When an audit blames the site for one template
Most audit findings are a count: some number of your pages are missing something. The count is usually correct and the unit is usually wrong, because the thing being counted was not decided page by page. It was decided once, in a template, and the count is the number of pages that template produced.
What this looks like in practice
A tool reports that a site has no share cards. Somebody goes to add them, and finds the most important pages on the site — the ones people actually share — already have complete card markup. The pages without were a different template entirely: a listing, or an archive, or something generated by a different part of the system.
That is a real case, from an audit this product ran. The count was accurate. The verdict was not, and the advice — add the card type — did not say where, so the only way to act on it was to go looking for pages that were already correct.
Why the unit matters more than the number
Almost every property an audit measures is set in a layout, a partial, a theme file or a component. Titles, descriptions, canonical tags, card markup, structured data, heading order. One edit, applied to every page that uses it.
So findings cluster by template, not by page, and a count without a location is close to useless in both directions: it overstates the work when one edit fixes everything, and it understates the danger when a template you thought was fine turns out to produce a third of the site.
A majority is not a site. "Most of your pages" and "your site" are different claims, and the gap between them is exactly where the useful information lives.
What to do with a finding like this
Ask which template the affected pages share, before doing anything. Sort the affected URLs by their path and the answer usually falls out immediately: everything under one section, or every page of one kind. That grouping is the actual finding.
Then check one page that was not flagged. If a page of the same kind is correct, you are looking at a template that was fixed once and not everywhere — a different problem from one that was never done.
Fix it where it was decided. If the answer is one file, the count was never the size of the job.
What a tool should do, and what it cannot
It should say every page only when it is every page, give a proportion otherwise, and name the URLs so the grouping is visible without exporting anything. A finding that reports a majority in the voice of a totality is making a claim it did not measure.
It cannot tell you which template is which. A crawler sees URLs and HTML; it does not have your theme, your components, or the knowledge that two paths come from the same partial. The grouping it can offer is a hint you confirm, not an answer — and a tool that claimed otherwise would be guessing at your CMS.
This one got it wrong in the direction described above, and the check now distinguishes "every page" from "most of them". A page citing a check is worth more for saying when the check was wrong than for implying it never is.
The narrower question about share cards
Cards are worth having, and they are worth raising only once the basic sharing markup is already in place — a site with neither has one problem, not two, and being told twice is noise rather than thoroughness.
The broader question of how to tell whether a tool is lying to you is a separate piece.
Common questions
Why do audit findings overstate the work?
Because they count pages, and most of what they measure was decided once in a template. The count is the number of pages that template produced, so one edit often fixes all of them.
How do I tell whether a finding is a template problem?
Sort the affected URLs by path. If they all sit under one section, or are all pages of one kind, that grouping is the real finding. Then check a page that was not flagged — if it is the same kind and correct, a template was fixed once and not everywhere.
Is 'most of your pages' the same as 'your site'?
No, and the gap is where the useful information lives. A finding that reports a majority in the voice of a totality is making a claim it did not measure.
Can a crawler work out which template a page came from?
Not reliably. It sees URLs and HTML, not your theme or components. Any grouping it offers is a hint you confirm rather than an answer, and a tool claiming otherwise is guessing at your CMS.
Do I need share cards if I already have basic sharing markup?
They are worth having, and worth raising only once the basic markup is in place. A site with neither has one problem, not two, and being told twice is noise rather than thoroughness.