Audit SEO by Evidence, Severity, Ownership, and Validation

The phrase "SEO audit" can describe anything from an automated scan to a detailed investigation. For this guide, an audit is a diagnostic process: it gathers evidence, identifies the likely constraint behind underperformance, assigns severity, names the person or team responsible for correction, and defines how the result will be validated.

That is different from an implementation list. The existing SEO mistakes resource can help you recognize common failure patterns, while the existing SEO checklist is useful for checking whether expected elements are present. An audit has a different job: explain what is actually happening on this site and what evidence supports the diagnosis.

Use four diagnostic layers in dependency order:

• Technical health. Evidence comes from crawl behavior, index status, canonical signals, response codes, rendering, and performance data. Severity depends on whether the issue blocks discovery, indexing, consolidation, or important user access. Owners are usually engineering, platform, or technical SEO. Corrective action should address the confirmed cause, and validation should re-crawl or re-test the affected URLs and confirm the intended search state.
• On-page relevance and content mapping. Evidence comes from query data, page content, search intent, duplication, internal links, and competing URLs. Severity depends on the commercial importance of the affected intent and whether pages conflict with one another. Owners are typically SEO, content, product marketing, or editorial. Corrective action may involve consolidation, rewriting, remapping, or internal-link changes, followed by validation in the page set and search performance data.
• Authority evidence. Evidence comes from referring domains, link context, lost links, competitor comparisons, and any known manual or security concerns. Severity should reflect the pattern and its likely effect rather than a tool score alone. Ownership may sit with SEO, digital PR, communications, or site security. Corrective action should match the evidence, and validation should confirm that the intended links, removals, or acquisition changes actually occurred.
• Root-cause synthesis. Evidence is the interaction among the earlier layers. Severity reflects business impact and the breadth of affected pages or queries. Ownership belongs to whoever coordinates the roadmap. Corrective action is the prioritized sequence of fixes, and validation is a re-check against the original hypothesis rather than a generic traffic expectation.

The order matters because later observations can be misleading when an earlier dependency is unresolved. A page cannot demonstrate the value of a content rewrite if it is not being indexed as intended. Likewise, a link-building recommendation is premature if important pages are competing with one another or are difficult for search engines to discover.

The audit is complete only when each important finding can answer five questions in plain language: what evidence shows the problem, how severe is it, who owns the fix, what action should be taken, and what test will confirm that the action worked. If one of those answers is missing, the item is still a hypothesis or task idea rather than a finished diagnostic finding.

Begin with first-party search evidence where it is available, then use a crawler and page-level tests to reproduce what the site is exposing. The goal is not to collect every warning. The goal is to determine whether important URLs can be discovered, fetched, rendered, indexed, and consolidated as intended.

Index and Coverage Evidence

Review the indexing information in Google Search Console and compare it with the URLs your site intends to make searchable. Investigate categories that contain important pages, especially when discovery, crawling, canonicalization, or index selection differs from the intended state. Do not assume that every excluded URL is an error; exclusions can be correct when the page is a duplicate, redirect, parameter variant, or intentionally non-indexable. If you need contextual benchmark material while reviewing the scale of a finding, use the existing SEO statistics resource without treating an aggregate benchmark as proof that your site has a problem.

Severity: critical when high-value pages are blocked, incorrectly canonicalized, or persistently absent from the index for a confirmed technical reason; lower when the state is intentional or affects non-essential variants. Owner: technical SEO plus the developer or platform owner. Corrective action: fix the specific directive, response, canonical, link, or rendering cause. Validation: re-test the affected URL set and confirm that the intended index state is visible after recrawl or reprocessing.

Core Web Vitals and Page Experience Evidence

Use Search Console field data where available and page-level diagnostics to identify templates or page groups with performance problems. Treat performance as one part of the user and search experience rather than as a universal explanation for weak rankings. Record which templates are affected, whether the problem is reproducible, and whether the affected pages matter commercially.

Severity: based on the breadth of affected templates and the user impact, not the presence of a warning alone. Owner: front-end engineering, platform, or performance specialists. Corrective action: address the measured bottleneck, such as render-blocking work, layout instability, oversized assets, or server delay. Validation: re-test the same templates using the same evidence source and compare the relevant field or lab measurements.

Crawl Simulation

Run a representative crawl with a tool such as Screaming Frog; its free mode can cover up to 500 URLs. Review broken internal references, redirect chains, canonical conflicts, robots directives, duplicate metadata, orphan-like pages that lack discoverable internal paths, and unexpected status codes. A 4xx response is only severe when the requested URL should resolve or valuable internal links still point to it.

