How do I know your audit is actually right?

Here is how crawl data goes wrong, so you know what to look for:

• Rate limiting reads as catastrophe. A crawler getting 429s reports pages as broken that are fine. I once had a crawl report a median of zero internal links across 192 product pages. The real cause was the theme rendering collection grids in JavaScript plus the crawler getting throttled. Every finding about internal links was an artifact.
• JavaScript rendering. If the crawler doesn't execute scripts, half your content is invisible to it and none of it is invisible to Google.
• Coverage percentages hide which pages. Ninety percent schema coverage sounds fine until the missing ten percent is your most expensive product.
• Staging and parameters. A crawler that wanders into faceted navigation will report tens of thousands of thin pages that no human will ever see.

What a defensible audit looks like

Findings dated. Tool named. The exact URLs listed, not just a count. A sentence on what was checked by hand versus what the tool asserted. And a flag process, so when you say "that one's wrong", it gets re-checked rather than defended.

This site publishes its own audit for the same reason — zero errors, every warning listed, including the two I'm not fixing and why.