% to 43% of all Share of Voice. The remaining dozens and hundreds of brands share what's left.

This means competition in AI is not a "everyone vs. everyone" fight. It's a fight for a spot in the top three. If you're in the top 3, you get a disproportionately large share of mentions, recommendations, and citations. If you're fourth or below, you're fighting for leftover attention.

The Share of Voice structure in AI resembles a power law: the first gets more than the second, the second more than the third, and the gap between third and fourth is steep. This is not a linear distribution — it's exponential.

For GEO optimization strategy, the takeaway is clear: identify your niche and fight for the top 3. It's better to be first in a narrow subcategory than tenth in a broad one. If you can't beat Yandex.Market across all of e-commerce — take first place in AI responses for a specific segment.

What Unites Leaders Across All Niches

We compared the strategies of five leaders — Netology, Yandex.Market, Alfa-Bank, Yandex Travel, Timeweb — and found five common traits.

Strong Digital Presence

All leaders actively maintain blogs, publish expert articles, and create educational content. Netology — dozens of articles about careers and education. Alfa-Bank — financial analytics and guides. Timeweb — technical guides and reviews. This content makes it into AI training data and forms a positive brand image.

Structured Data on the Website

Schema.org, OpenGraph, quality meta descriptions — all of this helps AI systems "understand" your site. Rating leaders invest in the technical side of GEO: structured data, proper navigation, semantic markup. More on the technical side — in the article GEO site audit.

Presence on Authoritative Platforms

vc.ru, Habr, RBC, Tinkoff Journal — all leaders actively publish on platforms that AI considers authoritative sources. This isn't just PR — it's shaping training data for neural networks. When AI sees a brand on ten authoritative platforms, it considers the brand authoritative.

Social Media Activity

Telegram channels, YouTube, podcasts — leaders are present where public opinion is formed. AI considers not only text content but the overall digital footprint of the brand.

High Positive Rate

All leaders except Netology (63.11%) show a positive rate above 80%. This is not a coincidence — positive AI tone is formed from positive content. Quality reviews, success cases, expert comments — all of this creates a positive context around the brand in the data AI is trained on.

What To Do: Practical Recommendations

For Brands Not in AI Responses

Problem: AI doesn't mention your brand at all or mentions it extremely rarely (mention rate < 20%).

Actions:

1. Audit current visibility. Check how AI responds to queries in your niche. Use GEO Scout for multi-provider monitoring — don't limit yourself to ChatGPT alone.
2. Content expansion. Start publishing expert content on authoritative platforms: vc.ru, Habr, industry media. AI is trained on this data.
3. Technical GEO. Implement structured data (Schema.org), optimize meta tags, create FAQ sections on key pages.
4. Monitor progress. AI visibility changes more slowly than SEO positions. Expect initial results in 2-3 months of systematic work.

For Brands in the Top 10 but Not the Top 3

Problem: AI knows your brand but places it 4th-10th. You get mentions but don't make it to the shortlist.

Actions:

1. Analyze the leaders. Study what brands in your niche's top 3 are doing. What content do they create? Where do they publish? What does AI cite?
2. Differentiation. Find a subcategory where you can be first. Better to be #1 for "WordPress hosting" than #7 for "hosting" overall.
3. Work on tone. If your positive rate is below 80%, look for negative triggers. What does AI know about your problems? How can this be fixed through content?
4. Domain citation. Create content that AI will want to cite: comparisons, rankings, guides, calculators.

For Niche Leaders

Problem: You're in the top 3, but ChatGPT ignores you. Or your recommendation rate is low.

Actions:

1. Multi-provider strategy. Don't focus on one provider. Your goal is stable presence across all 8. Weak spots with specific providers are growth opportunities.
2. English-language content. For ChatGPT — create content in English. This is the only way to break into its training data.
3. Recommendation-building content. To increase recommendation rate, publish objective comparisons, case studies with specific metrics, reviews with facts.
4. Defend your positions. Monitor competitors. If someone is rapidly growing in AI visibility — analyze what they're doing.

Full Rankings

This is an overview article — a summary of key findings. Detailed analysis of each niche with full tables, provider breakdowns, and individual recommendations is available in separate publications:

• EdTech brand rankings in neural networks — 139 brands, from Netology to Coursera
• E-commerce rankings in neural networks — marketplaces, from Yandex.Market to Amazon
• FinTech rankings in neural networks — banks and financial services
• Tourism brand rankings — from Yandex Travel to Booking
• Hosting provider rankings — from Timeweb to Selectel

Full rankings with daily updates — on the GEO Scout ratings page.

Conclusion

716 brands. 5 niches. 8 AI providers. Thousands of neural network responses. And one main conclusion: AI visibility is a new competitive advantage that cannot be bought with money.

Business size doesn't matter. Advertising budgets don't help. ChatGPT may not know your brand even if you're the market leader. But if you create the right content, publish on authoritative platforms, and work on the technical side of GEO — AI will mention, recommend, and cite you.

Neural networks are already creating shortlists for users. The only question is whether your brand will be on that shortlist — or you'll hand that spot to a competitor.

Start with an audit: GEO Scout will show you exactly how each of the 8 AI providers sees your brand today. And to avoid deciphering metrics on your own, the Command Center automatically turns monitoring data into a prioritized action plan: what content to create, which platforms to strengthen, what to fix technically.