E-commerce: product architecture for search and AI

Problem

Product sites fail in AI search for a specific reason: the facts a machine needs are in the photograph, in the variant selector or in a PDF, and never in text that can be extracted.

• Category and product pages competing for the same queries.
• Product attributes present visually but absent from readable content.
• Buying questions answered in reviews and chat, not on the page.
• Weak internal connections between a category, its products and related content.

Diagnosis

The site was mapped as a chain rather than a list of pages. Every gap in the chain is a point where an assistant loses the thread and recommends something else.

Each link was checked for three things: does the page exist, does it state the fact in text, and does it point to the next step.

Implementation

• Category architecture with one clear purpose per category page.
• Product page template carrying composition, use, size and suitability as readable text.
• FAQ content built from real buying questions and attached to the product it belongs to.
• Structured product information, including reviews where they exist.
• Internal linking between category, product and supporting content.
• Metadata and headings rewritten so each page states what it is on the first line.

Measurement

Search Console query and page coverage for category and product URLs, plus citation checks on product style questions of the form "which product for X" in ChatGPT with search, Perplexity, Google AI Overview and Gemini.

Result

{{RESULT_SENTENCE_OR_DELETE}}

Delivered: a product structure in which every commercially important attribute exists as text, every category has one job, and the path from a general question to a specific product is explicit rather than implied.