How to do an SEO audit step by step: practical guide 2026

Step 1: Indexation and crawling

Goal: confirm that Google can access your important pages and is not indexing pages it should not.

1.1 Coverage in Search Console

Open Google Search Console > Indexing > Pages. Here you will see the status of all URLs Google has found or attempted to crawl:

• •Indexed: how many pages are in Google's index. Compare this with the actual number of pages on your site.
• •Not indexed: review the reasons. The most serious: "Excluded by 'noindex' tag", "Crawled – currently not indexed", "Page with redirect".
• •Errors: pages with server errors (5xx) or not found (4xx) that Google keeps attempting to crawl.

Warning signal: if your site has 200 pages and Google indexes 40, there is a structural problem (misconfigured canonical, site-wide noindex, outdated sitemap).

1.2 Robots.txt

Access yourdomain.com/robots.txt. Check:

• •Is it blocking entire sections it should not? A Disallow: / left over from a staging environment pushed to production is one of the most common and devastating errors.
• •Is the sitemap referenced?

1.3 XML sitemaps

Submit or review the sitemap in Search Console > Sitemaps. Verify:

• •The URLs in the sitemap correspond to real, indexable pages (no noindex pages, no redirects, no 4xx errors).
• •The sitemap is updated when you publish new content.

1.4 Canonical tags

With Screaming Frog, export the canonicals report. Look for:

• •Pages with no canonical declared.
• •Canonicals pointing to a different URL than the page itself (unintentional cross-canonical).
• •Paginated pages with a canonical to the first page where this is not intended.

1.5 Accidental noindex directives

Filter in Screaming Frog for "noindex". Are any important pages carrying this tag? In WordPress, the "Discourage search engines from indexing this site" setting under Settings > Reading applies a site-wide noindex and is accidentally activated more often than it should be.

Step 2: Technical SEO

Goal: ensure the site is fast, secure and accessible to both users and search engines.

2.1 Core Web Vitals and page speed

Google has used Core Web Vitals as a ranking signal since 2021. There are three metrics:

• •LCP (Largest Contentful Paint): time until the main element of the page is visible. Target: under 2.5 seconds.
• •INP (Interaction to Next Paint): response to the first user interaction. Target: under 200 ms. Replaced FID in 2024.
• •CLS (Cumulative Layout Shift): visual stability of the page as it loads. Target: under 0.1.

How to measure them: Google PageSpeed Insights (lab and real-field data) and the Core Web Vitals report in Search Console (real user data segmented by device).

The most common issues:

• •High LCP caused by unoptimised images or incorrect application of the loading="lazy" attribute.
• •CLS caused by web fonts that are not preloaded or ads with no reserved space.
• •High INP caused by third-party JavaScript (chat plugins, ad networks, marketing tools) blocking the main thread.

2.2 Mobile-friendliness

Google indexes in mobile-first mode: Googlebot analyses your site as if it were a smartphone. Check:

• •Responsive design that works on screens 375–430 px wide.
• •Buttons and links with a minimum tap target of 44 px.
• •No content hidden on mobile that exists on desktop (Google sees both equally).
• •No text smaller than 16 px.

Tool: Lighthouse (built into Chrome DevTools) and DevTools device mode let you audit the mobile version. Note: Google removed the Mobile Usability report from Search Console in late 2023, so use Lighthouse or the mobile-friendly test instead.

2.3 HTTPS

The entire site must be served over HTTPS with a valid certificate. Verify:

• •No mixed content (images, scripts or stylesheets served over HTTP on HTTPS pages).
• •The certificate is not expired.
• •HTTP to HTTPS redirects are direct 301s, not chained (HTTP → HTTPS → www.HTTPS).

2.4 URL structure

URLs must be short, descriptive and consistent:

• •No unnecessary parameters in indexable URLs (?session_id=, ?ref=).
• •No special characters, uppercase letters or spaces.
• •Logical structure reflecting the site hierarchy (/services/seo-audit/ is better than /p?id=234).
• •Consistency: choose between www and non-www, between trailing slash and no trailing slash, and do not mix them.

2.5 4xx and 5xx errors

Using Screaming Frog or the Search Console coverage report, list all errors:

• •404s with backlinks: deleted pages that still receive external links. These represent authority being lost, fixed by a 301 redirect to the most relevant equivalent page.
• •Internal 404s: pages on the site that link to broken URLs. Update the link or redirect the destination.
• •5xx errors: server errors. They typically indicate hosting or code issues that require urgent attention.

