How artificial intelligence is changing user behaviour in search engines

AI Search is not just another small change in the way search engines work. It's a behavioural change. For an e-commerce owner, it means that a customer no longer just types in «men's sneakers black» or «best moisturizer for dry skin». He asks in natural language, compares options, asks for personalized suggestions, waits for a short summary and, in many cases, gets an answer without visiting any website. HubSpot's article on AI search behavior highlights exactly this shift: users aren't abandoning search, but are reframing it around conversational search, speed, trust and immediate utility. For brands, it's no longer just about «getting on Google first». It's to become the source worth mentioning, summarizing and recommending from AI-powered results, AI Overviews, ChatGPT Search and other discovery interfaces. See also: Digital Marketing & SEO, business automation & AI, website construction, e-shop construction.

AI Search: what changes in search behaviour

In traditional search, the user would enter a short phrase, scan ten organic results, open multiple tabs and compile the answer themselves. With AI Search, the process is condensed. The user describes the problem with more context, such as «I want shoes for walking in the city, comfortable, under 120 euros and to match smart casual outfits.» This change shifts the value from simple keyword matching to search intent, semantic understanding and quality of response. AI search favours content that answers real questions, explains trade-offs, provides examples, compares options and demonstrates experience. Simply put, SEO is becoming less mechanical and more strategic: you need to design content that helps humans while being clear enough for engines to understand.

For e-commerce, the change is even more critical because it affects the entire customer journey. Product discovery, product discovery, is no longer limited to an eshop's categories or sponsored results. It can start in an AI assistant, continue in a Google AI Overview and end in a marketplace, a comparison site or directly to the brand. If your content doesn't offer clear answers to price, usage, materials, dimensions, comparisons, reviews, after-sales and frequently asked questions, then the AI model has less reason to include you in the synthetic response. Conversely, a brand with strong brand authority, documented product pages, purchase guides, structured data and a consistent digital presence has a much better chance of showing up in the new environment.

Data shows fewer clicks and more responses on the page

Clicks are decreasing, but the influence of the brand does not disappear

The most practical consequence of AI Search is that it increases the likelihood that the user will get the information within the search environment itself. This doesn't mean that websites stop being valuable. But it does mean that the click is no longer a given, even when a brand influences the purchase decision. According to a Pew Research Center analysis of user behavior on Google, when an AI summary was displayed, users clicked on a traditional result at a rate of 8%, while when no AI summary was displayed the corresponding rate was 15%. Meanwhile, users terminated the browsing session at a rate of 26% when AI summary was present, compared to 16% when it was not. This is a strong signal for e-commerce teams: measuring success cannot be limited to organic traffic alone, but must take into account branded search lift, assisted conversions, visibility into AI responses and content quality.

As shown in the graph below, the presence of AI summary substantially alters user behavior after search.

The proliferation of AI Overviews also shows that change is not theoretical. Semrush recorded that AI Overviews appeared in 6,49% of desktop Google searches in January 2025 and 13,14% in March 2025. This rise, in such a short period of time, explains why many businesses are seeing fluctuations in organic clicks even when their rankings haven't crashed. For an eshop, especially in categories with intense pre-purchase research, such as technology, cosmetics, supplements, fashion, home goods or B2B products, the new reality requires a different content design: more documentation, better structure, cleaner answers and a stronger E-E-A-T.

The graph below illustrates the increase in the occurrence of AI Overviews in desktop searches according to the data available from Semrush.

Why classic SEO alone is no longer enough

Classic SEO is still essential: technical health, speed, crawlability, proper architecture, internal linking and quality backlinks remain critical. But AI Search adds a new layer: generative engine optimization, or GEO. GEO does not replace SEO, but extends it. Its goal is to make a brand's content more easily readable, trustworthy, and usable by generative systems that synthesize responses. This means we don't just write to get a SERP position. We write to prove we have the best, most useful, and most reliable answer to a specific problem.

In practice, this changes the priority of actions. A simple category page with products and a short 150-word SEO text at the bottom of the page will hardly meet the needs of a conversational search experience. Instead, a comprehensive category page that explains how to choose, what criteria matter, what differences there are between materials or technologies, how to measure size, how to compare models and what questions customers ask, has more value. Semantic SEO becomes a key mechanism here: we cover thematic entities, relationships, questions and sub-prefixes instead of mechanically repeating a keyword. The same goes for product pages. Product descriptions that simply copy the vendor are not enough. You need unique insights, actual usage, photos, reviews, FAQs and structured data.

