How Do You Perform a Complete SEO Audit?

1. Define the audit scope, business goals and baseline

A complete audit begins with a decision, not a crawler. Identify what the site must accomplish, such as qualified leads, ecommerce revenue, subscriptions, store visits or reduced support demand. Record the markets, languages, devices, search engines and URL classes in scope. A migration audit, local business audit and million-page marketplace audit require different evidence.

Establish a baseline for organic clicks, impressions, conversions, revenue, indexed pages, referring domains and Core Web Vitals. Segment results by template, directory, country, device and branded versus nonbranded demand. Annotate migrations, redesigns, tracking changes, manual actions and major traffic losses so that correlation is not mistaken for causation.

Collect Google Search Console, Bing Webmaster Tools, analytics, conversion, backlink and content inventory data. For large or frequently updated sites, obtain raw server logs. Preserve the crawl configuration and export dates so the audit can be reproduced.

Required audit deliverables

• An executive summary connecting search defects to business outcomes.
• An issue register listing evidence, affected URLs or templates, impact, confidence, effort, risk and owner.
• A roadmap divided into immediate fixes, development work, content work and strategic experiments.
• Acceptance tests and monitoring rules for every material recommendation.

2. Diagnose the correct layer before prescribing a fix

Search visibility is a sequence. A URL must generally be discovered, crawled, rendered when necessary, indexed, selected for relevant searches and persuasive enough to earn a conversion. Failure at one layer can resemble failure at another. A page described as ‘not indexed’ may never have been discovered, while a ranking decline can originate in changed intent rather than a technical defect.

This framework prevents expensive category errors, such as buying links for a page blocked by noindex or rewriting content when analytics tracking is broken.

3. Audit crawling, rendering and indexation controls

Crawl the site as both a desktop and mobile search agent, then compare the crawl with XML sitemaps, analytics landing pages, Search Console exports and known conversion URLs. Inspect status codes, robots.txt, robots meta directives, X-Robots-Tag headers, canonicals, redirects, pagination, hreflang and rendered HTML. Validate samples from every important template manually.

XML sitemaps should contain indexable canonical URLs that return successful responses. Google treats sitemap entries and canonical tags as signals, not guarantees. Confirm that canonical targets are accessible, equivalent and internally linked. Remove redirecting, blocked, duplicate or error URLs from sitemaps.

Map redirect chains and loops. Use direct permanent server-side redirects for durable moves. Google states that 301 and 308 redirects do not cause PageRank loss, although careless destination mapping can still damage relevance and user experience. Avoid redirecting unrelated expired pages to a homepage.

Inspect rendered pages when navigation, main content, metadata or structured data depends on JavaScript. Compare raw HTML with rendered output and test without stored cookies. For large sites, use server logs to identify wasted crawling on faceted combinations, parameters, internal search results, expired inventory and repeated errors.

Google says dedicated crawl-budget analysis is mainly relevant to sites with more than one million pages, rapidly changing medium-sized sites, or substantial inventories marked ‘Discovered, currently not indexed.’ Smaller sites should usually fix architecture, duplication and quality before pursuing speculative crawl-budget optimization.

4. Evaluate architecture, internal links and topical coverage

Build a URL map by template, topic, intent and business value. Important pages should be reachable through crawlable HTML links and should not depend solely on search forms, scripts or XML sitemaps. Measure click depth, orphan pages, incoming internal links, anchor relevance and links pointing to redirects or noncanonical URLs.

Use hub-and-spoke architecture where it reflects genuine user journeys. A broad service or topic hub should link to focused supporting pages, while those pages should link back to the hub and to relevant adjacent resources. This creates explicit relationships among entities without forcing every page to link to every other page.

Map query fanout: the primary question, prerequisite questions, comparisons, implementation tasks, troubleshooting needs and purchase considerations. Then identify whether each intent deserves a distinct page, a section within an existing page or no new asset. Consolidate overlapping pages when they compete for the same intent and neither offers unique value.

Review navigation and internal anchors at the template level. Repeated footer links may technically create links but provide weaker contextual guidance than useful in-copy references. Prioritize links from authoritative, relevant pages to strategically important URLs. Preserve natural wording and avoid mechanically repeating exact-match anchors.

5. Audit content quality, intent alignment and SERP opportunity

Inventory every indexable page and classify it as keep, improve, consolidate, redirect, exclude or remove. Base the decision on search demand, intent match, organic performance, backlinks, conversions, accuracy, uniqueness and strategic value. Low traffic alone is not sufficient reason to delete a page that supports customers, links or a narrow but valuable query.

For priority queries, compare the current results by page type, format, entity coverage, freshness, evidence and SERP features. Inspect titles, headings and opening passages for clear answers. A useful snippet-ready passage defines the subject, gives the decision rule and supplies essential constraints in a few self-contained sentences.

Improve thin or decaying pages with original evidence, expert review, examples, comparison criteria, process details and updated facts. Do not inflate word count. Ahrefs found that 53.4 percent of sampled AI Overview citations came from pages under 1,000 words, which weakens the assumption that longer content inherently earns citations.

Look for opportunities to create natural link demand: original datasets, statistics pages, calculators, benchmark reports, comparison assets, visual explanations and expert contribution programs. Refresh assets when their underlying evidence changes. Preserve URLs when possible and document material updates.

Run controlled title tests only on comparable page groups and measure clicks alongside rankings, conversions and seasonality. A higher click-through rate with lower qualified conversions is not necessarily a win. Avoid frequent uncontrolled changes that make causal interpretation impossible.

6. Test structured data, page experience and conversion paths

Validate structured data against both syntax requirements and visible page content. Markup may make a page eligible for a rich result, but display is not guaranteed. Remove unsupported claims, fabricated reviews and schema for information users cannot see. Confirm that search engines can crawl the page and its required resources.