2.6 Redirects

Review redirect chains. A chain of 3 or more hops (A → B → C → D) slows crawling and dilutes authority. Always shorten to a single direct redirect (A → D).

Step 3: On-page SEO

Goal: ensure each page is optimised for the searches it should capture, with no conflicts between pages on the same site.

3.1 Title tags

The <title> is the on-page factor with the most direct influence on rankings. Audit:

• •Does each page have a unique title?
• •Does the title include the target keyword for that page, ideally towards the beginning?
• •Length: between 50 and 60 characters so it is not truncated in search results.
• •No empty titles or CMS default titles ("Home – Site Name").

In Screaming Frog: filter "Page Titles" > export > sort by length and check for duplicates.

3.2 Meta descriptions

Not a direct ranking factor, but it affects CTR (the click-through rate in search results). Google rewrites it when it is not relevant to the query, but a well-written meta improves SERP performance:

• •Ideal length: 140–155 characters.
• •Includes the keyword and an implicit call to action.
• •Unique for each page.

3.3 Headings (H1–H6)

• •Each page must have exactly one H1 that aligns thematically with the title (not necessarily identical, but consistent).
• •H2 and H3 tags structure the content and are where you place keyword variants and related terms.
• •Do not use headings for visual styling: they are semantic hierarchy, not formatting.

3.4 Search intent

This is the most important shift in modern SEO. Google does not rank the page with the most keywords — it ranks the page that best answers what the user wants to do.

There are four types of intent:

• •Informational: I want to learn something ("what is an SEO audit").
• •Navigational: I want to go to a specific site ("Gecko Studio website").
• •Commercial: I am evaluating options ("best SEO agencies London").
• •Transactional: I want to hire or buy ("hire SEO audit").

For each target page, search that keyword in Google and identify what type of content dominates the top results. If "SEO audit" returns only informational guides, a sales landing page will not rank there.

3.5 Keyword cannibalisation

Cannibalisation occurs when two or more pages on your site compete for the same keyword. Google cannot decide which to rank, alternates between them and the end result is weaker than having one well-optimised page.

How to detect it:

• •In GSC: filter by keyword in the Performance report. If you see multiple URLs alternating positions for the same query, there is cannibalisation.
• •With Screaming Frog + GSC data: cross-reference the keywords assigned to each URL.

How to resolve it:

• •If one page is clearly stronger, redirect the weaker one to the stronger (301).
• •If both pages have distinct valuable content, clarify the intent of each and consolidate any overlap.
• •If the differentiation is artificial, merge the two into a single, more comprehensive guide.

Step 4: Content and EEAT signals

Goal: assess whether the content is sufficiently deep, current and trustworthy for Google to consider it the best available answer.

EEAT stands for Experience, Expertise, Authoritativeness and Trustworthiness. Google applies these criteria in its quality rater guidelines to evaluate content, particularly on topics that influence important decisions.

4.1 Content quality and depth

For each strategic page on the site, ask yourself:

• •Does the page answer the question more completely than the competitors currently ranking?
• •Is the content original, or a shallow rewrite of what already exists?
• •Are there data, examples, real cases or first-hand experience that differentiate the content?
• •Is the length appropriate for the intent? (A product page does not need 2,000 words; an in-depth technical guide on a complex decision does.)

4.2 Outdated or stale content

Content that was relevant 3–4 years ago may be losing positions because competitors have updated theirs. Identify:

• •Pages with visible old dates on topics that change (algorithms, regulations, tools).
• •Guides mentioning features or data that no longer exist.
• •Pages ranking in positions 8–15 with no movement for months: these are often prime candidates for an update.

4.3 Thin content

Thin content refers to pages with very little useful content: empty category archive pages, automatically generated pages with variable data, tag pages with a single article. Options:

• •Enrich with genuine content if there is a justification.
• •Consolidate similar pages.
• •Set to noindex if the page adds no value and receives no traffic.

4.4 Specific EEAT signals to review

• •Is the content author identified by name and profile?
• •Is there an "About us" or "Team" page with real profiles?
• •Are data sources and statistics cited?
• •Are there verifiable testimonials, case studies or reviews?
• •Are the privacy policy, legal notice and contact details accessible from every page?

Step 5: Internal linking and architecture

Goal: ensure that internal authority (PageRank) flows correctly from the strongest pages to the pages you want to rank.

