Over the last year, our SEO team has recorded a trend that can no longer be ignored. Informational queries that consistently drove traffic are now declining. CTR (the percentage of people clicking links in search results) is dropping even while rankings remain stable.
The cause is Google AI Overviews. This generated answer appears directly on the search results page, above any organic results. Users get the answer and don’t click through to the website. According to research, AI Overviews reduce CTR by 15-46% depending on the query type.
In Ukraine, AI Overviews are being tested since 2025. Those who adapt now will retain traffic. Those who wait will lose the rankings they built over years.
Contents
What is GEO and how it differs from SEO
GEO (Generative Engine Optimization) is the optimization of content for AI systems that generate answers to user queries. Google AI Overviews, ChatGPT Search, Perplexity, and Copilot all read your site, extract information, and present it in their own format.
The fundamental difference from classic SEO. In SEO, the goal is to reach the first page of results and get a click. In GEO, the goal is to become a source that AI cites. A user may never visit your site but will see your expertise in an AI-generated answer with a link.
This is not a replacement for SEO. It is an addition. Classic SEO works for commercial queries like “buy tires Kyiv” or “price of air suspension repair” – AI Overviews appear in only 4% of e-commerce queries. However, B2B technologies already receive AI answers for 70% of queries. Informational articles, guides, and reference materials – this is where generative engine optimization is critical.
AI Overviews in Ukraine: What is happening right now
Google began testing AI Overviews in Ukraine in 2025. Currently, not for all queries or all users. But the trend is clear – coverage is growing monthly.
Simultaneously, the share of people searching for information via ChatGPT and Perplexity is rising. Not instead of Google, but in addition. For complex questions, comparisons, and choosing between options, AI chats are becoming the first choice.
Numbers you should know. AI Overviews reduce organic CTR by 15-46% depending on the query type. Informational queries suffer the most – “how to set up,” “what is,” “difference between.” This is exactly the content many companies create to attract top-of-funnel traffic.
For our SEO clients, we already see this in analytics. Rankings are the same, impressions are growing – but clicks are falling. A classic symptom: Google shows your information in an AI Overview, but the user does not visit the site.
“We manage SEO for dozens of sites and see a clear picture: informational traffic is declining. Not due to falling rankings, but because Google answers on your behalf. The only way to stay in the game is to become the source that AI cites.”
– Volodymyr Kashalaba, CEO Guild of Marketing
Which content do AI engines cite and which do they ignore
AI systems do not cite everything. They select sources based on specific criteria. Understanding these criteria is the foundation of a generative engine optimization strategy.
What increases the chances of being cited
Original data and research
Statistics from your practice, survey results, comparative tables – things not found in other sources. AI systems value unique information because it makes their answer useful for the user.
Content with expert names
Studies show: materials with quotes from named specialists receive AI citations 2-3 times more often than anonymous content. Authorship, position, company – all these are signals of authority for AI algorithms (in Google, this is called E-E-A-T – Experience, Expertise, Authoritativeness, Trustworthiness).
Clear structure with direct answers
AI engines extract specific fragments: definitions, step-by-step instructions, comparisons. If your article starts with five paragraphs of fluff before the answer – AI will find another source.
What AI ignores
Generic content without specifics
“SEO is important for business” – there are thousands of such texts; AI sees no reason to cite yours specifically.
Text oversaturated with keywords
AI models evaluate semantics, not word matches. Keyword stuffing not only fails to help but reduces the chances of being cited.
Outdated content
AI systems prefer fresh materials. An article about “SEO trends 2023” in 2026 is dead weight.
How to optimize your site for generative search
Generative engine optimization is a set of specific actions. Not an abstract strategy, but changes in content, structure, and the technical part of the site.
Moving from keywords to topical coverage
In classic SEO, you target specific queries: “buy tires Kyiv,” “Performance Max price.” In generative engine optimization, a different logic applies. AI systems evaluate not individual pages, but the site’s topical authority.
What this means in practice. Instead of a single article “What is GEO,” you need a topical cluster: an article on GEO, an article on AI Overviews, a guide on optimizing schema markup, a case study on content adaptation. Each page is interlinked with others. AI sees: this site covers the topic deeply – therefore, it can be trusted as a source.
For our clients, for whom we do SEO promotion, we are already restructuring content strategies based on this principle. Not individual articles for queries, but connected clusters by topic.
Answers to specific questions
AI extracts direct answers. The most effective format is a question in a subheading, with the answer in the first sentence below it. A detailed explanation follows.
Look at the “People Also Ask” (PAA) block in Google. Each question from PAA is a potential fragment for an AI Overview. Your task is to provide a better answer than anyone else.
Original data and expert content
AI systems need facts not found in other sources. Your internal statistics, A/B test results, practice comparisons – this is what makes content citable.
At Guild of Marketing, we have data from hundreds of ad accounts over 13+ years. When writing about Performance Max or local SEO, we cite real numbers from our practice. It is exactly this type of content that AI algorithms highlight among thousands of identical “review” articles.
