Skip to main content
METRIC VAULT
← All posts

When a Site Audit Flags a Problem You Should Ignore

Not every red flag in your crawler deserves a fix. Here's how to tell the warnings that cost rankings from the ones that just burn client hours.

·

An analyst sits overwhelmed at a cluttered desk, hand on forehead, while a computer monitor glows with a red list of warning items.

Skip site audit warnings that don't touch crawling, indexing, or rankings: missing meta keywords, low word count on utility pages, multiple H1s, underscores in old URLs, text-to-HTML ratio, and alt text on decorative images. Handle indexing issues first, then real ranking signals, and write down what you skipped so clients understand why it isn't worth the hours.

Every crawler loves a long list. Open Screaming Frog, Ahrefs, or Semrush and you'll get hundreds of flagged items, color-coded in alarming red. The urge to fix all of them is strong, especially when a client is watching the health score. But plenty of what an audit surfaces has no measurable effect on traffic, and chasing it is one of the fastest ways to burn a retainer on busywork.

This guide sorts the warnings that move rankings from the ones you can close without touching code. The point isn't laziness. It's spending your limited hours where they actually change outcomes.

Why a site audit over-reports problems

Audit tools are built to be exhaustive, not prioritized. They flag anything that could theoretically matter, because missing a real problem looks worse than crying wolf. That design choice is fine, right up until a junior SEO or an anxious client treats every item as a to-do.

The health score at the top of most reports makes this worse. It's a weighted average with no link to revenue or rankings. A site can sit at 72/100 and dominate its niche, while a thin affiliate site scores 98 and ranks for nothing. Treat the score as a rough signal, never a target.

What separates a senior technical SEO from a checklist-runner is knowing what to ignore. That judgment is what clients are actually paying for.


Warnings you can almost always ignore

These turn up in nearly every site audit and rarely deserve a line in your deliverable.

Meta keywords are missing

Google dropped the meta keywords tag in 2009. Bing ignores it too. If your tool flags "missing meta keywords," close it and move on. Adding them does nothing except hand competitors a list of your targets if they view source.

Low word count on pages that don't need words

Crawlers flag any page under some arbitrary threshold, often 300 words. But a contact page, a login screen, a category page, or a well-built product page can be perfect at 80 words. Word count isn't a ranking factor; satisfying intent is. A thin content page competing for an informational query is a real problem. A thin utility page is not.

Multiple H1 tags

The old rule was "one H1 per page." Modern HTML5 allows multiple sectioning H1s, and Google's John Mueller has repeatedly confirmed it doesn't matter for ranking. Unless your heading structure genuinely confuses users or screen readers, a second H1 isn't worth a ticket.

"Low text-to-HTML ratio"

This metric is a relic. A heavy framework or inline CSS can tank the ratio while the page loads fast and ranks fine. Judge pages on actual Core Web Vitals and render performance, not on a ratio that correlates with nothing.

URLs containing underscores

Google reads hyphens as word separators and underscores as joiners, so technically blue_widget reads as one token. In practice it's a non-issue for existing URLs. Changing them means redirects, risk to link equity, and hours of QA to fix something that costs you nothing. Use hyphens on new URLs; leave old ones alone.

Images missing alt text on decorative elements

Alt text matters for accessibility and image search on content images. But spacer gifs, background decorations, and icon fonts don't need descriptive alt; an empty alt="" is correct for them. Tools flag the empty attribute anyway. Don't write "decorative divider line" alt text to satisfy a crawler.


Warnings that are context-dependent

These aren't automatic ignores, but they get over-fixed all the time. Investigate before acting.

Duplicate title tags

Real duplication across competing content pages is worth fixing. But paginated series, filtered views, and faceted navigation often carry near-identical titles by design, and are usually handled correctly with canonicals or noindex. Check whether the "duplicates" are pages you even want indexed before rewriting anything.

Redirect chains

A two-hop chain (A → B → C) is worth flattening when it sits on a high-traffic page or a frequently crawled template. A single extra hop on a page nobody visits isn't a priority. Fix chains in bulk during a migration, not one at a time as emergencies.

