Why Your Technical SEO Audit Didn’t Report Every Error

We run technical SEO audits and make recommended fixes for almost every client when we start working with them. We focus on the most critical errors first on the most critical pages and work our way down the list of priorities. Recently, one of our clients reached out to us saying that they’d received an email from another web company pointing out a bunch of errors. He sent us an email along the lines of, “Hey guys, I thought you were on top of this stuff. What gives?”

Technical SEO auditing tools like the ones we use from Semrush can produce different results across different runs. What it surfaces in one run may not reflect every error present on your website. In this article, I’m going to explain why.

The audit isn’t a snapshot of your whole site; it’s a snapshot of what was crawled

Tools like Semrush, Screaming Frog, Ahrefs, and Sitebulb don’t inherently know about every page on your domain. They crawl. That means they start somewhere (usually your homepage or a specified seed URL) and follow internal links outward, page by page, until they hit a limit. This is a page cap based on your plan, a time limit, a crawl depth setting, or the boundaries of how your internal linking is structured.

If your site has more pages than the crawler’s limit, or if large sections of the site are buried deep in your architecture (few internal links pointing to them, deep folder structures, orphaned pages), the first crawl won’t reach everything. It’s not evaluating your entire site; it’s evaluating the subset it managed to reach in that session.

Duplicate-content checks are relative, not absolute

This is the part that catches people off guard, specifically with duplicate H1s, titles, and meta descriptions. These checks work by comparing pages that were crawled in the same session against each other. If Page A and Page B share an identical H1, but only Page A got crawled in round one, there’s nothing for it to match against, so no duplicate gets flagged.

Fix the H1 on Page A, run the audit again, and now the crawler reaches Page B for the first time. Suddenly it matches something else on the site (maybe a page you never touched), and a “new” duplicate pair shows up. It’s not new; it just hadn’t been compared against its match yet.

Other reasons errors aren’t surfaced in audits

• Crawl prioritization: many crawlers favor pages with fewer clicks from the homepage first. Fixing issues near the surface can shift what the crawler reaches deeper into the site on the next pass.
• Timeouts on large or slow sites: a crawl can get cut short before it’s actually finished, and the report still generates as if it were complete.
• Genuinely new problems: content published or updated between audits can introduce fresh duplicates that weren’t errors before.
• Settings scope: subfolder restrictions, parameter handling, or crawl depth settings can quietly exclude sections of the site without anyone noticing.

Takeaways for running and interpreting audits

One clean audit report doesn’t mean a clean site. It means a clean crawled sample. These takeaways will ensure you get a clear picture:

1. Check crawled-page count against your actual site size, not just the error count. If your sitemap says 40,000 URLs and the audit crawled 5,000, you’ve only seen 12% of the picture.
2. Increase crawl limits or run segmented audits (by subfolder or template type) if your plan or tool can’t cover the whole site in one pass.
3. Expect iterative discovery, not one-and-done fixes. Budget for multiple rounds of cleanup as the crawler reaches deeper into the site, especially on large or older domains.
4. Cross-reference with server logs or Google Search Console when possible. These tools show you what’s actually being crawled and indexed by search engines, independent of your audit tool’s crawl limits.

Technical SEO audits are critical to uncovering errors that could be holding your website back from ranking as well as it could for SEO and AEO. Don’t hesitate to bring up findings with your web partner if you hear from a company that’s uncovered errors, but beware that there’s often context behind the findings that requires discussion.