How to Run a Complete SEO Audit in 2026: 5-Phase Methodology and 40-Point Checklist

A complete SEO audit in 2026 looks nothing like the audit of 2020. Core Web Vitals, E-E-A-T signals, AI engine visibility (GEO), and indexation via Search Console now form an interconnected whole that you cannot audit in silos. This five-phase methodology gives you a reproducible framework, whether you are working on a small niche site or a multi-domain architecture.

Why Structure the Audit in Phases

A flat checklist (200 items to tick off) creates the illusion of thoroughness but does not tell you in what order to act or how to prioritise fixes. A phased approach allows prioritisation: technical blockers (phase 1) before content optimisations (phase 2), which themselves precede authority adjustments (phase 3).

The practical rule: if Google cannot crawl and index correctly, nothing else matters. Resolve access blockers first, then content, then links.

Phase 1 — Technical Audit: Access, Crawl and Performance

The technical audit verifies that Google can access your pages, crawl them efficiently, and render them quickly. This is the foundation.

Crawlability and Indexation

Run a full crawl with Screaming Frog, Sitebulb or equivalent. Check:

• Noindex pages: are they intentional? Staging pages, admin pages, internal search result duplicates should be noindex — but not your category pages or pillar articles.
• Robots.txt: does it accidentally block important directories? Check the Disallow: lines against your high-value URLs.
• Canonical tags: does every page have a correct canonical? Paginated pages, e-commerce filters, and UTM parameters generate duplicates if the canonical is absent or mispositioned.
• Crawl budget: for sites over 10,000 pages, check server logs to identify wasteful URLs Googlebot is crawling (redirect loops, infinite facets, session parameters).

In Google Search Console, open the Indexing > Pages report and classify exclusions: "Alternative page with proper canonical tag", "Duplicate without user-selected canonical", "Crawled – currently not indexed". The last status, when it hits high-value pages, is a strong signal of thin content or insufficient quality.

Core Web Vitals and Performance

CWV have been a ranking factor since 2021. In 2026, Google's tolerance for poor scores has decreased, particularly on mobile.

Target metrics:

• LCP (Largest Contentful Paint): < 2.5 s
• INP (Interaction to Next Paint): < 200 ms (replaced FID since March 2024)
• CLS (Cumulative Layout Shift): < 0.1

Tools: PageSpeed Insights for lab data, Search Console > Experience > Core Web Vitals for field data (real, CrUX-based). Prioritise URLs in "Poor" status in field data — these are the only ones that negatively impact ranking.

Common causes of high LCP: uncompressed images, poorly configured lazy loading, high server TTFB, render-blocking fonts. Causes of CLS: images without defined dimensions, dynamically injected ads, cookie banners pushing content down.

HTTPS, Redirects and URL Structure

• Are all pages on HTTPS? Check for mixed content resources (HTTP images on HTTPS pages).
• Are there redirect chains? A → B → C → D instead of A → D directly. Each extra hop dilutes PageRank and wastes crawl budget.
• Are URLs clean, short and descriptive? Avoid dynamic parameters in the index without proper canonicalisation.

Structured Data

Rich snippets (FAQ, Article, Product, BreadcrumbList) depend on well-implemented schemas. Test with Google's Rich Results Test tool. In 2026, Article schemas with author, datePublished, dateModified fields and FAQ schemas are the most directly impactful for content sites.

Phase 2 — Content Audit: Quality, Relevance and E-E-A-T

The content audit answers a simple question: do your pages deserve to rank well? Google evaluates the usefulness of a page for the user, not its conformity to a keyword grid.

Thin Content and Low-Value Pages

A crawl reveals short pages (< 300 words excluding boilerplate), but length is not the only criterion. Ask: does this page answer the search intent better than the current positions 1–3?

Actions:

• Consolidate: two articles on the same topic with low traffic → merge into one pillar article + 301 redirect.
• Enrich: page ranking at position 8–20 with traffic but low CTR → deepen the content, add fresh data, rework the title and meta description.
• Deindex: pages with no SEO value (empty tag archives, internal search result pages) → noindex.

Keyword Cannibalisation

Cannibalisation occurs when multiple pages target the same query and compete against each other. Search Console reveals the problem: if two URLs alternate in position for the same query over 90 days, there is likely cannibalisation.

Resolution: identify the master page (the one with the best traffic history and best backlinks), merge content from the other into the master, redirect the cannibalising URL with a 301.

E-E-A-T Signals (Experience, Expertise, Authoritativeness, Trustworthiness)

Since the Helpful Content updates and quality rater guidelines, E-E-A-T signals influence ranking for YMYL sites (health, finance, law, news) and increasingly for all content sites.

E-E-A-T checklist:

• About and team pages: is the author identifiable? Is there a link to LinkedIn profiles or external publications?
• Author pages: does each author have a dedicated page with their bio and expertise?
• Sources and citations: are factual claims sourced to credible studies or publications?
• Update dates: are articles timestamped? Are old articles on fast-moving topics refreshed?
• Transparency: legal notice, privacy policy, editorial policy accessible from the footer?

Content Freshness

Google favours fresh content for time-sensitive queries (news, prices, legislation, rankings). Audit your articles published more than 12 months ago on topics that evolve quickly. Strategy: update the content, modify dateModified in the Article schema, and resubmit via Search Console.

