Why does your company need a GEO report? Summary report based on analysis of 400+ Polish companies
Average GEO score: 40.6/100. 66% of companies below 50/100. 95% without llms.txt. 5 sectors: Telecommunications (39.5), E-commerce (42.3), Banking (47.1), Insurance (41.7), Media (49.5). No company reached 80/100.
Telecommunications — average 39.5/100
- T-Mobile.pl (64/100): 73 FAQ with JSON-LD, but no llms.txt and outdated sitemap
- Orange.pl (63/100): Sitemap from 2022 (4 years outdated), no Schema Product/Offer
- Plus.pl (Polkomtel) (49/100): WordPress with numerous plugins, no FAQ, no authors
- Play.pl (P4) (42/100): No Schema, no FAQ, minimal educational content
Scenario: User asks AI: "How much does an unlimited internet plan cost at Play?" — ChatGPT does NOT cite play.pl — because Play lacks Schema Product/Offer on its offer pages, content is CTA-driven ("Buy now!") instead of informational. AI cites comparison sites, telepolis.pl and Reddit. The user gets prices from 3 months ago.
None of the 4 operators reaches the "Good" threshold (60+). None formats content in optimal citable blocks. The telecom industry is massively invisible in AI search.
E-commerce — average 42.3/100
- Allegro (62/100): DataDome blocks AI crawlers on WAF, no llms.txt
- Ceneo (52/100): No FAQ JSON-LD, CSR hinders indexing
- Decathlon (52/100): No llms.txt, limited Schema on products
- Zalando (52/100): No dedicated AI directives in robots.txt
- x-kom (48/100): React CSR — site invisible to AI crawlers, Cloudflare blocks
- Answear (44/100): No AI rules in robots.txt, weak Schema
- Eobuwie (41/100): No structured data, minimal content
- Empik (38/100): No llms.txt, no FAQ, poor citability
- Media Expert (29/100): Critically low score — missing almost all GEO signals
Scenario: User asks AI: "What laptop for remote work under 4000 PLN do you recommend? Where to buy it?" — ChatGPT recommends laptop models, but when asked "where to buy" doesn't direct to x-kom or Media Expert — because their pages render client-side (CSR/React), and AI crawlers see an empty page. AI links to YouTube reviews and Ceneo.
The Allegro Paradox: Allegro launched Allegro GPT in ChatGPT (API integration), but simultaneously its WAF (DataDome) blocks OAI-SearchBot. Allegro works with OpenAI — and simultaneously blocks its crawlers.
Banking — average 47.1/100
- Santander (62/100): Leader — best Schema and FAQ, but no llms.txt
- mBank (54/100): Good content base, but no AI-specific structured data
- Pekao (52/100): No llms.txt, limited FAQ
- Bank Millennium (49/100): No llms.txt, weak financial product citability
- BNP Paribas (49/100): No llms.txt, no AI rules in robots.txt
- PKO BP (49/100): Largest bank in Poland, but weak AI visibility
- Credit Agricole (42/100): No structured data, minimal content
- Nest Bank (32/100): Critically weak AI visibility
- Alior Bank (28/100): Missing almost all GEO signals
Scenario: User asks AI: "Which bank has the best free personal account in 2026?" — ChatGPT provides general information about accounts, but doesn't cite current terms from banks — because sites lack Schema Product/Offer on banking products. AI relies on months-old comparison articles (bankier.pl, money.pl).
PKO BP — Poland's largest bank — scores 49/100. No bank has an llms.txt file. None presents financial product terms in an AI-readable format.
Insurance — average 41.7/100
- Allianz (59/100): Closest to "Good" threshold, but no llms.txt
- UNIQA (41/100): Weak citability, no FAQ JSON-LD
- Ergo Hestia (38/100): No structured data on products
- PZU (38/100): Largest insurer in Poland, but weak AI visibility
- Warta (33/100): No llms.txt, minimal Schema
Scenario: User asks AI: "What travel insurance do you recommend for a trip to Asia?" — ChatGPT mentions PZU, Warta and Allianz by name — because they are large and present on Wikipedia. But it doesn't provide offer details (coverage, price, exclusions), because insurer websites don't present this data in an AI-readable format. Instead, AI directs to rankomat.pl.
