Product page analyser for cross-border selling

A product detail page that converts at home often fails abroad for reasons that never show up in analytics: the size chart assumes a local standard, the ingredient list is only in one language, the hero image shows a use case that does not translate. This tool reads the page, looks at the images, and reports the gaps that matter for a specific market.

Paste a product page URL. We read the page the way a buyer in another country meets it — including the images, which is where most of the real information lives — and tell you what would stop it from selling abroad.

How it works

We fetch the page and pull its text, structured data and images.
The images go to Claude as images — not as filenames. Ingredient panels, size charts, certification marks and usage shots are read the way a person reads them.
If reviews sit in an embedded widget, we read that frame too.
You get a readiness score, the specific gaps with the evidence behind each, and a market-by-market read.

What it cannot see

Anything drawn only after the page loads, by JavaScript.
Review pages you reach by clicking through pagination.
Anything behind a login or a region block.

The result names exactly what was read, so you can judge how much weight to give it.

FAQ

Does it actually look at the product images?

Yes. The images are sent to the model as images, so the ingredient panel, size chart and certification marks on them are read directly. Most of what decides a cross-border purchase is in the imagery rather than the HTML.

Will it work on a marketplace listing?

Usually. Amazon, Shopify, Cafe24 and most storefronts serve enough in the initial HTML to analyse. Some marketplaces render their detail section entirely in JavaScript, and the result will say so rather than guess.

Does it read customer reviews?

It reads reviews that are in the page text or in an embedded review frame. It does not click through paginated review pages, and the result states which of those applied.

Is my URL stored?

No. The page is fetched, analysed and the result returned. Nothing about the request is written to a database — this site has no database.