Phase 3 — Authority Audit: Backlinks and Internal Linking

Backlink Profile

Use Ahrefs, Semrush or Majestic to extract your full backlink profile. Analyse:

• Referring domain distribution: 100 links from 100 different domains > 100 links from 1 domain. Diversity matters.
• Domain quality: a link from a site with strong topical authority is worth more than 50 links from generic directories.
• Toxic links: obvious PBN domains, spam sites, over-optimised exact-match anchor links. Create a disavow file if you notice an algorithmic or manual penalty linked to links.
• Lost links: referring domains that deleted their links are often recoverable via a simple email or URL update.

Internal Linking

Internal linking distributes PageRank between your pages and helps Google understand your site's topical structure.

Internal linking audit:

• Do your most important pages (pillar articles, conversion pages) receive internal links from other pages?
• Are there orphan pages (no internal inbound links) visible in Search Console?
• Are internal link anchors descriptive (link text = topic of target page) or generic ("click here")?
• Is click depth reasonable? An important page more than 4 clicks from the homepage is difficult to crawl and hard to build authority for.

Phase 4 — Search Intent Audit

A technically perfect article that answers the wrong search intent will not rank. Search intent is the real reason someone types a query into Google.

The four intent types:

• Informational: the user seeks an answer ("how to do X")
• Navigational: the user seeks a specific site ("Flipagora login")
• Commercial: the user is comparing options before buying ("best hybrid inverter")
• Transactional: the user is ready to buy ("buy 6kWp solar kit")

For each page, open a private window and type the target query into Google. Analyse the SERP:

• What format do the top 3 results use (long article, list, video, product page)?
• What is the approximate length of the top 3 pages?
• Is there a Featured Snippet, a "People Also Ask" section?

If your page format does not match what the SERP displays, that is an intent mismatch signal — fix it before optimising content.

Phase 5 — GEO Audit: Visibility in AI Engines

In 2026, a growing share of searches begins on ChatGPT, Perplexity, Google AI Overviews, Copilot or Claude. The GEO (Generative Engine Optimization) audit checks whether your site is cited in these responses.

AI Crawler Accessibility

Verify that your robots.txt and meta robots do not accidentally exclude major AI bots:

• GPTBot (OpenAI)
• ClaudeBot (Anthropic)
• PerplexityBot (Perplexity)
• Googlebot (Google AI Overviews)
• Bingbot (Copilot/Bing)

A User-agent: GPTBot / Disallow: / in robots.txt makes you invisible to ChatGPT. This is sometimes intentional — but often it is inherited from an overly restrictive robots.txt template.

llms.txt File

The emerging llms.txt standard (proposed by Jeremy Howard, 2024) allows your site to indicate to language models which sections of your content are authorised for training and citation. Created at your domain root (/llms.txt), it lists your priority URLs and usage conditions.

It is not yet an official standard, but major AI bots are beginning to read it. For high-value content sites, it is an early authority signal worth implementing.

Citation Testing

Test your site in ChatGPT, Perplexity and Google AI Overviews on your target queries. If you do not appear while competitors are cited:

• Are your pages accessible to AI crawlers?
• Does your content directly answer the question (Q&A format, structured lists, concrete data)?
• Do you have sufficient external presence (Wikipedia, studies, press) to be perceived as an authoritative source?

40-Point Summary Checklist

Technical

• Full crawl completed, URLs analysed
• Search Console "Pages" report audited (exclusions classified)
• Robots.txt verified (no accidental Disallow)
• Correct canonical on all indexed pages
• LCP < 2.5 s (CrUX field data)
• INP < 200 ms (CrUX field data)
• CLS < 0.1 (CrUX field data)
• Full HTTPS, no mixed content resources
• No redirect chains > 1 hop
• JSON-LD schemas validated (Article, FAQ, BreadcrumbList)
• XML sitemap submitted and up to date in Search Console
• Crawl budget analysed (server logs) for large sites

Content

• Thin content pages identified and treated (enrichment or deindexation)
• Cannibalisation identified and resolved (merge + 301)
• About page and author pages in place
• Sources cited for factual claims
• Publication and update dates present
• Legal notice and privacy policy accessible from footer
• Old articles on evolving topics scheduled for update
• Meta descriptions optimised (< 160 chars, compelling)
• Title tags optimised (< 60 chars, primary query included)

Authority

• Backlink profile exported and analysed
• Toxic links identified, disavow file updated if necessary
• Orphan pages identified and linked from relevant pages
• Pillar pages receive internal links from cluster articles
• Internal link anchors are descriptive
• Click depth < 4 clicks for important pages
• Recoverable lost links identified

Search Intent

• SERP analysed for each strategic page (format, length, intent)
• Content format aligned with dominant SERP
• Featured Snippet targeted on informational queries
• People Also Ask integrated into content where relevant

GEO / AI

• GPTBot, ClaudeBot, PerplexityBot not blocked in robots.txt
• Meta robots verified (no noindex on value pages)
• llms.txt file created or planned
• Site tested in ChatGPT, Perplexity and Google AI Overviews
• Complete structured data for AI-preferred formats (Q&A, lists, numbered steps)
• External presence (press, reference directories, Wikipedia if relevant) audited

How Often Should You Run an SEO Audit?

A complete 5-phase audit represents 8 to 20 hours of work depending on site size. Here is the recommended frequency by context:

Continuous monitoring (weekly Search Console, CWV alerts, position tracking) complements periodic audits. The full audit remains irreplaceable for in-depth diagnostics.