5.1 Crawl depth

No important page should be more than 3 clicks from the homepage. Pages deep in the hierarchy receive less internal authority and are crawled less frequently.

In Screaming Frog: use the "Crawl Depth" report > export and filter for strategic pages more than 3 hops away.

5.2 Orphan pages

These are indexed pages that receive no internal links. Google may find them via the sitemap, but without internal links they receive virtually no authority. Locate them using Screaming Frog's "Orphan Pages" report (requires importing the sitemap) or by cross-referencing the sitemap against the list of crawled URLs.

5.3 Anchor text and context of internal links

The anchor text of an internal link is a semantic signal for Google. Always using the same generic anchor ("click here", "find out more") is a missed opportunity. The anchor should be descriptive and relevant to the destination page.

5.4 Authority distribution

Identify the pages with the most external backlinks (using any tool that shows the link profile). Make sure those pages link internally to the pages you most want to rank. Internal link sculpting is one of the highest effort-to-result levers in SEO.

Step 6: External link profile

Goal: get a clear picture of the quantity, quality and distribution of sites linking to yours.

An exhaustive analysis of every backlink is not necessary in a first SEO audit. What you do need to know:

• •Number of unique referring domains: more representative than the total number of backlinks (a single domain can link once or a thousand times; what matters for authority is the distinct domain count).
• •Average domain quality: tools such as Ahrefs (Domain Rating), Moz (Domain Authority) or Semrush (Authority Score) provide a reference score.
• •Anchor distribution: do backlinks use your brand name, naked URLs or exact-match keywords? A profile with many exact-match keyword anchors can signal artificial link building.
• •Time trend: is the number of referring domains growing, stable or declining? A sharp drop can indicate a penalty or loss of relevance.
• •Broken backlinks: external pages linking to URLs on your site that no longer exist. Recover that authority by redirecting those URLs to active equivalent pages.

On low-quality links: if you detect a clear pattern of spam backlinks (low-value directories, link farms, artificial private blog networks), document it. The actions to take depend on volume and context.

Step 7: SEO beyond Google (2026 differentiator)

This is the section most audits ignore, and in 2026 that omission is no longer trivial.

7.1 Bing Webmaster Tools

Bing holds between 5% and 9% market share in many English-speaking markets, and its results feed Microsoft Copilot and several secondary search engines. Check:

• •Is the site verified in Bing Webmaster Tools?
• •Is the sitemap submitted?
• •Are there crawl errors specific to Bingbot that do not appear in GSC?

Setup takes under 20 minutes and prevents Bing-specific issues from going unnoticed for months.

7.2 IndexNow

IndexNow is an instant content-change notification protocol used by Bing, Yandex, Seznam and other search engines. Instead of waiting for a crawler to reach your new URL, the site actively notifies the search engine.

How to enable it: several WordPress SEO plugins integrate IndexNow natively (RankMath supports it out of the box; others via an add-on). Verify it is enabled. For other platforms, implementation requires the API or a webhook.

7.3 llms.txt

An llms.txt file in the root of the domain is the emerging convention (2024–2025) for signalling to language model crawlers which content is suitable for indexing or citation in AI-generated answers. It is the AI-era equivalent of robots.txt.

Check whether the site has yourdomain.com/llms.txt. If it does not exist and the client wants visibility in AI-powered searches, adding it is a low-cost, increasingly relevant action.

7.4 AI Overviews and generative search

Google AI Overviews (the generative summary at the top of results) now appear in many markets for informational queries. Pages cited in those summaries need:

• •A clear structure with direct answers to specific questions.
• •Structured data markup (FAQ, HowTo, Article).
• •Strong EEAT: the model favours sources with verifiable authority.

There is no canonical tool for auditing AI Overviews presence systematically yet, but you can manually sample the top 20 keywords for the site and check whether it appears as a cited source.

Step 8: Prioritising findings

A typical SEO audit generates between 80 and 300 findings. The real work of a professional audit is not finding them — it is ordering them.

Prioritisation framework (Impact × Effort matrix):

Criteria for raising a finding's priority:

1. 1.It affects pages that already receive traffic or backlinks.
2. 2.It blocks crawling or indexation (serious technical issue).
3. 3.There is a keyword opportunity with low competition and high commercial intent.
4. 4.The direct competitor has already addressed it and you have not.

Present findings as concrete actions with an owner and a date, not as a list of abstract problems. "Optimise SEO" is not a task; "update the H1 on the 5 highest-traffic product pages to include the city modifier" is.