Four models, one editor
Claude, ChatGPT, Gemini and Grok answer the same question; an editor checks the picks on Amazon.
Product Guides / Publishing (US) · Case Study
An English guide site that asks Claude, ChatGPT, Gemini and Grok the same buying question, shows where they agree and checks every pick on Amazon.

Results
The Brief
Readers already ask an AI their product question. A guide site could stand out by asking four models at once and showing the result honestly. The old site's invented expert quotes and fixed ratings undermined that trust.
The Approach
Every guide opens with a "how this page was made" block. We removed star and review counts that broke Amazon's rules and show price as a band. New guides are produced in the "AI Panel" format.
See the full process → 47-Point AI-Ready SEO Audit methodology.
Tech Stack
The Build
Guides come in two formats: Amazon customer-review research and an AI Panel built from four models' answers. Each page states its format at the very top.
Under Amazon Associates rules, customer stars and review counts may only be shown via the official API. Visible stars and schema ratings were removed.
The pre-launch audit cleared a wildcard redirect that sent 19 real pages to 404, invented "Dr." quotes, and a fixed rating schema on hundreds of pages.
Steps Applied
Claude, ChatGPT, Gemini and Grok answer the same question; an editor checks the picks on Amazon.
Every guide states its source and method on the first screen.
Price shown as a band, ratings only where the API allows. No fabricated rating schema.
Honest Reflection
We first built the site around "experts tested it" language. The pre-launch audit found that claim on hundreds of pages and not in any shared component, so fixing it meant rewriting 270 files sentence by sentence. Today I'd put the method statement in a single component from day one.
Screenshots


Same 47-point methodology, $499 fixed price, 5-7 business days. Book the audit.