Find the schema bug that repeats on a thousand pages
Checking structured data one URL at a time tells you a page is broken. This reads your sitemap, checks up to 40 pages spread across your page types, and tells you which problems belong to a template — so you fix one file instead of chasing a thousand identical errors.
We find your sitemap and check up to 40 pages spread across your page types — not the first 40, which would all be the same template. Two requests at a time, and we obey your robots.txt. Nothing is kept unless you ask for a share link.
Why does a single-page check mislead at scale?
Because it answers the wrong question. A page-level validator hands you a list of findings for one URL, and on a site of any size that list is indistinguishable from the list you’d get for any other URL built from the same template. You end up with a thousand copies of one problem and no way to tell that’s what you’re looking at.
The useful unit is the template, not the page. “Every product page is missing priceCurrency” is a sentence a developer can act on in one commit. “This product page is missing priceCurrency” is a ticket that gets closed and immediately reopens somewhere else.
How do you decide which pages to check?
By grouping your URLs into path templates and sampling across them, rather than taking the first pages in your sitemap. This is the single most consequential decision in the tool.
Sitemaps are almost always ordered by section. Take the first forty URLs from a typical shop and you get forty blog posts — a perfectly executed check that tells you nothing whatsoever about the pages that take money. Grouping by template and sampling round-robin means every page type gets looked at before any page type gets looked at twice, so a site with three thousand posts and four product pages still gets its shop audited.
The report says exactly what was sampled and what was skipped. If your sitemap index lists twenty-four sitemaps and we read five, it says so — including that the URL count you see is from those five, not from the whole site.
When will you say a problem is site-wide?
Only when every sampled page of a template failed, and we sampled at least two of them. That’s when you’ll see “fix the template once and it corrects 1,240 pages”.
If three of five failed, the report says three of the five we checked, and makes no claim about the rest. One page failing proves nothing about a template at all — it is an anecdote, and we label it as one. A sample can establish that a pattern exists; it cannot establish that a pattern is universal, and a tool that blurs those two is a tool whose numbers you can’t repeat to a client.
What does the audit actually check?
The same rule library as the single-page schema validator — syntax, required properties per type, Google’s rich result requirements, and whether AI answer engines can reach the page at all — applied across your sample and rolled up by template. Crawler access is reported once at site level rather than repeated forty times, because it’s a property of your domain, not of any one page.
You also get two coverage numbers that only make sense across a whole site: how many pages have no structured data at all, and how many have it only in JavaScript — which means it exists for Google’s renderer and does not exist for ChatGPT, Perplexity or Claude, none of which run JS.
Frequently asked questions
Why check a whole site instead of one page?
Because a broken page is rarely a broken page — it's a broken template. Checking one product URL tells you that URL is missing a currency. Checking twenty tells you every product page is, which is one fix in one file rather than a list of a thousand identical problems. That difference is the entire reason this tool exists separately from the single-page validator.
How many pages do you check?
Up to 40, and we tell you exactly which ones and what we skipped. Fetching an entire large site would take hours and put real load on your server for very little extra information — once you've checked twenty pages built from the same template, the twenty-first tells you nothing new.
How do you choose which pages to check?
We group your sitemap's URLs by path template — /blog/*, /products/* and so on — then sample round-robin across the groups. This matters more than it sounds. Sitemaps are usually ordered by section, so simply taking the first 40 URLs means checking 40 near-identical blog posts and learning nothing about the shop. Round-robin means a site with 3,000 posts and 4 product pages still gets its product pages looked at.
You checked 40 of my 5,000 pages. Can I trust a conclusion from that?
Only where we say you can. We claim a problem affects a whole template when every sampled page of that template failed and we sampled at least two. If three of five failed, the report says 'three of the five we checked' and extrapolates nothing. A sample can prove a pattern exists; it can't prove one is universal, and we don't pretend otherwise.
Will this hammer my server?
No. Two requests at a time, at least 250ms apart, and we honour a Crawl-delay directive if your robots.txt sets one. We also obey your robots.txt for ourselves — if it asks crawlers not to fetch a page, we skip it and say so. A tool that lectures people about crawler etiquette while ignoring theirs has no standing to say anything.
Does this need my sitemap?
Yes, and we look for it in the usual places: the Sitemap: line in your robots.txt first, then /sitemap.xml and the common variants, including gzipped ones. If there isn't one, that's worth fixing on its own — without a sitemap, search engines and answer engines have to guess which pages exist. You can still check pages individually with the schema validator.
How often should I re-run it?
After any template change, and after a migration. Structured data breaks silently — nothing on the page looks different when a field stops being emitted — so it's the kind of thing that stays broken for months unless something checks it.
Found a problem that spans your whole catalogue?
Template-level structured data is part of the SEO and development work we do for small and medium businesses. Send us the audit and we’ll tell you what it would take to fix.
Checking a single page or some markup you’ve written? Use the schema validator — it validates pasted JSON-LD instantly in your browser. · All free tools