PZU — Poland's largest insurer — scores 38/100. AI "knows" the brand but can't cite a specific offer. The purchasing decision shifts to intermediaries.
Media — average 49.5/100
- Onet.pl (63/100): Leader — extensive structured data, SSR
- Benchmark.pl (61/100): Strong technical base, good E-E-A-T
- Medonet (61/100): Good medical Schema, authors with credentials
- TVN24 (58/100): Good base, but no llms.txt and AI files
- Wirtualna Polska (57/100): Solid base, but no AI rules in robots.txt
- Filmweb (52/100): Good Schema Movie/Review, but no llms.txt
- Pudelek (47/100): Weak citability, no structured data
- Gazeta.pl (42/100): No llms.txt, limited Schema Article
- 90minut.pl (12/100): Critically weak AI visibility
Scenario: User asks AI: "What happened in Poland today? Give me the top news." — ChatGPT cites Onet and TVN24 more often than other portals — because they have better structured data (Schema Article, NewsArticle with authors and dates). 90minut.pl with a score of 12/100 is practically invisible: sports AI search uses ESPN and Goal.com instead of Polish sources.
Media has the highest average (49.5/100). This is natural — they produce a lot of content. But the gap between the leader (Onet 63) and the bottom of the market (90minut 12) is enormous.
Why does your company need a GEO report?
Summary report based on analysis of 400+ Polish companies across 5 sectors. Hard data. Real problems. User scenarios.
Key finding: Out of 400+ analyzed Polish companies, none reached a score of 80/100 in the GEO audit. The average score is just 40.6/100. The Polish digital economy is massively unprepared for the AI search engine era.
GEO determines whether your brand is visible in responses from ChatGPT, Gemini, Perplexity, Copilot and Google AI Overviews. GEO today is what SEO was in 2010.
The weakest optimized sector in Polish digital economy
| Company | Score | Key issue |
|---|---|---|
| T-Mobile.pl | 64 | 73 FAQ with JSON-LD, but no llms.txt and outdated sitemap |
| Orange.pl | 63 | Sitemap from 2022 (4 years outdated), no Schema Product/Offer |
| Plus.pl (Polkomtel) | 49 | WordPress with numerous plugins, no FAQ, no authors |
| Play.pl (P4) | 42 | No Schema, no FAQ, minimal educational content |
"How much does an unlimited internet plan cost at Play?"
ChatGPT does NOT cite play.pl — because Play lacks Schema Product/Offer on its offer pages, content is CTA-driven ("Buy now!") instead of informational. AI cites comparison sites, telepolis.pl and Reddit. The user gets prices from 3 months ago.
Key finding: None of the 4 operators reaches the "Good" threshold (60+). None formats content in optimal citable blocks. The telecom industry is massively invisible in AI search.
Paradox: most online transactions, weak AI visibility
| Company | Score | Key issue |
|---|---|---|
| Allegro | 62 | DataDome blocks AI crawlers on WAF, no llms.txt |
| Ceneo | 52 | No FAQ JSON-LD, CSR hinders indexing |
| Decathlon | 52 | No llms.txt, limited Schema on products |
| Zalando | 52 | No dedicated AI directives in robots.txt |
| x-kom | 48 | React CSR — site invisible to AI crawlers, Cloudflare blocks |
| Answear | 44 | No AI rules in robots.txt, weak Schema |
| Eobuwie | 41 | No structured data, minimal content |
| Empik | 38 | No llms.txt, no FAQ, poor citability |
| Media Expert | 29 | Critically low score — missing almost all GEO signals |
"What laptop for remote work under 4000 PLN do you recommend? Where to buy it?"
ChatGPT recommends laptop models, but when asked "where to buy" doesn't direct to x-kom or Media Expert — because their pages render client-side (CSR/React), and AI crawlers see an empty page. AI links to YouTube reviews and Ceneo.
Key finding: The Allegro Paradox: Allegro launched Allegro GPT in ChatGPT (API integration), but simultaneously its WAF (DataDome) blocks OAI-SearchBot. Allegro works with OpenAI — and simultaneously blocks its crawlers.