"Too many" internal or outbound links

The old 100-links-per-page guideline was a crawling constraint from a different era. A navigation-heavy page or a thorough resource list can carry hundreds of links without harm. Only step in if link equity is being diluted away from pages you care about.

Pages blocked by robots.txt

A crawler flags these as "not indexable," implying a problem. Often it's deliberate; you meant to block the admin area, the cart, or staging. Verify intent before you "fix" a block that's doing its job.

WarningDefault actionWhen it actually matters
Missing meta keywordsIgnoreNever
Low word countIgnoreThin content page targeting a real query
Multiple H1sIgnoreHeading order confuses users/screen readers
Underscores in URLsIgnore for old URLsSetting conventions for new URLs
Duplicate titlesInvestigateCompeting indexable content pages
Redirect chainsInvestigateHigh-traffic or heavily crawled pages
Blocked by robots.txtInvestigateYou blocked something you wanted indexed

How to set real SEO audit priorities

Once you've cleared the noise, rank what's left by impact, not by how red it looks. Good seo audit priorities follow a simple logic: fix what affects indexing first, then what affects rankings, then what affects efficiency.

Tier 1 — Indexing and access. If Google can't crawl or index a page, nothing else matters. Look for accidental noindex tags, broken canonicals pointing to the wrong URL, robots.txt blocks on important sections, and 5xx server errors on live pages. These are rare but devastating.

Tier 2 — Ranking signals that actually move. Slow Core Web Vitals on key templates, missing or duplicated canonical logic on money pages, broken internal links to important content, and genuinely thin or duplicated content competing for the same query. This is where most real technical SEO wins live.

Tier 3 — Hygiene and efficiency. Minor redirect chains, image compression, orphan pages with no business value, and crawl-budget tidying. Worth doing eventually, never worth a fire drill.

A useful filter: ask "if I fix this, which URL ranks better or gets indexed, and roughly how much traffic could that URL pull?" If you can't answer, it's probably not a priority.


How to communicate ignored items to clients

The hardest part isn't deciding what to ignore. It's explaining to a client why you're not fixing a page of red warnings they can see in the shared report.

Get ahead of it. In your audit deliverable, add a short section called something like "Flagged but not actionable." List the categories you're deliberately skipping and one sentence on why. This does two things: it proves you reviewed everything, and it heads off the inevitable "but the tool says there are 400 problems" conversation.

Here's a template line you can reuse:

The audit flagged 312 items. Of these, 240 fall into categories
that have no measurable effect on indexing or ranking (missing
meta keywords, decorative-image alt text, text-to-HTML ratio,
legacy URL formatting). We are not actioning these. The
remaining 72 are prioritized below by traffic impact.

Framed this way, ignoring warnings stops looking like cutting corners and starts looking like expertise, which is exactly what it is.


The cost of fixing everything

Chasing every flag has a real downside beyond the wasted hours. Large-scale "fixes" bring risk: rewriting title tags site-wide can tank CTR if you get them wrong, mass URL changes can orphan pages or break redirects, and aggressive canonical changes can deindex pages you wanted to keep. Every edit is a chance to break something that currently works.

The second cost is opportunity. Hours spent writing alt text for decorative images are hours not spent on the content gap that would actually win a competitive keyword. A site audit should free up time for strategic work, not spawn an endless backlog of cosmetic tickets.

A quick pre-flight checklist

Before you add any flagged item to your action plan, run it through four questions:

  • Does fixing this change whether a page gets crawled or indexed?
  • Does it affect a page with real or potential traffic?
  • Is the "problem" actually intentional (a block, a canonical, a thin utility page)?
  • Does the effort match the likely upside?

If an item fails all four, close it. Note that you reviewed it, and spend the recovered hours on Tier 1 and Tier 2 work.

The takeaway

A site audit is a diagnostic tool, not a to-do list. Its job is to surface everything; your job is to filter. The best technical SEOs stand out less for how many issues they fix and more for how confidently they ignore the ones that don't matter. Lead with indexing, follow with ranking signals, and let the cosmetic warnings sit in red forever. Your rankings won't notice, and your clients' budgets will thank you.