March 23, 202615 min read

Yandex with Alisa vs ChatGPT: How They See Russian Brands Differently

Comparing YandexGPT and ChatGPT recommendations across 5 niches: EdTech, e-commerce, FinTech, travel, hosting. GEO Scout data from March 2026 — who the Russian AI knows vs the international one.

Vladislav Puchkov

Founder of GEO Scout, GEO optimization expert

The two most popular AI assistants in Russia are Yandex with Alisa (YandexGPT) and ChatGPT. One was built in Moscow, trained on the Russian-language internet. The other in San Francisco, on an English-language corpus with added multilingual data. When a user asks "which online course should I choose" or "where to buy hosting," each gives its own answer. And these answers differ radically.

At GEO Scout, we daily monitor how 9 AI providers see Russian brands. In this study, we compare YandexGPT and ChatGPT head-to-head for the first time — across 5 niches, dozens of brands, and hundreds of responses. The results expose a systemic issue: a brand visible in one AI can be invisible in another.

Methodology

Data collected by the GEO Scout platform in March 2026.

Parameters:

• AI providers: YandexGPT (Yandex with Alisa) and ChatGPT (GPT-4o)
• Niches: EdTech, e-commerce, FinTech, travel, hosting
• Metric: Share of Voice (SoV) — the percentage of responses in which AI mentions the brand
• Prompts: commercial Russian-language queries — "which bank to choose," "best hosting," "where to take a Python course," and similar
• Domain citation: percentage of responses in which AI cites the brand's domain (includes a link)

Share of Voice of 96.7% means: out of 30 relevant responses, AI mentioned the brand in 29. The higher the SoV, the more frequently the brand appears in commercial query responses.

Full rankings across all 9 providers on the GEO Scout ratings page.

EdTech: YandexGPT knows everyone, ChatGPT knows only Coursera

Online education is the niche where the gap between providers is greatest. YandexGPT confidently navigates Russian EdTech, while ChatGPT systematically misses it.

Netologia and Skillbox are undisputed leaders in YandexGPT: 96.7% SoV for both. Alisa mentions them in virtually every online education response. ChatGPT gives these same brands just over half — 43-53%.

Yandex Practicum shows the largest absolute gap: 90% in YandexGPT versus 30% in ChatGPT. A 60 percentage point difference. YandexGPT, predictably, promotes its own ecosystem product. ChatGPT knows about it but doesn't consider it a priority.

Coursera is the mirror situation. ChatGPT recommends the international platform in 86.7% of responses. YandexGPT — in 3.3%. For YandexGPT, Coursera barely exists as an option for the Russian user.

Skypro is a brand invisible to ChatGPT: 0% SoV, complete absence from responses. Meanwhile, YandexGPT not only mentions Skypro (46.7%) but cites the domain in 40% of responses. For Skypro, ChatGPT is a dead channel, while YandexGPT is a traffic source.

The EdTech takeaway: if you're a Russian EdTech platform counting on AI traffic from ChatGPT — without targeted GEO optimization for English-language sources, you won't be there. If you're an international platform like Coursera — YandexGPT doesn't know you.

E-commerce: rare agreement between two providers

E-commerce is the only niche where YandexGPT and ChatGPT show similar results for top brands. Marketplaces dominate the information space so thoroughly that even AI providers with different data corpora reach the same conclusions.

Yandex.Market is the only brand in the entire study with a perfect 100% SoV from both providers. Both AIs consider it mandatory to mention in every response.

Ozon and Wildberries get 100% from ChatGPT and 80% from YandexGPT. An interesting nuance: ChatGPT knows Russian marketplaces better than YandexGPT. This is an exception to the general rule, explained by Ozon and Wildberries being actively represented in international English-language media — financial reports, IPO history, mentions in Forbes and Bloomberg.

Avito shows the opposite: 43.3% from YandexGPT and just 16.7% from ChatGPT. Avito is a service with a strong local brand but weak international presence. ChatGPT doesn't perceive it as a first-tier marketplace.

AliExpress predictably gets more from ChatGPT (83.3% vs 56.7%) — an international brand well-represented in English-language sources.

E-commerce demonstrates: when a brand achieves a certain scale and media coverage, differences between AI providers smooth out. But for second-tier brands, the gap persists.

FinTech: the Sberbank paradox in YandexGPT

Fintech delivered the study's most unexpected insight: YandexGPT doesn't know Sberbank well. The country's largest bank, an ecosystem with dozens of products, a brand with near-100% recognition — yet only 36.7% SoV in YandexGPT.

Sberbank at 36.7% in YandexGPT, versus 56.7% in ChatGPT. The Russian AI knows Russia's largest bank worse than the American one. This is one of the study's most counterintuitive results.

