How to Do a Technical SEO Audit Step by Step
Written by SEOdiag Team · Published on 2026-05-11
A technical SEO audit is a systematic analysis of a website's infrastructure to identify problems that prevent or hinder its crawling, indexing, and ranking in search engines. It's not about reviewing keywords or analyzing backlinks — it's about making sure the technical foundation that all SEO rests on is healthy.
This guide covers the complete process, from preparation to report delivery, designed for SEO professionals who need to audit client sites efficiently and repeatably.
Step 1: Define the scope
Before launching any crawler, you need to answer three questions:
How many URLs does the site have? A 50-page site can be audited manually in an afternoon. A 15,000-product ecommerce store requires automated tools. The scope determines the tool.
Are there subdomains or localized versions? A site with blog.example.com, shop.example.com and versions in /en/, /es/, /pt/ requires you to decide whether you'll audit everything or just the main domain.
What's the business priority? If the client is losing traffic on product pages, focus the audit on the transactional section before reviewing the blog.
Step 2: Crawl the entire site
The crawl is the foundation of every audit. A crawling tool visits every URL on the site following internal links, and records: the HTTP status code, titles, meta descriptions, headings, broken links, redirects, canonicals, hreflang, and URL structure.
What you need from a good crawler:
- Redirect chain tracking to detect chains and loops.
- JavaScript rendering for SPA sites or those with dynamic content (more on this topic in our JavaScript SEO guide for SPAs).
- WAF evasion capability if the site is behind Cloudflare or another firewall that blocks bots.
- Structured data extraction (Schema.org, Open Graph).
Step 3: Check indexability
Once you have the complete crawl, the first analysis is verifying what Google can see. For a detailed catalog of what to look for in each category, check out our article on the 12 most common technical SEO errors.
Robots.txt: Are there rules blocking important sections? Check that CSS and JS assets needed for rendering aren't blocked.
Noindex tags: Are important pages marked with <meta name="robots" content="noindex">? It's surprisingly common to find this on sites that migrated from staging to production without cleaning up meta tags.
Canonical tags: Do pages with URL variations (with/without trailing slash, with parameters) have a canonical pointing to the correct version?
Sitemap.xml: Does it exist? Is it up to date? Does it match the site's actual URLs? A sitemap with URLs returning 404 or noindex sends contradictory signals to Google.
Step 4: Analyze internal link architecture
The internal link structure determines how authority (PageRank) flows within the site and how efficiently the crawler can discover all pages.
Click depth: How many clicks separate important pages from the homepage? If a product page is more than 3 clicks deep, Google assigns it lower crawl priority.
Orphan pages: Are there URLs that receive no internal links? These pages are virtually invisible to search engines.
Link distribution: Does the homepage concentrate 90% of internal links while category pages barely have 2 or 3? An extreme imbalance wastes authority.
Step 5: Evaluate on-page elements
For each crawled URL, check:
- Title tag: unique, descriptive, ≤60 characters, including the primary keyword.
- Meta description: unique, persuasive, ≤155 characters.
- H1: only one per page, consistent with the title.
- Images: all with descriptive
altattributes, modern format (WebP), declared dimensions to prevent CLS. - Structured data: valid Schema.org for the content type (Product, Article, FAQ, LocalBusiness).
Step 6: Measure speed and Core Web Vitals
The metrics that matter for ranking are:
- LCP (Largest Contentful Paint): the largest visible element should load in under 2.5 seconds.
- INP (Interaction to Next Paint): the response to the user's first interaction should be under 200ms.
- CLS (Cumulative Layout Shift): cumulative visual shift should be under 0.1.
Check these metrics in both field reports (real user data from Chrome UX Report) and lab data (Lighthouse, PageSpeed Insights).
Step 7: Verify multilingual configuration
If the site has versions in multiple languages or for multiple regions:
- Hreflang: each page must point to all its alternate versions, and references must be reciprocal.
- URL structure: verify that the URL structure is consistent (
/en/productvs/product?lang=en). - Translated vs. duplicate content: pages in different languages must have effectively translated content, not the same text in one language with meta tags from another.
Step 8: Prioritize findings by severity
Not all errors have the same impact. A practical classification:
| Severity | Criteria | Examples |
|---|---|---|
| Critical | Prevents indexing or causes mass de-indexation | noindex on key pages, robots.txt blocking everything, redirect loops |
| High | Directly affects ranking of important pages | duplicate titles on transactional pages, massive duplicate content, LCP > 4s |
| Medium | Degrades overall quality but doesn't block | missing meta descriptions, images without alt, 3-hop redirect chains |
| Low | Best practices not met | URLs with uppercase, underscores instead of hyphens, incomplete structured data |
Step 9: Generate the report
The audit report should have two formats:
Executive (for management/client): one-page site health summary, global score, top 5 critical issues with estimated impact, and prioritized recommendations.
Technical (for development): detailed backlog with every affected URL, the error type, severity, and specific corrective action. Ideally in spreadsheet format so the team can assign tasks.
Automating the process
Doing all of this manually takes 4 to 8 hours per site. For agencies auditing multiple clients, automation isn't a luxury — it's an operational necessity.
SEOdiag is a SaaS platform that automates the entire technical SEO audit process described in this guide. It crawls the site in the cloud, detects errors, classifies them by severity, and automatically generates two deliverables: an executive PDF report with AI-powered diagnostics and a technical Excel backlog ready for the development team. Multi-tenant to manage multiple clients from a single account, with plans starting at USD 1.