Severity: highest for patterns that affect many important pages or create discovery and consolidation errors. Owner: technical SEO for diagnosis and engineering or content operations for implementation. Corrective action: repair internal destinations, simplify redirects where justified, correct directives, and align canonical targets. Validation: repeat the crawl with the same scope and confirm that the affected URLs now return the intended state.

Server and Page-Speed Evidence

Test representative high-value templates with Google PageSpeed Insights and supporting browser or server diagnostics. A previously published example used TTFB above 600ms as a prompt to investigate hosting, caching, application work, or upstream latency; treat that figure as an operational trigger that still requires context rather than a universal pass-fail rule.

Severity: determined by repeatability, user impact, and template coverage. Owner: infrastructure, platform, or front-end engineering depending on the cause. Corrective action: fix the measured bottleneck rather than applying generic speed work. Validation: re-test the same pages under comparable conditions and confirm that the targeted bottleneck changed without introducing regressions.

Once important pages are technically available to search engines, test whether each page has a clear job. This stage should answer which intent the page serves, whether another URL competes for the same intent, whether the content satisfies the query context, and whether internal navigation helps search engines and users find the page.

Keyword-to-Page Mapping Evidence

Start with the top 20 pages by organic impressions or another business-relevant page set from Search Console. For each page, review the queries associated with it and compare those queries with the page's intended purpose. Ask whether the ranking URL is the one the business wants to own that intent and whether another internal URL is appearing for the same query family.

Severity: high when commercially important queries alternate among competing pages, the wrong URL ranks, or several pages are materially duplicating the same purpose. Owner: SEO with editorial or product stakeholders. Corrective action: clarify intent ownership through consolidation, differentiation, canonical or redirect decisions where appropriate, and internal-link changes. Validation: re-check the competing page set and query-to-page relationship after the changes have been processed.

Title and Header Evidence

Review whether the title and visible heading structure accurately describe the page and help a reader understand the subject. A previously published operating check used 60 characters as a title review threshold, but truncation and display behavior vary, so treat that figure as a review prompt rather than a guarantee. The H1 and H2 hierarchy should reflect the content structure rather than repeat mechanically or force keyword variants.

Severity: usually moderate unless titles are duplicated at scale, materially misleading, or preventing pages from being distinguished. Owner: SEO and editorial. Corrective action: rewrite titles and headings to match the page's real intent and distinguish it from nearby pages. Validation: crawl the affected set again and manually review representative pages to confirm uniqueness, accuracy, and readable hierarchy.

Content Coverage Evidence

Compare the page with the questions, subtopics, examples, and decision information present in the current search results and in your own customer knowledge. Word count is not a ranking target. If a page contains 400 words while a competing page contains 1,200 or more, that difference is only a prompt to inspect missing coverage; it is not proof that adding words will improve performance.

Severity: based on whether important user questions or decision information are genuinely absent. Owner: content, subject matter experts, and SEO. Corrective action: add, remove, reorganize, or consolidate content to satisfy the intended query more completely. Validation: review the revised page against the original content-gap evidence and monitor whether the intended query set is associating with the correct URL.

Internal Linking Evidence

Use a crawl or internal-link report to identify important pages with weak contextual support, excessive depth, broken references, or links from irrelevant contexts. Do not use an arbitrary internal-link count as a universal target. The question is whether important pages are discoverable and supported by relevant navigation and contextual paths.

Severity: high when important pages are effectively isolated or reachable only through weak paths. Owner: SEO, content operations, or engineering depending on where the links are generated. Corrective action: add or repair relevant internal paths and remove misleading destinations. Validation: re-crawl the site and confirm that the target pages now have appropriate discoverable paths from relevant sections.

Backlink analysis should determine whether external references support the pages and topics you expect to compete on, whether important links have been lost, and whether there is a pattern that warrants remediation. It should not turn a third-party metric or an isolated low-quality link into a diagnosis by itself.

Baseline Authority Evidence

Use a backlink analysis platform to review referring domains, linked pages, anchor context, topical relevance, link acquisition patterns, and lost links. Focus on identifiable sources and pages rather than only a composite authority score. Ordinary sites often accumulate low-quality or irrelevant links without any deliberate action, so the audit should distinguish background noise from a pattern that creates a real concern.

Severity: high only when there is credible evidence of a harmful pattern, a major loss affecting important pages, or a clear competitive deficit relevant to the target topics. Owner: SEO, digital PR, communications, or security depending on the finding. Corrective action: recover legitimate lost references, improve link-worthy assets, pursue relevant editorial mentions, or investigate suspicious patterns where evidence justifies it. Validation: confirm the change in the referring sources and the pages receiving support rather than relying solely on a vendor score.

Competitor Gap Evidence

Compare competitors that genuinely compete for the same search intents, not simply large brands in the same industry. Review which pages earn relevant references, what kinds of publications or organizations cite them, and whether your site lacks comparable evidence of usefulness or authority.

