What Adobe Analytics data revealed

The analytics company Adobe Analytics published data on June 15, 2026, mapping the behavior of American consumers arriving at e-shops from AI tools — primarily from ChatGPT by OpenAI and Google Gemini. The results are clear for online sellers: AI-referred traffic is the most valuable you can have today.

According to the report, covered by Reuters, visitors coming from large language models (LLMs) generate 53% higher revenue per visit than those coming from traditional sources such as search engines, social networks, or email newsletters.

Key figures from May 2026:

• Conversion rate: AI-referred visitors were 54% more likely to make a purchase than others.
• Time spent on site: They spent 53% more time on e-shops.
• Browsing: They viewed more individual pages per visit.
• Year-on-year growth in AI traffic: Traffic from AI tools to retail websites surged 138% year-on-year in May.

Why do AI customers spend more? Personalization is key

Vivek Pandya, director of digital insights at Adobe, explained that when a product appears in AI recommendations, merchants are able to offer a much higher level of personalization to customers who then click through to complete the purchase.

It works simply: a user asks ChatGPT or Gemini, for example, "which vacuum cleaner is the quietest for an apartment with pets?" and the AI responds with specific products and links. The customer arrives at the e-shop with a clear idea and high purchase intent — unlike someone who is just aimlessly browsing.

Moreover, AI acts as a filter. It won't direct you to a product that doesn't match the query. This dramatically increases the relevance of the visit — and with it, the willingness to spend.

What this means for e-commerce: the SEO era expands to include AIO

This data confirms a trend that analysts call AIO — Artificial Intelligence Optimization. While until now e-shops focused mainly on search engine optimization (SEO), they now must add optimization for language models as well.

In practice, this means that product pages must be structured so that AI models can easily read, understand, and index them. Key factors include:

• Structured data (schema.org markup) — product information in a machine-readable format
• High-quality and up-to-date descriptions — AI models prefer detailed, factually rich texts
• Availability of APIs or product feeds — enables AI models to work directly with the catalog

For Czech e-shops like Alza, Rohlik, Mall, or Notino, this means that investment in the technical quality of product data today is not just a matter of better SEO, but directly affects whether they appear in AI assistant responses at all.

The European dimension: AI Act and model openness

While Adobe Analytics data maps the American market, the trend is rapidly spilling over into Europe. ChatGPT is available in the Czech Republic for free as well as in paid versions (ChatGPT Plus for $20 per month, approximately 460 CZK). Google Gemini is fully functional in Czech and integrated into Google Search, which dominates in the Czech Republic.

The European AI Act, which came into effect in 2024 and whose key provisions are becoming binding throughout 2026, adds another dimension. It requires transparency in how AI systems work with data. For e-shops, this means a double challenge: to be visible in AI responses while also complying with European personal data protection rules (GDPR) and algorithmic decision-making transparency.

Moreover, the Czech Republic has become home to the Czech AI Factory in Ostrava, signaling the growing importance of AI infrastructure in our region. Companies that start preparing their data for AI readability today will have a head start.

How to prepare: three steps for Czech e-shops

1. Implement structured data. Schema.org markup for products (with price, availability, rating) is the absolute minimum today. Without it, AI models simply won't find you.

2. Keep product feeds up to date. Google Merchant Center and similar tools are now the primary source from which Gemini draws product information. Outdated prices or unavailable goods will disqualify you.

3. Think about query context. Users don't ask "laptop 16 GB RAM", but "which laptop to buy for university with a budget of 15 thousand". Adapt your descriptions accordingly — speak the customer's language, not catalog numbers.

It's not just about ChatGPT and Gemini

Although Adobe Analytics data primarily tracks ChatGPT and Gemini, the same principle applies to all AI assistants — including Microsoft Copilot (integrated into Bing and Windows), Claude by Anthropic, or Perplexity AI. Each of these tools can generate highly qualified traffic — if your website allows them to.