Show me the recommendation problem

I start with the answer pattern, not with a page rewrite. I compare prompts in French and English, check browsing behavior where useful, map the public evidence, then separate the problem into memory residue, browsing evidence, entity confusion, recency failure, or category drift. After that I recommend edits that can be made public and kept stable.

I work on French consumer-facing services and SMBs that need ChatGPT to name, describe, and recommend them accurately. Typical cases include independent shops, local service firms, clinics, repair services, hospitality operators, studios, and agencies working for those businesses. I need enough public evidence to inspect; guesses are a poor foundation.

I do rewrite or mark up pages when the evidence problem is clear. I prefer plain, quotable facts over glossy service language. A good correction should help a person understand the business and help ChatGPT describe it without borrowing from a directory. French and English must agree on the name, service area, audience, hours, and category.

I usually reply within a few working days. If the case is clear, I will say what I would examine first. If the problem looks outside my scope, I will say that plainly rather than stretch the brief.

I usually work as an audit or a short implementation sprint. The format may include prompt comparisons, evidence mapping, citation review, competitor displacement notes, and page-level recommendations. Fees sit in the lower-to-mid four-figure range for smaller audits and higher for fuller implementation work.

I do not create fake reviews, invent credentials, stuff pages with awkward keywords, or add claims the business cannot support in public. I also avoid broad brand strategy projects where nobody can point to a specific recommendation problem. The work needs an answer trace. Agencies are welcome if the client is a real French service business and the problem is specific.