Best in the comparison, but far from optimal
| Company | Score | Key issue |
|---|---|---|
| Santander | 62 | Leader — best Schema and FAQ, but no llms.txt |
| mBank | 54 | Good content base, but no AI-specific structured data |
| Pekao | 52 | No llms.txt, limited FAQ |
| Bank Millennium | 49 | No llms.txt, weak financial product citability |
| BNP Paribas | 49 | No llms.txt, no AI rules in robots.txt |
| PKO BP | 49 | Largest bank in Poland, but weak AI visibility |
| Credit Agricole | 42 | No structured data, minimal content |
| Nest Bank | 32 | Critically weak AI visibility |
| Alior Bank | 28 | Missing almost all GEO signals |
"Which bank has the best free personal account in 2026?"
ChatGPT provides general information about accounts, but doesn't cite current terms from banks — because sites lack Schema Product/Offer on banking products. AI relies on months-old comparison articles (bankier.pl, money.pl).
Key finding: PKO BP — Poland's largest bank — scores 49/100. No bank has an llms.txt file. None presents financial product terms in an AI-readable format.
A public trust industry invisible to AI
| Company | Score | Key issue |
|---|---|---|
| Allianz | 59 | Closest to "Good" threshold, but no llms.txt |
| UNIQA | 41 | Weak citability, no FAQ JSON-LD |
| Ergo Hestia | 38 | No structured data on products |
| PZU | 38 | Largest insurer in Poland, but weak AI visibility |
| Warta | 33 | No llms.txt, minimal Schema |
"What travel insurance do you recommend for a trip to Asia?"
ChatGPT mentions PZU, Warta and Allianz by name — because they are large and present on Wikipedia. But it doesn't provide offer details (coverage, price, exclusions), because insurer websites don't present this data in an AI-readable format. Instead, AI directs to rankomat.pl.
Key finding: PZU — Poland's largest insurer — scores 38/100. AI "knows" the brand but can't cite a specific offer. The purchasing decision shifts to intermediaries.
Closest to the threshold, but still far from ideal
| Company | Score | Key issue |
|---|---|---|
| Onet.pl | 63 | Leader — extensive structured data, SSR |
| Benchmark.pl | 61 | Strong technical base, good E-E-A-T |
| Medonet | 61 | Good medical Schema, authors with credentials |
| TVN24 | 58 | Good base, but no llms.txt and AI files |
| Wirtualna Polska | 57 | Solid base, but no AI rules in robots.txt |
| Filmweb | 52 | Good Schema Movie/Review, but no llms.txt |
| Pudelek | 47 | Weak citability, no structured data |
| Gazeta.pl | 42 | No llms.txt, limited Schema Article |
| 90minut.pl | 12 | Critically weak AI visibility |
"What happened in Poland today? Give me the top news."
ChatGPT cites Onet and TVN24 more often than other portals — because they have better structured data (Schema Article, NewsArticle with authors and dates). 90minut.pl with a score of 12/100 is practically invisible: sports AI search uses ESPN and Goal.com instead of Polish sources.
Key finding: Media has the highest average (49.5/100). This is natural — they produce a lot of content. But the gap between the leader (Onet 63) and the bottom of the market (90minut 12) is enormous.
| Sector | Average | Leader | Weakest | llms.txt |
|---|---|---|---|---|
| Media | 49.5 | Onet (63) | 90minut (12) | 0% |
| Banking | 47.1 | Santander (62) | Alior (28) | 0% |
| E-commerce | 42.3 | Allegro (62) | Media Expert (29) | 0% |
| Insurance | 41.7 | Allianz (59) | Warta (33) | 0% |
| Telecom | 39.5 | T-Mobile (64) | Play (42) | 0% |
- -95% of companies lack llms.txt — the basic instruction file for AI. It's like having no sitemap in 2005.
- -No Schema Product/Service/Offer — AI can't compare offers, prices or terms.
- -Sales content instead of informational — AI cites facts, not CTAs. Pages with "Buy now!" are skipped.
- -WAF and CSR block crawlers — DataDome, Cloudflare and React CSR make sites invisible.
- -No authors and E-E-A-T — AI prefers sources with experts. Anonymous content is skipped.
Find out what AI says about your company
The AIVS™ report will show you exactly where you stand — and what to fix so AI search engines start recommending your company.
4,900 PLN net · one-time · delivered in 7 days