March 23, 202615 min read

Blocked Services in AI: How Booking, Airbnb, and Others Lose Visibility in Neural Networks

Booking receives 0% recommendations from all AI providers, YandexGPT completely blocks services restricted in Russia, and Aviasales captures 36.6% of recommendations. GEO Scout data breakdown: how blockings transform AI visibility in the travel niche.

Vladislav Puchkov

Founder of GEO Scout, GEO optimization expert

When AI answers the question "which service is best for hotel booking," it doesn't just search for information. It evaluates the practical usefulness of the answer for the specific user. And if a service is blocked in the user's country — AI knows this.

At GEO Scout, we daily monitor brand visibility across 9 AI providers. Analyzing travel niche data, we discovered a pattern that changes the rules of the game: service blocking doesn't just reduce its visibility in neural networks — it completely zeros out recommendations and redistributes traffic to alternatives.

Here's a detailed breakdown with numbers.

Which services are blocked and how it affects AI

In recent years, several major international services have been blocked in Russia: Booking.com, Airbnb, and various Google services. For traditional SEO, this meant reduced organic traffic. But for AI visibility, the consequences turned out to be different — and significantly more serious.

More on the metrics we use: what is brand AI visibility and Share of Voice in AI.

When a user asks a neural network about hotel booking, AI must decide: should it mention the blocked service? Should it recommend it? Different providers answer this question differently.

Key observation: all AI providers separate two actions — mentioning and recommending. Mentioning a blocked service is possible as a fact ("Booking.com is one of the largest booking services, but it's blocked in Russia"). Recommending means suggesting the user actually use it. And no AI provider wants to recommend blocked services.

Booking.com: known by all, recommended by none

Booking.com is the world's largest hotel booking service. Billions in revenue, hundreds of millions of users. And zero recommendations in AI.

Overall metrics

A third of AI responses to travel queries mention Booking. But how exactly? Positive rate is only 8.86% — one of the lowest in the niche. Meanwhile, negative rate is 5.06% — AI actively criticizes and warns about access issues.

For comparison: Aviasales has a positive rate of 71.43%, Yandex Travel — 84.15%. The difference is orders of magnitude.

Breakdown by AI provider

YandexGPT — absolute zero. Yandex has completely excluded Booking from responses. Not a single mention. As if the service doesn't exist. This is a deliberate policy: YandexGPT doesn't recommend what the user can't use.

Claude — the most informed. 51.7% mention rate — more than half of responses contain a Booking mention. But these are purely informational: "Booking.com is a major service, however it's blocked in Russia, consider alternatives." Recommendation — zero.

ChatGPT knows but doesn't advise. 41.4% — ChatGPT mentions Booking in the context of the global market, but when answering for a Russian-language audience, immediately adds a caveat about the blocking. Recommendation rate — 0%.

A detailed ranking of all travel brands in our research.

Airbnb: balanced on the edge of negativity

Airbnb is the second major casualty of blockings in the travel niche. But its situation differs from Booking's.

Overall metrics

Note the sentiment: positive and negative rates are exactly equal — 12.24%. This is a unique situation. Most brands have sentiment skewed one way. Airbnb has a perfect balance between "great service" and "but you can't use it."

This means AI doesn't know how to feel about Airbnb. On one hand, training data is full of positive reviews. On the other, the blocking makes those reviews irrelevant for Russian users. The result is cognitive dissonance in numbers.

Breakdown by AI provider

YandexGPT again delivers zero. The same pattern as with Booking: complete exclusion.

Gemini knows better than ChatGPT. 36.7% versus 20.7%. Gemini is likely trained on more recent data where Airbnb is still actively mentioned in the context of the global vacation rental market.

ChatGPT is restrained. Just 20.7% — one in five responses. For a brand of Airbnb's scale, this is catastrophically low. Before the blocking, Airbnb would have been in the top three most-mentioned services.

Google Flights: the product ecosystem paradox

Google Flights is an interesting case. It's a Google product, and it has specific behavior in AI responses.

Overall metrics

Breakdown by AI provider

Gemini promotes its own ecosystem product. 73.3% — nearly three-quarters of responses. Gemini is a Google product, Google Flights is a Google product. Coincidence? Not exactly. This demonstrates an important principle: AI providers have built-in preferences toward their own products and services.

ChatGPT also knows it well. 69% — one of the highest mention rates for any travel service in ChatGPT. The English-language training base plays a key role: Google Flights is a standard tool for American and European users.

YandexGPT — predictable zero. Google Flights doesn't work in Russia, YandexGPT doesn't mention it. The logic is airtight.

More on differences between AI providers in the article ChatGPT vs Claude vs Gemini: who they recommend.

Kayak: invisible to most neural networks

Kayak is a flight metasearch engine with global presence. But in AI responses for the Russian audience, it has virtually disappeared.

Overall metrics

Breakdown by AI provider

Kayak is perhaps the most illustrative example of selective invisibility. Three AI providers don't know about it at all: YandexGPT, Gemini, and Google AI Mode. Yet ChatGPT mentions it in 55.2% of responses.

Why such variation? ChatGPT is trained on English-language data where Kayak is a standard tool. Gemini and Google AI Mode prefer their own Google Flights. YandexGPT prefers Russian services. The result: Kayak exists only in ChatGPT's reality.

More on how brands become invisible to neural networks in the research AI blind spots.

Five patterns: how AI reacts to blockings

Data from four blocked services reveals consistent neural network behavior patterns.

Pattern 1: YandexGPT — total filtering

YandexGPT consistently delivers 0% for all blocked services: Booking, Airbnb, Google Flights, Kayak. These aren't random zeros — this is systemic policy. YandexGPT targets Russian users and excludes services that are physically unavailable from responses.

For brands, this means: if your service is blocked in Russia, for YandexGPT's audience (tens of millions of Yandex with Alisa users), you don't exist.

Pattern 2: mention without recommendation

0% recommendation rate across all providers for all blocked services. This is the most important finding. Even ChatGPT, which mentions Booking in 41.4% of responses and Google Flights in 69%, never recommends them for the Russian audience.

AI has learned to separate knowledge from recommendation. "I know Booking exists" and "I recommend Booking to you" are two fundamentally different actions. With blocking, the second is completely zeroed out.

Pattern 3: rising negativity

Negative rates for blocked services are anomalously high. Booking: 5.06% — AI warns about unavailability. Airbnb: 12.24% — equals its positive rate. Neural networks don't just ignore — they actively inform users about problems. Typical response: "Booking.com is the largest hotel aggregator, but it's blocked in Russia. We recommend looking at..."

This creates a counter-advertising effect: the user not only doesn't receive a recommendation but also learns about problems with the service.

Pattern 4: ChatGPT — a window to the global world

ChatGPT consistently shows the highest mention rates for blocked services: Booking 41.4%, Google Flights 69%, Kayak 55.2%. The reason is the English-language training base. ChatGPT is trained on terabytes of English text where these brands dominate.

For international brands, this is partial consolation: at least one AI provider knows them. But without recommendations, mention rate doesn't convert to users.

Pattern 5: redistribution to alternatives

The most strategically important pattern. Recommendations that could have gone to blocked services don't disappear — they're redistributed. Aviasales receives 36.6% recommendation rate. Yandex Travel — 77.87% mention rate and first place in mentions. Competitor blocking is a direct bonus for available alternatives.

Who benefits from blockings: the beneficiary map

Each blocked service leaves a "visibility vacuum," and this vacuum is filled by specific brands.

Aviasales: the main beneficiary

Aviasales captures recommendations that, without blockings, would be distributed among Booking, Skyscanner, and Kayak. A 36.6% recommendation rate is the absolute record in the travel niche. Grok recommends Aviasales in 70% of responses.

Yandex Travel: the new #1

With a mention rate of 77.87%, Yandex Travel holds first place instead of Booking. Under normal circumstances, Booking with its global dominance would be in the top three. Now its place belongs to Yandex.

Claude's data is particularly telling: domain citation 62.1% for Yandex Travel. Claude links to travel.yandex.ru as an authoritative source in the majority of travel responses.

Ostrovok and Sutochno: growth in the accommodation niche

Airbnb left a vacuum in the short-term rental niche. Ostrovok (hotels) and Sutochno.ru (apartments by the day) are actively filling it. AI increasingly mentions these services as alternatives when users ask about accommodation booking in Russia.

Onetwotrip: capturing aggregator traffic

Onetwotrip positions itself as a universal aggregator — flights, hotels, rail. Without Booking and Kayak, it captures some of the queries that international aggregators previously handled.

What this means for the travel market

Blockings have created a unique situation: the AI market in Russia develops by different rules than the global one.

Two parallel worlds

In the global AI space (ChatGPT, Claude, Gemini), Booking remains a major player. In the Russian AI space (YandexGPT, and increasingly, Russian-language responses from other providers), it doesn't exist. This creates two parallel rankings with minimal overlap.

