Phase 2 · Foundation

Fix Broken Links
(External)

Phase 2 — Foundation  |  Frequency: Monthly  |  Duration: ~2 hrs/client  |  Deliverable: Fixed outbound links + Zoho comment

Identify outbound external links pointing to dead third-party pages. Crawl in external link mode, validate each broken link, then update to the live URL, replace with an authoritative alternative, or remove.

Overview

What It Involves

Crawl the site in external link mode (Configuration → Spider → Check External Links) using Screaming Frog. For each broken outbound link: find if the target page moved to a new URL on the third-party site; if yes, update. If not, replace with a different authoritative source on the same topic. If no good replacement exists, remove the link.

External broken links are lower priority than internal, but they appear in any technical audit the client runs against us — Ahrefs, Semrush, or similar. A 2-hour fix that clears the external link audit is easy, defensible value.

Why It Matters

  • User trust — clicking a "learn more" link and landing on a 404 creates a poor experience and damages credibility.
  • Client audit exposure — external 404s appear in Ahrefs and Semrush crawls which clients sometimes run independently.
  • Defensibility — proactively fixing these before the client spots them demonstrates thoroughness.

Crawl Setup

Configure Screaming Frog for External Link Checking

In Screaming Frog: Configuration → Spider → tick "Check External Links."

This causes Screaming Frog to follow outbound links and report their HTTP status codes from third-party domains.

Important — run as a separate crawl

  • External link checking significantly slows crawl speed
  • Run this as a separate crawl from your standard internal crawl
  • Do not combine them on large sites — timeout risk is high
  • Enter the client domain and run once configuration is set

Identify Broken External Links

Export Broken External Links

Filter: Response Codes → 4xx (client errors from external domains).

Also check for 5xx responses (server errors on external sites) — these are often temporary but worth noting.

Switch to the Inlinks tab — this shows which page on the client's site contains the broken outbound link.

Before acting

  • Export as CSV and note the total count
  • Save the export file immediately — before making any changes
  • Do not act on any 4xx until validated manually in a browser (see Step 3)

Resolve Each Link

Validate Each Broken Link Manually

Not every 4xx in Screaming Frog is actually broken. Some sites block crawlers (Cloudflare, bot protection) but serve real pages to browsers.

Open each flagged URL in a browser. If it loads fine, record it as "False positive — page accessible in browser" and move on without making any change.

Only act on confirmed broken links

  • Genuinely 404s or shows "page not found" in the browser: proceed to fix
  • False positive (loads fine in browser): record and skip — do not alter the link
  • 5xx in Screaming Frog: check in browser; if also failing, note as temporary server error

Resolve Each Genuinely Broken External Link

For each confirmed broken outbound link, work through the following decision process:

Resolution decision tree

  • Check if the page moved: paste the URL into archive.org (Wayback Machine) to see what it used to say, then search for the current version on the same domain using site:domain.com "original title"
  • If found at a new URL: update the link on the client's page to the new destination
  • If not found on that domain: find an alternative authoritative source covering the same topic — government (.gov.au), industry association, or well-known publication preferred
  • If no good alternative exists: remove the link entirely. A sentence that referred readers to an external source can be rewritten to stand on its own, or removed if it was filler.
Quality bar for replacements: Only replace with authoritative sources. A low-quality or loosely related page is worse than removing the link — it damages credibility and signals poor editorial judgement to both users and Google.

Special Cases

Links Requiring Client Sign-Off Escalate

Do not change the following without checking with the account manager first:

Categories requiring escalation

  • Partner or sponsor links — these are commercial relationships. The client may want to be informed even if the partner's page is 404ing.
  • Affiliate links — clients sometimes have revenue-share arrangements. Changing or removing these requires explicit sign-off.
  • Government or regulatory citations — if the regulation or standard has genuinely moved, confirm with the client before updating (they may have a preference for which version of the document is cited).

Record these in the Zoho comment as "escalated to client for sign-off" with a brief explanation of why.

Document & Log

Log in Zoho

Add a comment to the client's Zoho task. The comment must include:

Required fields in Zoho comment

  • Crawl date and total external broken links found (pre-fix count)
  • Number fixed (updated to new URL on same domain)
  • Number replaced (different authoritative source)
  • Number removed (no suitable replacement)
  • Number escalated to client — with brief reason for each
  • Time logged

Standards

Completion Checklist

Every item must be met before the task is marked complete

  • External link check run as a separate crawl — not combined with the internal crawl
  • Every flagged 4xx validated manually in browser before acting — no false-positive removals
  • