For the past month or so, I’ve been studying major retail brands in Europe and found that they are far from ready for the new Agentic Commerce era. This creates a great opportunity for smaller e-commerce brands.

The list includes 30 brands spread across 7 categories: Grocery (10), Fashion (5), Home Improvement (4), Furniture (3), Pet Care (3), Electronics (3), and Health & Beauty (2).

Tests were performed semi-automatically and included the following steps:

• AI Overview searches using structures such as “best for under ” and “how to ”;
• Website crawling with intelligent page categorization and interpretation;
• Internal structure and concept correlation on a dedicated website graph;
• Agentic Commerce protocol implementations (ACP, UCP, AP2, MCP, A2A, and others).

What was found:

• Little to no results for AI Overview prompts: Recommendations were often attributed to international brands or highly customized editorial articles. For example, for “Best chair for desk job under €300”, I expected Gemini to surface brands like IKEA or JYSK, but instead CNET and Shinoo were recommended.
• Most brands use highly structured websites, which is expected given that some have over 10,000 products. The system had little to no issue categorizing each page level and produced accurate predictions when given a random page URL.
• Most sites correlated products with other products, but this was usually category-based and functioned more as a comparison system rather than an upsell or contextual recommendation engine. This was especially noticeable in Fashion, Electronics, and Home Improvement. On the other hand, Grocery sites were able to relate specific products within recipe pages (which is impressive), but not the other way around (which is amusing).
• Not a single brand appeared to have implemented any of the tested Agentic Commerce protocols, which is odd considering that some of these protocols have been available for over six months.

So… how can you leverage this for your small e-commerce website?

1. Structure your website with hierarchical categories: This allows LLMs to more easily identify and crawl your products.
2. Write a Q&A section for every product: As users move from traditional website browsing to chatbot-driven recommendations, communication shifts from active to passive. LLMs tend to answer questions rather than lecture on a subject. Provide contextualized information and you will likely bypass generic text generation.
3. Build a highly correlated system across your site: Products ↔ concepts ↔ context. If possible, make it dynamic and relevant to the topic discussed in each blog article.
4. Use standard e-commerce providers: Many are already adopting major protocols such as UCP. Accept payments immediately, as this will likely push your product higher in recommendation systems.

As a small business, you have far more agility to build this than large brands with thousands of listed products. Leverage LLMs to write text, summarize content, and generate meaningful questions.

In my opinion, this is still an emerging topic, which is why it receives little attention. However, I’m confident that within the next 12 to 18 months, all 30 of these brands will have implemented these (and other) recommendations, making competition even tougher than it is today. For those who move faster now, the likelihood of staying on top is significantly higher.