One error type can occur on several pages, and one page can have several findings. That’s why you can’t add the number of error occurrences to the number of affected URLs.
Suppose four pages were checked. Three types of findings were identified, with five individual occurrences, but only three URLs have issues. Let’s count them row by row so the summary doesn’t inflate the list of work.
Name what you’re counting
An error type is the rule or category used to identify a deviation. An occurrence is a specific instance of that deviation. An affected page is a URL with at least one finding under consideration.
One page can have several different findings. One type can occur on many pages. Some checks apply to the site or a general response, so you shouldn’t force every row into a separate problem URL.
“Duplicate Title” is a type; one match found on a page is an occurrence; that page’s URL is the address. Title means the page name in a meta tag, usually visible in the browser tab. To interpret a value in a report, read its label first: different checks may count occurrences differently.
Count occurrences in the table
| URL | Duplicate Title | Incorrect link | Missing page description |
|---|---|---|---|
| /a | Yes | Yes | No |
| /b | Yes | No | No |
| /c | No | Yes | Yes |
| /d | No | No | No |
There are three types under consideration. The first column has two “Yes” entries, the second has two, and the third has one. That makes five marked occurrences: 2 + 2 + 1.
There are findings on /a, /b, and /c. That’s three unique pages, not five. For the first two types, the address lists are [/a, /b] and [/a, /c]. Their union is [/a, /b, /c], because /a appears twice.
If two different links are broken on one page, a single “Yes” isn’t enough to count the links. Add them as separate rows with their destination addresses. For general guidance on interpreting a report, see the SEO audit guide.
State the proportion only for the pages checked
If the audit covered only some of the pages, state the proportion relative to that set. In this example, three of the four URLs examined have the listed findings. This doesn’t prove that the same proportion applies to the entire site.
An automated daily audit of a connected site has an upper limit of up to 100 pages. The scope of a full manual audit is determined by the selected package. So it’s important to include the number of addresses actually covered and any limitations in the report, rather than using “full” as a promise of an unlimited crawl.
For unchecked addresses, an additional technical audit is needed if their condition affects the work plan.
Turn the numbers into specific tasks
Five occurrences may require one template change, three editorial edits, or several independent investigations. The count alone doesn’t tell you how much effort is involved. To determine that, compare examples and causes.
Choose one or two URLs from each group. Check whether the error type is related to a shared block, content entry, or a specific address. The article on template errors can help you choose control pairs for this review.
For a shared template, agree on one task based on the selected examples. For an incorrect link, identify the source page and the intended destination. Don’t multiply the cost of one fix by the number of rows before investigating the cause.
Check the count before sending the summary
A concise note for this example: “We examined four URLs. Three types of findings produced five occurrences across three different pages. We’re checking the shared title block and the link; the status of other sections is not yet known.”
In the next report, compare the same list and the checks performed. Four occurrences instead of five could mean a fix was made or the list of pages examined changed. To close a specific task, repeat the original address and action.
Attach a short list of URLs before sending. A colleague should be able to reproduce the total from the same rows without adding overlapping groups together. If the number can’t be reconstructed, clarify its label and counting method.