SEO and AI visibility for Shopify & e-commerce.

In retail, it is not the advertising text that decides, but the product data. AI systems and buying agents read the feed, schema, and product page; and recommend what they fully understand. If information is missing or contradicts itself, no agent asks, they name a different product.

Whether fashion, furniture, technology or spare parts: We work on product titles, attributes, variants, category texts, feeds and availability - the places where catalogs actually fail. For hundreds of items, software takes over the checking; the approval remains with you.

Product research is shifting. According to HubSpot State of AI (2025), 43% of consumers use AI assistants to replace Google when researching products, and 68% trust AI recommendations when making purchasing decisions. For e-commerce, the average AI visibility score in our measurements is 25 out of 100 - the range between average and top 10 percent (75+) is hardly greater in any industry.

A second shift is technical: Since October 2025, ChatGPT has been able to trigger purchases via the Atlas Browser and the Agentic Commerce Protocol (4% merchant fee), and since January 2026 there has been a fee-free alternative with the Universal Commerce Protocol from Google and Shopify. This means that the purchasing agent is no longer a hypothetical scenario, but rather a channel with its own data requirements.

What slows shops down is rarely marketing. It's product data: missing attributes, conflicting information between feed, schema, and product page, and descriptions that don't answer buyers' questions. An agent who can't resolve a claim doesn't recommend the product; he doesn't ask.

The quantity makes matters worse. A catalog with several hundred items cannot be checked for completeness by hand, and automatic optimization tools solve the problem by making it invisible: They change hundreds of texts at once without anyone being able to understand which change had which effect. The catalog is thus filled, but the question of the effect remains unanswered.

A division of labor therefore makes sense: software takes care of checking and preparing the entire catalog, while people decide what goes live. This is exactly how SPHEN.AI, our Shopify app developed together with NICCOS, is built. It synchronizes the catalog, finds gaps for each product and presents drafts for approval - nothing is published automatically.

The same practical approach applies to prioritization as in the rest of our work: the products with the largest share of sales and margins come first, not alphabetical order. In most catalogs, a few dozen items account for the majority of the results - they determine whether the work is measurable.