Free Product Page Report
Free product page report: paste one product URL to see what search engines, feeds and AI agents can read, scored on a 100-point rubric with evidence and fixes.
Paste the address of one public product page. The report fetches it the way a search engine or AI shopping agent would, scores what it can read on a 100-point rubric, shows the evidence behind every finding, and lists what to fix first.
Product Page Report
We fetch the page once as ClaroDemo/1.0, respect robots.txt, and read only what the server sends (no login, no JavaScript rendering).
The URL and the report are processed by Claro; identical URLs are cached for an hour.
What the report checks
The report reads the HTML your server returns and scores five things a buyer, a catalog team and a machine all need:
| Section | Points | What it looks for |
|---|---|---|
| Identity | 25 | Brand and manufacturer part number, in markup or labelled on the page; a GTIN with a valid check digit; pack wording (“2-pack”, “case of 12”) that the markup reflects; variants with their own SKU or GTIN |
| Specifications | 25 | Visible spec rows, with numbers carrying their units; a fit or compatibility statement |
| Structured data | 20 | Product markup (JSON-LD) with name, image and offers that parses cleanly; markup that agrees with the visible name, price and availability; a variant group when the product has variants |
| Evidence | 15 | Linked documents such as datasheets, manuals or SDS; images with alt text; a revision or update date |
| Transaction | 15 | Price with currency (a quote path counts), availability or lead time, and shipping, returns or an inquiry route |
Checks that don’t apply to the page, such as variant markup for a product with no variants, or a GTIN when brand and part number already identify it, are marked N/A and left out of the score. The score is the share of applicable points earned.
Each finding is Found, Missing, Conflict (for example the markup price differs from the price on the page, or a GTIN fails its check digit) or N/A, with what we observed: the JSON-LD value or a short snippet of the visible text. The fix list comes from the failed findings, most severe first. No AI model writes any part of it, so the same page always gets the same report.
Reading the result
- 80 and above: the essentials are present and consistent. Next, check that the category’s key attributes are on the page. The Attribute Library lists them per category.
- 50 to 79: something a buyer or a machine relies on is missing or contradicts itself, usually identity, spec detail or the markup.
- Below 50: machines see little beyond a title and a price. Expect the page to be skipped by shopping surfaces and misread by AI assistants.
- Not measured: we couldn’t assess the page, and the report says why: robots.txt, an error status, an unreachable site, a non-HTML response, a JavaScript-only shell, or a page that isn’t a single product. That is a finding, not a zero.
From one page to a whole catalog
A single page is a sample. The gaps it shows (a missing part number, specs that live only in a PDF, variants sharing one GTIN) usually repeat across a supplier’s whole range. Claro runs the same checks across every SKU, fills missing attributes from the manufacturer’s documents with a source for each value, and writes approved values back to your PIM or ERP.
What does the product page report check?
It fetches one public product page and scores it on a 100-point rubric in five sections: identity (brand, part number, GTIN check digit, pack and variant scope), specifications (spec rows with units, fit and compatibility), structured data (Product markup and whether it agrees with the visible page), evidence (documents, images, revision dates) and transaction (price, availability, shipping, returns or a quote path). Every finding shows what was observed and how to fix it.
Why did my page get no score?
If we can’t read the page, we don’t score it, and we tell you why: robots.txt blocks crawlers, the server returned an error, the site was unreachable, the address isn’t an HTML page, the page is an empty shell built by JavaScript, or it isn’t a single product page. An unreadable page is a finding in itself, because search engines and AI agents that fetch it the same way hit the same wall.
Does it render JavaScript?
No. The report reads the HTML your server sends, as many crawlers and AI agents do. If that HTML is an almost empty application shell, the report says so instead of scoring it. If only some product data is added by JavaScript, that data shows as missing, and so it will for any crawler that does not render pages.
Does a high score mean my product data is correct?
No. The report checks that the facts are present, machine-readable and consistent on this one page. It cannot tell whether a value is accurate or current for the exact variant, a valid GTIN check digit does not prove who owns the code, and it does not see your ERP, PIM or supplier documents. Validate important values against the manufacturer’s current source.
Is it free, and what happens to the URL?
It’s free for a few reports a day with no login; confirm your email to unlock 25 a day. Claro’s crawler, identified as ClaroDemo, fetches the URL once and respects robots.txt. Reports for pages that could be read are cached for an hour. No AI model is involved: the same page always gets the same score.
Claro
See where your catalog breaks — free
Claro runs this automatically: resolve identity, fill missing attributes, validate updates, and write clean records back into your PIM/ERP. Upload a sample supplier file for a free catalog audit.
Get a free catalog audit