Technical settings: schema, llms.txt, robots.txt
Besides content, there is a technical component. It is not complex, but without it, AI engines may simply not see your site.
Schema markup (structured data)
Schema markup is code that explains to search engines and AI engines what is on the page. Not for humans – for algorithms.
Five types of schema critical for generative engine optimization:
Article - for articles and blog posts
Indicates author, publication date, and update date. AI sees: the article is fresh, the author is real.
FAQ - for Q&A blocks.
One of the most common sources for AI Overviews.
HowTo - for step-by-step instructions
Ideal for guides formatted as “how to set up,” “how to do.”
Organization - for the homepage
Name, contacts, social media, area of activity. AI systems use this data to verify the source.
Breadcrumb - navigation breadcrumbs
Helps AI understand the site hierarchy and the context of each page.
llms.txt file
A new practice gaining momentum. The llms.txt file is placed in the site root (similar to robots.txt) and describes your site for AI crawlers (robots that collect information).
What to write in llms.txt: company name, area of activity, main services or products, key expertise, links to the most important pages. Essentially, a brief for the AI system: “this is who we are, this is our expertise, this is where to find our best information.”
The llms.txt standard is not yet finalized but is already supported by several AI crawlers. Implementation takes 15 minutes – and puts you ahead of 99% of Ukrainian sites that don’t know about it.
robots.txt: Do not block AI bots
Check your robots.txt today. Some CMS (Content Management Systems) by default block unknown bots. If GPTBot (OpenAI), ClaudeBot (Anthropic), or PerplexityBot are blocked – your content will not appear in AI answers. Period.
This does not mean “open everything.” Pages with personal data, admin panels, technical URLs – block them as usual. But content pages – articles, services, case studies – must be accessible to AI crawlers.
“The technical part of generative engine optimization is not rocket science. Schema markup, llms.txt, checking robots.txt – this is a few hours of work. But those few hours determine whether AI systems see you or if you simply do not exist for them.”
– Volodymyr Kashalaba, CEO Guild of Marketing
Checklist: What to do this month
A practical action plan. Not theory – specific steps that can be completed in 2-4 weeks without a large budget.
Audit current traffic
Compare CTR for the last 6 months with the previous 6 months. If rankings are stable but clicks are falling – AI Overviews are already affecting you. Informational queries will decline the most: “how,” “what is,” “why.”
Check robots.txt
Search for mentions of GPTBot, ClaudeBot, PerplexityBot. If there is a Disallow rule – fix it. 5 minutes of work.
Implement schema markup
Start with Article and FAQ for blog posts, Organization for the homepage. Validate via Google Rich Results Test.
Create llms.txt
A brief description of the company and expertise for AI crawlers. Place in the site root next to robots.txt.
Audit content for "citability"
Review your top 20 articles. Do they have direct answers to questions? Do they have original data? Is the author indicated? Update anything that does not meet these criteria.
Restructure content strategy
From individual articles for keywords to topical clusters. Each topic is covered by 3-5 interconnected materials with internal linking.
Monitoring
Track if your site appears in AI Overviews. Tools: Semrush, Ahrefs, or manual checking of key queries in incognito mode.
FAQ
-
Will generative engine optimization replace classic SEO?
No. GEO complements SEO, it does not replace it. Commercial queries ("buy," "price," "order") continue to work under classic rules - AI Overviews appear in only 4% of e-commerce queries. Generative engine optimization is critical for informational queries where AI answers are already eating traffic.
-
Are AI Overviews already working in Ukraine?
Yes, Google is testing AI Overviews in Ukraine since 2025. Coverage is growing monthly. Simultaneously, the use of ChatGPT and Perplexity for information search is rising. It is better to adapt now while competitors have not started.
-
How much does generative engine optimization cost?
Basic technical steps (schema markup, llms.txt, robots.txt) - a few hours of developer work. Strategic content restructuring depends on the site volume. Leave a request for an SEO audit and we will evaluate specifically for your project.
-
Which sectors will suffer most from AI search?
B2B technologies already have AI Overviews for 70% of queries. Medical, legal, and financial informational queries are next in line. E-commerce is currently the least vulnerable - only 4% of queries have AI answers.
-
What is llms.txt and is it mandatory to create it?
llms.txt is a file in the site root that describes your business and expertise for AI crawlers. An analog of robots.txt, but for artificial intelligence. Not mandatory, but gives an advantage - 99% of Ukrainian sites do not have it. Implementation takes 15 minutes.
AI search will not wait for you to be “ready.”Free SEO audit with analysis of AI Overviews’ impact on your traffic. 30 minutes. We will show where you are losing clicks and how to fix it.
Guild of Marketing has been doing SEO promotion for businesses across Ukraine. 13+ years of experience. 300+ ad accounts, a comprehensive approach: PPC, SEO, SMM, AI integration. We are already adapting clients’ content strategies for generative search.
More about SEO promotion:
guildofmarketing.ua/en/seo-promotion/
Free
30-minute consultation
We will analyze your funnel and show you areas for growth.
Free • Specific recommendations • Quick start