The window of opportunity is narrowing

Currently, Russian alternatives capture the vacuum "by default" — simply because competitors are blocked. But this window isn't permanent. As the AI market grows and new players emerge, competition for the freed positions will intensify. Brands investing in GEO optimization now will secure their positions. The rest will lose even what they received for free.

A precedent for other niches

The patterns we see in travel are beginning to appear in other niches. Any blocked service — whether a social network, cloud platform, or financial tool — risks facing the same set of problems: zero recommendations, rising negativity, traffic redistribution to alternatives.

GEO strategy for alternatives: how to lock in the advantage

If you're a Russian service benefiting from competitor blockings, here's a strategy for securing your positions.

1. Build recommendation rate, not just mention rate

Mention rate grows "on its own" — AI mentions you because competitors are blocked. But recommendation rate is a deliberate neural network choice. For AI to recommend you, you need to:

• Create expert content (comparisons, guides, reviews)
• Earn independent reviews on platforms that AI cites
• Strengthen E-E-A-T signals (expertise, experience, authoritativeness, trust)

More in the article how to get into neural network recommendations.

2. Work with each AI provider separately

The data shows: each provider is a separate ecosystem. YandexGPT completely ignores blocked brands, ChatGPT remembers them. Strategy for each provider should differ:

• YandexGPT: you're already winning here by default, but you need to reinforce positions with Russian-language content
• ChatGPT/Claude: blocked competitors are still mentioned here, you need to push them out of mention rate
• Gemini: Google promotes Google Flights, but there's room for hotels and recreation

On provider differences in the article Yandex Alisa vs ChatGPT: the difference in recommendations.

3. Increase domain citation

Domain citation — the percentage of responses where AI links to your website as a source. This is the strongest GEO signal. Aviasales has domain citation of 28.94%, significantly higher than most competitors.

How to increase domain citation:

• Publish unique data (price indexes, destination statistics)
• Create content that AI will want to cite: analytical, not promotional
• Use structured data on your website

More on citation in domain citation rate: who AI cites directly.

4. Monitor changes daily

The AI visibility landscape shifts quickly. Blockings can be lifted, competitors may find workarounds, and AI models are updated and retrained. Daily monitoring through GEO Scout lets you track any changes across all 9 AI providers.

5. Prepare for competitors' return

If blockings are lifted, Booking and Airbnb will return to AI responses with accumulated global authority. Brands that by then have secured high recommendation rate and domain citation will hold their positions. Those who relied only on competitor absence will lose everything.

Summary table: blocked services in AI

Overall pattern: YandexGPT — 0% across all. ChatGPT — knows them best. Recommendation — 0% from all providers for all blocked services.

How to monitor the impact of blockings on your brand

If you operate in the travel niche — or any niche with blocked competitors — you need to understand how AI visibility is being redistributed.

GEO Scout daily monitors brand visibility across 9 AI providers: ChatGPT, Claude, DeepSeek, Gemini, Google AI Mode, Google AI Overview, Grok, Perplexity, and YandexGPT.

What you can do right now:

1. Check your brand — go to geoscout.pro and see how neural networks talk about you
2. Compare with competitors — including blocked ones, to understand what vacuum you can fill
3. View the ranking — an interactive dashboard with provider filters is available in the rating
4. Set up daily monitoring — the free plan includes 3 prompts across 3 AI providers

Key takeaways

Blocking = zero recommendations. No AI provider recommends blocked services. That's 0% recommendation rate across the board — from ChatGPT to YandexGPT. AI strives to give practically useful answers and won't suggest an unavailable service.

YandexGPT is the strictest filter. Complete exclusion of blocked brands from responses. For tens of millions of Yandex with Alisa users, these services don't exist.

ChatGPT remembers but doesn't recommend. The English-language base makes ChatGPT the most informed, but awareness without recommendation doesn't convert to users.

Alternatives get a bonus, but not forever. Aviasales (36.6% recommendations), Yandex Travel (77.87% mentions), Ostrovok, Sutochno — all benefit from blockings. But this is a window of opportunity that needs to be secured through GEO optimization.

Sentiment works against the blocked. AI doesn't just stay silent — it warns. Booking: positive 8.86%, negative 5.06%. Airbnb: positive = negative. This is active counter-advertising, not passive ignoring.

Monitoring is a necessity. The blocking situation is dynamic. Daily monitoring through GEO Scout lets you track changes and react in real time. The Command Center at GEO Scout automatically turns this data into a prioritized plan — how to leverage the vacuum from blocked competitors and what content to create to secure your positions.