AI agent as a gateway to the consumer

Traditional advertising and SEO are no longer enough. Customers are increasingly using AI assistants, who analyze needs, compare products, and make choices for them. If your product data is not prepared to be read by these systems, your offer may not appear in their recommendations at all.

The McKinsey Report (2025) calls this phenomenon agentic commerce - the moment when:

McKinsey forecasts that by 2030 agent-mediated shopping could generate $3-$5 trillion in revenue globally. Deloitte data shows that 72% of retail companies are already testing or implementing integrations with generative AI in the sales process.

It's not an experiment - it's a redefinition of the sales channel. Today, when a consumer asks ChatGPT "recommend shoes for running in the woods", generative AI artificial intelligence selects products based on data structures and context, not CPC rates. If your product doesn't exist in a model-readable structure, it won't be shown - even if you have a better offer.

According to Bain & Company, already 60% of queries on AI-based search engines end up without a click on the page - the user gets a response directly from the agent. In practice, this means that a store whose data is not understood by AI models is out of business.

Data instead of ads - the semantics of new SEO

SEO in generative models is based not on links, but on AI models' understanding of product data.

McKinsey emphasizes that:

It is Generative Engine Optimization (GEO) - a new field that combines product feed technology, user language and context, and data structures. In practice, this means that AI doesn't need your advertising. It needs a product description that is precise, contextual and machine-understandable.

Data from Synerise Research Hub (RecSys 2025) confirm that context-based models increased the accuracy of product recommendations by 25%, and the conversion rate of users from AI channels is about 18-22% higher than from search engines.

Case study: retail without clicks and commissions

The Polish sporting goods retailer (about 800 SKUs) has incorporated the XML feed into AI channels in the Semly.ai. Within three months:

• the number of product recommendations in ChatGPT and Gemini increased by 136%,
• conversion from AI traffic was 6,9% (versus 1.8% from Google Ads),
• the cost of customer acquisition has dropped from 11 EUR to 0.5 EUR (only the cost of the monthly subscription),
• traffic from AI was responsible for 17% revenue at zero cost per click.

This confirms the main thesis of the report Deloitte (2025):

What determines visibility in AI?

An analysis of the Data Economy Congress (2025, Warsaw, Poland) identifies four key factors determining whether a product will appear in AI agents' responses:

1. Quality of product data - detailed technical descriptions, properties, categories, applications.
2. Timeliness of XML feeds - no errors and regular refreshing of metadata.
3. Semantic context - product linkage with use ("ski jacket for cold weather up to -20°C").
4. Agent's trust in the source - stores with positive reviews, fast deliveries and clear return policies are more likely to be quoted.

Synerise experts add:

Change in definition of sales channel

In the "agentic" world, the customer becomes not the person with the browser, but the AI agent interpreting its needs.

McKinsey writes directly:

This forces a new way of thinking: not keyword campaigns, but data under linguistic intent.

Companies that learn to provide agents with context will gain precedence as "sources of product truth."

What to do now?

1. Modernize your product data - define a set of information about each product (material, application, target, conditions of use).
2. Treat AI as a new distribution channel - just as mobile commerce or social commerce used to be.
3. Analyze in which queries your products appear. If they are not in the conversational results - no context.
4. Use data monitoring tools with AI, for example. Semly - analyzes intentions, and behavioral data increases the accuracy of recommendations.

These are the four pillars of the new SEO for generative models.

As he put it mcKinsey report:

E-commerce is entering a stage where the key is no longer traffic, but rather understanding the product through artificial intelligence models.

Where neither ads nor Google reach, AI is already directing customers to a store that has done its data homework.

Share:

Copy link