The rise of AI assistants reinforces this need. OpenAI has announced a significant increase in ChatGPT usage, from 100 million weekly active users in November 2023 to 200 million in August 2024, 400 million in February 2025 and 800 million in April 2025. These numbers are not exclusively about search, but show how quickly the public is becoming accustomed to environments where responses are interactive, direct and synthetic. The more users learn to ask AI tools, the more they will expect a similar experience from search engines, eshops and customer support channels.

The graph below shows the rapid increase in weekly active users of ChatGPT, which serves as an indication of wider adoption of AI interfaces.

Step-by-Step guide to e-commerce strategy in AI Search

The first step is to map intents, not just keywords. Start with your core product categories and capture what the customer is really trying to solve. For example, behind the keyword «office chair» there may be intents such as «chair for long hours», «chair for small space», «ergonomic chair for back pain», «chair up to 150 euros» or «office chair for tall person». This analysis helps you create content that answers real needs and not just search terms. Combine data from Google Search Console, internal eshop search, customer questions, reviews, customer service tickets and keyword research tools. The result should be an intent map by category, prioritized by commercial value and frequency of queries.

The second step is to upgrade the pages that play a role in the purchase decision. Don't necessarily start with blog posts. Start with category pages, top product pages and buying guides. Add short answers to critical questions at the top of the page, comparison tables where they make sense, clear information about availability, shipping, returns, warranties and after-sales, and content that explains when a product is suitable and when it is not. This honesty reinforces brand authority, because AI systems and humans alike better evaluate content that doesn't look like generic advertising. If a product isn't right for everyone, say so. Credibility often converts better than over-promising.

The third step is to implement structured data. Use Product schema, Offer schema, AggregateRating, Review, FAQPage where it's really useful, BreadcrumbList and Organization schema. Structured data doesn't guarantee display in AI Overviews or rich results, but it helps engines understand exactly what's on the page. For e-commerce SEO, this is important because products have many structured attributes: price, currency, availability, SKU, brand, color, size, material, ratings and shipping policies. The more clearly stated, the less uncertainty for the engines and the better the likelihood of your information being leveraged in enriched or synthetic results.

Practical application in 30 days

In the first 7 days, select 10 pages with high commercial relevance and extract their queries from Google Search Console. Group them into informational, commercial investigation, and transactional intent. From day 8 to day 15, enrich the pages with answers to the most frequently asked questions, improve titles and meta descriptions, add internal links to relevant products and guides, and remove duplicate or weak descriptions. From day 16 to day 23, apply structured data and test the result with Rich Results Test and Schema Markup Validator. From day 24 to day 30, create a thematic content cluster around your most profitable category: a buying guide, a comparison article, a FAQ resource, and improved product descriptions. After 30 days, measure not only clicks and rankings, but impressions, CTR, branded searches, assisted revenue, and on-page user behavior.

How to measure whether you are winning in the new environment

In AI Search, success is not always immediately visible through a classic organic traffic report. A brand can influence the purchase decision without getting the first click, especially in zero click searches. That's why a broader measurement framework is needed. Track impressions and average position in Google Search Console, but analyze them by query intent, not just overall. Check if branded searches are increasing, because a user may see your brand in AI response and later search for you directly. Measure assisted conversions in GA4, especially for users returning via direct or paid after an organic contact. Track engagement metrics such as scroll depth, add-to-cart rate, product comparison usage and internal search queries. If you have on-site search, user searches within the eshop are a goldmine for content optimization.

At the same time, create a periodic AI visibility check process. Search Google, Bing, Perplexity, ChatGPT Search where available and other AI interfaces for your customers' key queries. Record whether your brand is featured, which competitors are mentioned, what types of sources are used and what content gaps are identified. Don't treat this process as an absolute science, because AI results are variable. Treat it as a qualitative research layer that reveals where you need to strengthen documentation, comparisons, evaluations and topic coverage. AI Search rewards brands that have clear expertise and a consistent presence across the ecosystem: website, reviews, social proof, PR, marketplaces, YouTube, knowledge panels and trusted third-party references.

The strategic direction is clear: keep investing in SEO, but adapt it for a world where the answer comes before the click. Create content that is accurate, useful, well-structured and commercially relevant to the customer experience. Reinforce E-E-A-T with real experience, writers or experts where it makes sense, clear policies, reviews and proof of credibility. Improve product pages to act not only as selling points, but also as knowledge sources. AI Search doesn't eliminate the need for a website; instead, it raises the bar for which website is worth using as a resource. Those e-commerce brands that move early will gain an advantage not because they «outsmarted» an algorithm, but because they have better organized their knowledge, expertise and marketing proposition.

Frequently Asked Questions