Possible explanations:

1. The "obvious" effect: YandexGPT may consider Sberbank so obvious that it deliberately suggests lesser-known alternatives to be more helpful
2. Competitive context: the Yandex ecosystem competes with Sber across multiple verticals — from payments to cloud services
3. Prompt structure: for "which bank to choose" queries, AI more often suggests banks with distinct differentiators, while Sber is perceived as the "default" not needing recommendation

T-Bank is the only fintech brand with identical SoV from both providers: 63.3%. Notably, ChatGPT also cites T-Bank's domain in 26.7% of responses — the bank actively invests in content and English-language presence.

MTS Bank shows the classic "Russian brand without international presence" pattern: 50% from YandexGPT, just 13.3% from ChatGPT.

Yandex Bank — 36.7% from YandexGPT and 6.7% from ChatGPT. Even Yandex's own banking product doesn't get preferential treatment in YandexGPT — the AI doesn't favor it over competitors.

Uralsib is another brand invisible to ChatGPT: 0% SoV. For regional banks, ChatGPT is a closed channel.

Travel: blocked services and two different worlds

The travel niche reveals the direct impact of geopolitics on AI recommendations. YandexGPT completely ignores Western services unavailable in Russia, while ChatGPT ranks them at the top.

Booking, Google Flights, Kayak — 0% in YandexGPT. Three of the largest international services are completely absent from Alisa's responses. YandexGPT doesn't recommend services that have restricted operations in Russia or are unavailable to Russian users. This is a fundamental difference from ChatGPT, which recommends these services in 41-69% of responses without considering geographic availability.

For the user, this is critically important: ChatGPT might recommend booking a hotel through Booking, which doesn't work with Russian cards. YandexGPT won't do that.

Aviasales is the most interesting brand in the niche. ChatGPT knows it better than YandexGPT: 89.7% versus 66.7%. Aviasales is a Russian service with a strong international brand, an English-language version, and presence in international rankings. This proves: international presence works for ChatGPT visibility even for Russian companies.

Yandex Travel leads in YandexGPT (76.7%) but also scores a high 65.5% in ChatGPT. The brand is large enough for ChatGPT to know it, but in YandexGPT it gets a predictable ecosystem bonus.

Onetwotrip follows the typical pattern: 53.3% from YandexGPT, 27.6% from ChatGPT. A Russian service with limited international presence.

Hosting: two completely different worlds

Hosting is the niche with the most invisible brands. The hosting market is geographically divided: Russian providers dominate in one AI, international ones in the other. There's almost no overlap.

AdminVPS — 76.7% from YandexGPT, 0% from ChatGPT. The largest absolute gap among all brands in the study. For AdminVPS, ChatGPT doesn't exist as a channel. All AI traffic comes exclusively through YandexGPT.

Cloud.ru — the study's anomaly. A Russian cloud provider gets 23.3% from YandexGPT and 80% from ChatGPT. An inversion of the typical pattern. The likely reason is the English-language brand name. ChatGPT may perceive Cloud.ru as an international service. This is a valuable case: an English-language brand name for a Russian company can significantly boost ChatGPT visibility.

Timeweb and Beget are Russian hosting leaders in YandexGPT (83.3% and 76.7%), but ChatGPT knows them much worse (43.3% and 23.3%). The Russian hosting market is practically invisible to ChatGPT.

DigitalOcean and AWS — 0% in YandexGPT. International cloud providers are completely absent from Alisa's responses. ChatGPT recommends them in 56.7% of responses. For a Russian user asking Alisa about hosting, AWS doesn't exist.

Selectel is an interesting case: 10% from YandexGPT and 43.3% from ChatGPT. Selectel actively positions itself as an enterprise solution with English-language documentation and international presence, explaining its higher SoV in ChatGPT.

Summary table: most significant discrepancies

The average SoV gap between YandexGPT and ChatGPT across all brands in the study is 28 percentage points. This isn't statistical noise — these are two fundamentally different information spaces.

Three patterns that explain the discrepancies

1. The data language barrier

YandexGPT is trained on a Russian-language corpus and relies on Russian-language sources: reviews, rankings, articles, forums. ChatGPT on an English-language corpus with added multilingual data. A brand actively represented in the Russian-language internet but absent from the English-language web will be visible in YandexGPT and invisible in ChatGPT.

Example: AdminVPS actively works with Russian-language reviews and forums — 76.7% in YandexGPT. But has no English-language mentions — 0% in ChatGPT.

2. The geopolitical filter

YandexGPT doesn't recommend services that are unavailable or restricted for Russian users. Booking, Google Flights, Kayak — 0% in YandexGPT. ChatGPT doesn't apply such a filter and recommends them freely.

