How to Use Screaming Frog for SEO Audit: 10 Powerful Steps

Most people download Screaming Frog, run a crawl, see a wall of data, and have no idea what to do with it.

The tool is extraordinary. But it only produces value if you know which reports to look at, what the data is telling you, and how to export it in a format you can actually act on.

Whether you have just downloaded it and are not sure where to start, have been using it for basic crawls and want more depth, or want a systematic process that produces actionable outputs every time, this guide covers every step from the first configuration decision to the final export.

What Screaming Frog Does (And Why It Matters)

Screaming Frog SEO Spider is a desktop crawler for Windows, macOS, and Ubuntu. It mimics what Googlebot does: visits every accessible URL, reads HTTP headers and HTML content, follows links, and records technical data about every page encountered.

The advantage over cloud-based tools: it crawls from your computer with real-time response data. Every redirect chain, every canonical tag value, every response code you see reflects the actual current state of your server.

Free version: Up to 500 URLs with all core reports. Sufficient for small sites and focused section audits.

Paid version (£199/year): No URL limit, JavaScript rendering, GSC integration, GA integration, scheduled crawls, near-duplicate content detection, and server log analysis. For agency or ongoing work, it pays for itself on the first large audit.

How to Use Screaming Frog for SEO Audit: 10 Steps

Step 1: Configure Before You Crawl

The most common Screaming Frog mistake: crawling immediately without checking configuration. Wrong settings produce incomplete or misleading data.

Under Configuration, then Spider:
Enable “Store HTML.” This lets you search page content and use the Near Duplicates report later.

Under Configuration, then Rendering:
This is the highest-stakes configuration decision.

For React, Vue, Angular, or client-side Next.js: select JavaScript. Without this, Screaming Frog reads the near-empty initial HTML and misses all content that appears after scripts execute. The trade-off: crawl time increases significantly. A 200-page site that crawls in 3 minutes in Raw HTML mode may take 20 minutes in JavaScript mode. It is worth the wait.

For WordPress, Shopify, or any server-rendered site: leave as None (Raw HTML). Faster and produces accurate results because all content is in the initial HTML.

Under Configuration, then Speed:
On shared hosting, reduce thread count to 1 to 3. Too many simultaneous requests triggers rate limiting and floods your response codes with “Too Many Requests” errors that obscure real audit data.

Step 2: Run the Crawl and Let It Finish

Enter your root domain with https://. Select Spider mode. Press Start.

Monitor the status bar. A sudden flood of 5XX status codes means your server is struggling. Pause, reduce thread count, restart.

Do not interrupt a crawl mid-way to start analysing. Partial crawl data produces misleading finding sets. Let it complete fully.

Step 3: Start With Response Codes

Click the Response Codes tab. This is the most important first view after any crawl.

4XX (broken pages): Click the 4XX filter. Every URL here is a broken page. For each one, click the Inlinks tab to see every source page linking to it. This is your broken link map. These pages waste crawl budget, leak PageRank, and create dead ends for users.

3XX (redirects): Check for 302s where 301s are needed. A temporary redirect on a permanently moved URL does not pass PageRank as reliably. Every 302 on a URL you intend to permanently redirect is a fix item.

5XX (server errors): Critical severity. These pages cannot be crawled or indexed. Investigate server health immediately.

2XX (success): Scan for any expected important pages that are absent. If a key page is missing from the 2XX list, it has a redirect, a robots.txt block, or a DNS resolution issue.

Step 4: Audit Page Titles, Meta Descriptions, and H1s

Page Titles tab. Filter for each issue:

• Missing: Google generates a replacement automatically, almost always poorly
• Duplicate: competing pages sharing a title, strong indicator of content duplication
• Over 60 characters: title truncated in search results
• Below 30 characters: typically too vague to compete for meaningful queries

Repeat under Meta Descriptions and H1 tabs.

Export each issue list as CSV. Every row becomes a row in your fix tracker.

Step 5: Build the Complete Broken Link Map

Go to Reports, then All Inlinks. Export as CSV. Open the export and filter the “Status Code” column for 404.

What you have now: every broken link on your entire site, mapped to its exact source page and anchor text. No manual page-by-page checking. The work is done automatically.

This export is the most immediately actionable output from any Screaming Frog audit. Use it as the input to your broken link fix sprint.

Full guide: How to Fix Broken Links in SEO: 6 Proven Steps That Work

Step 6: Identify All Redirect Chains

Reports, then Redirect Chains. Sort by Chain Length, descending.

For every chain of 3 or more hops: record the source URL and the final destination. Your fix is to update all internal links pointing to the source to point directly to the final destination, then update the redirect rule to bypass all intermediate hops.

If you see a URL appearing as both a source and a destination in the redirect data, you have a loop. ERR_TOO_MANY_REDIRECTS. Fix this before any chains.

Full guide: How to Fix Redirect Chains: 5 Proven Recovery Steps

Step 7: Review Canonical Tags

Canonicals tab. The statuses that require action:

“Non-Indexable Canonical”: canonical pointing to a noindexed page. Authority cannot consolidate onto a page that is excluded from the index.

“Canonical Points to Redirect”: canonical pointing to a URL that itself redirects. Unnecessary chain before Google reaches the intended canonical.

“Missing”: no canonical tag on the page. Vulnerable to parameter-based duplication from tracking codes, sorting, and filtering.

Step 8: Find All Accidental noindex Tags

Directives tab. Filter by noindex. Review every single result.

This check finds the most damaging hidden problem on many poorly maintained sites. An accidental noindex on a blog post template. A Yoast SEO setting accidentally set to “No” for a content type. A staging-environment directive that survived migration.

Every result here should be an intentionally excluded page. If any are not, they are the most urgent fix in your entire audit.

Step 9: Near Duplicates (Paid Version)

Reports, then Near Duplicates. This report identifies pages with similar body content using similarity scoring.

Most valuable for: eCommerce sites using manufacturer descriptions across products, multi-location sites with near-identical location pages, and blog content covering overlapping topics.

The output shows which pages are competing against each other and diluting each other’s ranking authority.

Step 10: Export Your Complete Finding Set

After all tabs are reviewed, export:

1. 4XX errors with Inlinks
2. Redirect Chains
3. Duplicate and missing Page Titles
4. Duplicate and missing Meta Descriptions
5. Missing and multiple H1s
6. noindex pages (complete list)
7. Canonical errors

Open each export in Google Sheets. Add columns: Severity, Fix Method, Fix Owner, Deadline. This spreadsheet is your audit tracker.

For the complete 14-step audit process this tool feeds into: How to Conduct a Technical SEO Site Audit: 14 Proven Steps

Full audit reference: Technical SEO Audit: 14 Proven Steps to Dominate Rankings

Frequently Asked Questions

Key Takeaways

• Configure rendering settings before crawling. Wrong settings produce incomplete data that misses critical issues.
• The All Inlinks export filtered by 404 is the single most actionable output. It maps every broken link to its source automatically.
• The Directives tab noindex check finds the most damaging hidden problems. Review every result manually.
• Export findings to CSV immediately after the crawl. Build your audit tracker from these exports before beginning analysis.
• Cross-reference Screaming Frog data against GSC Coverage data. Discrepancies reveal the most interesting problems.