Severity: higher when the gap appears consistently across the exact query set you are trying to win and lower when it exists only at the domain level. Owner: SEO, content, and digital PR. Corrective action: identify assets or expertise that merit citation and build outreach around those specific opportunities. Validation: verify newly earned relevant references and whether the target pages are receiving stronger external support.

Potentially Harmful Link Patterns

Investigate links that appear manipulative, compromised, or part of a deliberate scheme, but do not prescribe a disavow action simply because a tool labels a link as toxic. Review source quality, pattern, intent, and any known manual-action context before escalating.

Severity: depends on the strength of the evidence and whether there is a repeated pattern. Owner: senior SEO with legal, security, or communications input when needed. Corrective action: document the pattern, remove or neutralize links where appropriate, and use Google's available remediation mechanisms only when the evidence supports doing so. Validation: confirm that the targeted links or patterns changed and retain a record of what was reviewed and why.

The synthesis stage turns individual observations into an ordered remediation plan. This is where you decide whether several warnings share one cause, whether a visible symptom is actually secondary, and which correction should happen first.

For every important issue, write an evidence statement before the recommendation. Then assign severity, an owner, the corrective action, and the validation step. This prevents generic recommendations from entering the roadmap without a reason.

How to Connect Symptoms to Causes

• Correct page ranks poorly for a valuable query: compare intent match, competing internal URLs, content coverage, internal support, and external authority before deciding which layer is constraining the page.
• Wrong page ranks: inspect duplication, internal anchors, canonical signals, redirects, page purpose, and query overlap before rewriting content indiscriminately.
• Traffic changes after a migration: compare old and new URL mappings, redirects, canonicals, internal links, index states, templates, and lost content before attributing the change to a broad search update.
• New pages remain undiscovered or unindexed: verify crawl paths, directives, rendering, duplication, canonicalization, content purpose, and site architecture before concluding that authority is the only issue.
• Local visibility varies: separate genuine location and service-area differences from profile accuracy, website relevance, reviews, citations, and broader prominence evidence. Do not assume one local signal explains every market.

Severity: should combine business impact, affected scope, and confidence in the diagnosis. Owner: assign one accountable coordinator even when several teams must contribute. Corrective action: fix the highest-confidence upstream constraint before downstream symptoms. Validation: return to the original evidence and confirm whether the diagnosed state changed; if not, revise the hypothesis rather than simply repeating the same action.

Build the Remediation Roadmap

Group findings by dependency and implementation risk. High-impact items with clear evidence and low implementation risk usually move first, while architecture changes should wait until the diagnosis is strong enough to justify the wider blast radius. Record the affected URLs or templates, the evidence source, severity, owner, corrective action, validation test, dependencies, and status.

The output is not a list of everything that could be improved. It is a sequence of verified problems and testable corrections. That distinction makes the audit useful to developers, content teams, and decision-makers because each task has a reason and a way to determine whether it was completed successfully.

A self-audit is appropriate when the site is understandable, the evidence is accessible, and the person running the review can implement or coordinate the fixes. Escalation is useful when the evidence conflicts, the site architecture is complex, several layers fail at once, or the team cannot determine which issue is upstream.

Self-Audit Is Usually Manageable When

• The site has under 200 pages and a relatively simple architecture.
• The main issue can be reproduced in first-party search data or a crawl and has a clear owner.
• The team can make the correction, document it, and re-test it without introducing wider platform risk.
• The remediation can be monitored over a 3-4 month observation stage without the business depending on an immediate outcome.

In a self-audit, keep the same discipline used elsewhere in this guide. Evidence: save the Search Console export, crawl result, page test, or backlink evidence that supports the finding. Severity: rank the issue by business impact and affected scope. Owner: name the person who can actually make the change. Corrective action: describe the smallest action that addresses the confirmed cause. Validation: define the re-test before implementation so success does not become subjective afterward.

Professional Review Becomes More Useful When

• The team has implemented evidence-based corrections and the monitored condition still has not changed after 90 days.
• A migration, redesign, template change, or search event coincides with a broad decline and several possible causes overlap.
• The site contains over 500 pages, multiple systems, or URL patterns that make sampling and dependency analysis difficult.
• Technical, content, authority, and measurement findings interact in ways that make sequencing uncertain.
• The business needs an independent review of a high-risk implementation before engineering or editorial resources are committed.

A professional audit should still meet the same standard as a self-audit. It should show the evidence behind each material finding, state severity, assign or recommend an owner, describe the corrective action, and specify how the result will be validated. A larger toolset or a longer document does not replace those requirements.

If outside help is appropriate, use request a professional SEO audit from our team as the next step. The useful deliverable to request is a prioritized diagnostic roadmap that your team can evaluate and implement, not an unranked export of tool warnings.