This isn't a bug but a feature: YandexGPT provides practically applicable recommendations for the Russian context. ChatGPT provides globally relevant but not always Russia-applicable ones.

3. The English-language brand effect

Brands with English-language names and international presence get a boost in ChatGPT, even if they're Russian. Cloud.ru — 80% in ChatGPT with 23.3% in YandexGPT. Aviasales — 89.7% in ChatGPT with 66.7% in YandexGPT. Selectel — 43.3% in ChatGPT with 10% in YandexGPT.

English-language naming and international positioning are a lever for boosting ChatGPT visibility.

What this means for GEO strategy

The study's main conclusion: optimizing for only one AI provider leaves up to 50% of AI traffic on the table. YandexGPT and ChatGPT form two parallel channels, each requiring its own approach.

For brands strong in YandexGPT but weak in ChatGPT

This includes most Russian brands: EdTech platforms, hosting providers, regional banks, travel services.

What to do:

• Create English-language content. Reviews, case studies, comparisons in English that ChatGPT can use as a source
• Work with international rankings. Mentions on G2, Trustpilot, Product Hunt, TechCrunch increase the likelihood of appearing in ChatGPT
• Optimize English-language Wikipedia. ChatGPT actively uses Wikipedia as a factual source about brands
• Publish on English-language platforms. Medium, Dev.to, HackerNoon — content from these enters ChatGPT's training data

For brands strong in ChatGPT but weak in YandexGPT

International brands operating in the Russian market: Coursera, DigitalOcean, Hetzner.

What to do:

• Build Russian-language presence. Articles on Habr, vc.ru, reviews in Russian-language blogs
• Work with Russian-language reviews. Reviews on Otzovik, iRecommend, specialized forums
• Localize content. Not just translation, but adaptation for the Russian context — Russian client case studies, comparisons with local competitors
• Collaborate with Russian-language bloggers. Reviews and mentions in content indexed by YandexGPT

For all brands

• Monitor both providers. Without data, you can't understand where you're strong and where you're invisible. GEO Scout monitors YandexGPT, ChatGPT, and 7 more providers daily
• Track domain citation. Brand mention is good. Domain citation is conversion. Skypro gets domain citation of 40% in YandexGPT — that's real traffic
• Test hypotheses. Cloud.ru showed that English-language naming boosts ChatGPT visibility. Aviasales — that international presence works. Test which channels influence each AI provider

Recommendations by niche

EdTech

Your audience in Russia uses both Alisa and ChatGPT. In YandexGPT, Russian EdTech brands get 30-97% SoV — the channel works. ChatGPT is the weak link: even leaders get under 54%. Priority: English-language graduate case studies, publications in international education media, presence on Coursera/edX as a partner (for cross-visibility).

E-commerce

The top 3 marketplaces are covered by both providers. If you're not Ozon, Wildberries, or Yandex.Market — work on presence in both channels. Avito, for example, loses 26.6 pp in ChatGPT. For niche marketplaces and D2C brands, GEO optimization for both providers gives a competitive edge.

FinTech

The Sberbank paradox shows: even the largest brands can underperform in AI. Don't rely on offline recognition — AI forms recommendations by different criteria. Monitor positions, work on content. T-Bank is an example of balance: 63.3% from both providers and domain citation of 26.7% in ChatGPT.

Travel

YandexGPT won't recommend Booking — that's a fact. If you're a Russian service, your competitor in YandexGPT is other Russian services, not international ones. In ChatGPT, it's the opposite: you compete with Google Flights (69%) and Kayak (55.2%). Two different competitive environments, two different strategies.

Hosting

The most polarized niche. Russian hosting is invisible in ChatGPT, international hosting in YandexGPT. If you're a Russian provider, ChatGPT is territory to conquer. The Cloud.ru case (80% in ChatGPT) shows: English-language positioning and international communication break the barrier.

Conclusion

YandexGPT and ChatGPT aren't competitors — they're two parallel information worlds. YandexGPT lives in the Russian-language internet and accounts for Russian specifics: from service availability to local rankings. ChatGPT lives in the English-language internet and evaluates brands on an international scale.

For business, this means one thing: GEO strategy must be bilingual. Optimizing only for YandexGPT leaves the ChatGPT audience to competitors. Optimizing only for ChatGPT loses traffic from Alisa, which tens of millions of Russians use daily.

Start with diagnostics: check how both AI providers see your brand right now. Current data across all 9 providers is available at GEO Scout, and niche summary rankings on the ratings page.

Data updates daily. In a week, the landscape may change — AI learns fast. The question is whether it's learning about your brand. The Command Center at GEO Scout automatically analyzes differences between YandexGPT, ChatGPT, and other providers — and builds a prioritized action plan for each of them.