Assess Core Web Vitals with field data. At the 75th percentile, Google’s good thresholds are Largest Contentful Paint at 2.5 seconds or less, Interaction to Next Paint at 200 milliseconds or less and Cumulative Layout Shift at 0.1 or less. Chrome User Experience Report data represents real-user experiences on eligible popular destinations. Lighthouse is valuable for diagnosis, but a laboratory run is not a substitute for field evidence.

Segment performance by template, device and geography. Find shared causes such as oversized hero media, late-loading fonts, third-party scripts, consent tooling, unstable ad slots or long main-thread tasks. Fix the shared component before optimizing isolated URLs.

Complete the path from search result to conversion. Test forms, calls, checkout, store locators, account creation and analytics events on mobile and desktop. Confirm that thank-you pages and revenue events are not double counted. SEO recommendations cannot be valued when conversion measurement is unreliable.

7. Review authority, competitors, local signals and risk

Audit referring domains, relevant page-level links, anchor patterns, lost links and suspicious acquisition spikes. Compare competitors with a link-intersect analysis to find publications, associations, suppliers, resource pages and datasets that cite several peers but not the audited site. Reclaim broken links and convert accurate unlinked brand mentions through direct outreach.

Favor digital PR and assets that deserve references over purchased placements. Risky tactics include scaled guest-post networks, expired-domain manipulation, private link networks and aggressive exact-match anchors. They may produce temporary movement, but create detection, removal and reputational risk. Do not recommend hacked links, cloaking, doorway pages, hidden text, fake reviews or deceptive redirects.

For local organizations, verify business identity, categories, landing pages, location details, opening hours and review processes across owned properties and major platforms. Check whether each location page provides distinct, useful information. Do not create near-duplicate city pages for places where the business has no meaningful presence.

Authority should be evaluated relative to the topic and competing result set, not as a single third-party score. A relevant citation from an industry body or original research can be more defensible than many unrelated directory links.

8. Audit visibility in AI answers and answer engines

Google states that AI Overviews and AI Mode use the same foundational search requirements: a page must be indexed and eligible to appear with a snippet. There is no special AI schema or separate AI technical file required. Clear answers, accessible supporting evidence and strong topical relationships can help retrieval, but they do not guarantee selection.

Audit likely query rewrites rather than checking one head term. Test definitions, comparisons, follow-up questions, troubleshooting prompts and purchase-stage questions across Google, Bing or Copilot and ChatGPT where relevant. Record whether the brand is cited, linked, mentioned without a link, mischaracterized or absent.

Do not assume one platform represents all answer systems. Ahrefs analyzed 16.975 million cited URLs and found that 86 percent of top-mentioned sources were not shared across ChatGPT, Perplexity and Google AI Overviews. Its freshness findings also differed by platform. Measure each ecosystem separately and retain dated samples because interfaces and source selection change.

Bing recommends established SEO fundamentals for Bing and Copilot. IndexNow can accelerate notification of added, updated or deleted URLs, but it does not guarantee crawling, indexing or ranking. Use it as a discovery protocol, not a ranking tactic.

Track citation share, linked citations, referenced pages, answer accuracy, branded search demand and assisted conversions. Conventional rank tracking remains useful, but it cannot fully represent synthesized answers.

9. Prioritize fixes and turn findings into implementation

Score each issue using impact, confidence, effort and risk. Impact should consider affected demand, revenue, strategic value and the number of URLs sharing the defect. Confidence reflects evidence quality. Effort includes development, content, QA and coordination costs. Risk includes traffic loss, rendering regressions, legal exposure and reversibility.

1. Protect: Resolve accidental blocking, broken canonicals, migration errors, tracking failures and revenue-path defects.
2. Unlock: Fix template issues preventing valuable pages from being discovered, rendered or indexed.
3. Improve: Strengthen intent alignment, internal linking, page experience and conversion paths.
4. Expand: Build missing topic coverage, original assets and legitimate authority.
5. Experiment: Test titles, answer formats and emerging AI visibility opportunities with controlled samples.

Every ticket should state the problem, affected URLs or template, evidence, expected outcome, implementation owner and rollback plan. An acceptance test might require a successful status, self-referencing canonical, rendered main content, valid schema and inclusion in the correct sitemap. Test in staging where possible, but repeat validation in production because edge rules, caching and consent scripts can differ.

10. Monitor outcomes and distinguish evidence from uncertainty

After release, recrawl affected templates and inspect representative URLs. Monitor errors, indexed-page trends, organic clicks, conversions and field performance. Use shorter monitoring windows for blocking or migration defects and longer windows for content or authority work. Keep a change log so later gains or losses can be evaluated against deployments.

What is proven

Official documentation establishes that technical eligibility does not guarantee crawling, indexing or ranking. Canonicals and sitemap entries are signals rather than commands. Good Core Web Vitals thresholds use 75th-percentile field measurements. Structured data creates eligibility, not guaranteed rich-result display.

What reflects broad practitioner consensus

Experienced auditors generally combine tools, validate critical findings manually and prioritize template-level defects by business impact. Community discussions also repeatedly warn that large automated reports without interpretation are poor deliverables. These observations are useful but remain anecdotal rather than controlled evidence.

What remains uncertain

AI answer citation systems are volatile and platform-specific. Observational studies can reveal associations, but not a universal formula for citation. The incremental ranking effect of individual content changes is also difficult to isolate when competitors, result formats and search demand change simultaneously.

Run a focused technical check monthly on active sites, a broader quarterly review for most established sites and a full audit before migrations, redesigns or major platform changes. High-volume publishers and marketplaces may require continuous crawl, log and indexation monitoring.