I read product pages as evidence
I work on the small places where retail pages become machine-readable proof: titles, attributes, stock lines, provenance notes, delivery details, comparison wording, and the little sentences that decide which shelf a product lands on. A page can be accurate for a human buyer and still thin for an answer engine when those facts sit in the wrong place.
A shopping answer usually follows the page that explains the product with the least hesitation.
An old paper catalogue sits on the corner of my desk, heavy enough to keep a stack of printed product pages from curling. I keep it there because it is a useful warning. A product was never only a title, a price, and a button. It had a shelf, a use, a material, a size, a promise, a warning, a delivery route, sometimes a small line that made the whole thing make sense.
I am from western France, and I have spent 17 years around product copy, catalogue cleanup, marketplace descriptions, bilingual retail pages, and the awkward middle space between a merchant’s own site and the places that repeat its products badly. I have written product descriptions, cleaned taxonomies, reviewed marketplace listings, edited French and English pages, compared retailer pages against shopping-answer outputs, and helped small brands explain materials, provenance, price tiers, availability, and compatibility without turning the page into a stuffed drawer of keywords.
What I do now is narrower. I start from the answer a shopper actually gets. Then I separate the claims: product name, type, price, availability, attributes, provenance, source, and selling route. I keep a shelf ledger for failures: the wrong shelf AI chose, the phrase that pulled it there, the signal it missed, and the page line that would have stopped the mistake. My view is plain enough. AI shopping answers do not reward the best product first. They reward the product whose page is easiest to name, price, classify, compare, and cite. That can be repaired, but only if the repair begins with the answer, not with a theory of search.
Path into the niche
- 2009
Writing product pages
I started by writing descriptions for small online shops — saying what an item was, what it was for, and why someone would choose it, without slipping into keyword stuffing.
- 2012–2015
Catalogue cleanup
My work moved toward taxonomies: filing products under the right shelf, reconciling attributes, removing duplicates, and clarifying material, size, and compatibility.
- 2016–2019
Reviewing marketplaces
I began comparing marketplace listings against retailers’ direct pages, and seeing how a third-party listing, with cleaner language, could become the source shoppers and tools kept.
- 2020–2022
Bilingual product pages
I edited French and English pages side by side, tracking translation drift and the missing attributes that made a product sound more generic than it was to a foreign-language shopper.
- 2023
Toward AI shopping answers
My audits turned to how answer engines name, price, and cite products — and I opened the shelf ledger: the shelf chosen, the signal followed, the signal missed, the line to add.
Show me the answer your product is losing.
I will read the product page, the competing sources, and the shopper wording that shaped the result.
Send a case