Audit tools deduct points per instance, which produces a scoring problem nobody warns you about. Nine hundred images missing alt text can pull a score down more than a single noindex tag on your main service page, even though one costs you nothing measurable and the other removes you from search results entirely.
So a low score tells you something is wrong, but not whether anything serious is wrong. Sorting the findings by actual severity is manual work, and it is worth doing before anyone starts fixing things in the order the report happened to list them.
Here is how the common findings rank by what they actually cost.
Severe: Problems That Suppress Pages Entirely
These are rare; they are usually invisible from the front end, and any one of them can make the rest of your SEO work irrelevant.
Noindex tags left in place.
The usual origin is a staging site that got pushed live with its directives intact, or a developer who blocked a section during a rebuild and never removed it. The page loads perfectly for visitors. Google simply will not include it. This is the highest-severity finding an audit can return, and it often sits on a template rather than a single page, which means it affects everything using that template.
Robots.txt disallow rules.
Same effect from a different direction. A line blocking a directory that made sense two years ago now blocks the blog, or the product catalog, or occasionally the entire site. Worth reading the file line by line rather than trusting a summary, because the one damaging rule tends to sit among several reasonable ones.
Canonical tags pointing somewhere wrong.
When every page in a template canonicals to the homepage, you have told Google that none of those pages is worth indexing on its own. This happens surprisingly often after a CMS migration,n and it is hard to spot without looking at the source, since nothing about the page appears broken.
Server errors at scale.
Intermittent 500 responses, timeouts under load, or a section returning errors only to crawlers. Check your server logs rather than relying on a crawl, because a tool that visits once may catch a good moment.
Content that only exists after JavaScript runs.
Google renders JavaScript, but not reliably or instantly, and sites that build their main content client-side often get indexed with the shell and none of the substance. If a page’s text does not appear in the raw HTML source, that is worth investigating.
Certificate and mixed content failures.
An expired certificate produces a browser warning that stops visitors cold. Mixed content, where a secure page loads insecure resources, is less dramatic and still worth clearing.
Moderate: Problems Whose Cost Scales With Volume
None of these will remove you from search. All of them drag on performance, and the damage scales with how many pages carry the issue.
Redirect chains.
After two or three site migrations, a URL can bounce through four hops before landing. Each one loses a little, and internal links still pointing at the original URL make it worse. Update the links rather than just the redirects.
Core Web Vitals failures.
Largest Contentful Paint is usually the first to check, and the cause is usually images. Unoptimized hero images, missing dimensions causing layout shift, render-blocking scripts in the head. These are real, and they are also frequently overweighted in audit scores relative to what they cost in rankings.
Duplicate or missing title tags across a template.
Forty pages titled with the same phrase compete against each other and give Google nothing to differentiate them. This is a scale problem with a scale fix, usually a template change rather than forty manual edits.
Broken internal links and 404s.
A handful is normal. A few hundred means something structural happened, like a URL pattern change nobody propagated.
Mobile rendering failures.
Content wider than the viewport, tap targets crowded together, text that requires zooming. Google indexes the mobile version, so this is not a secondary concern.
Sitemaps full of URLs that should not be there.
Redirects, 404s, and noindexed pages listed in the sitemap send mixed signals and waste crawl budget on larger sites.
Orphan pages.
Pages nothing links to are hard for Google to find and are telling you, by their existence, that your internal linking has gaps.
Cosmetic: Findings That Cost Points and Little Else
These fill out the bottom of most reports and generate a disproportionate share of the busywork.
Alt text on decorative images is more of an accessibility consideration than a ranking one, and a background pattern doesn’t need a description. Multiple H1 tags on a page have not been a real problem for years. Meta keywords are ignored entirely. Underscores versus hyphens in URLs is a preference, not a rule, and rewriting established URLs to fix it causes more disruption than it resolves. Keyword density warnings are based on a model of how search engines work that has not been accurate in a long time.
Fix these when you are already editing the page. Do not build a project around them.
Why the Score Gets This Wrong
Most tools deduct by count rather than by consequence, which means a site with one catastrophic issue and clean everything else can score better than a site with no serious problems and many small ones.
Different tools also weight the same finding differently, so the same site can score 58 in one and 79 in another. Neither number is wrong. They measure against different internal checklists, and neither is failing to meet an external standard
This is why the score works as a tracking line for one site over time and poorly as a comparison between two sites, or as a target in itself.
How to Work Through the List
Run the site, then resort the output before assigning anything.
Everything in the severe category gets handled this week regardless of how few pages it touches, because those findings cap what all your other work can achieve. The moderate category gets prioritized by how many pages carry the issue and how commercially important those pages are, which means the fix affecting six service pages usually outranks the one affecting two hundred blog posts. The cosmetic category goes on a list for whenever someone is in those templates anyway.
The part worth building into your process is verification. Crawlers produce false positives on JavaScript-heavy sites, staging environments, and configurations that look like errors without context. Confirm severe findings by hand before anyone touches production, because an SEO report card tool will find the problem quickly and cannot tell you whether it was intentional.
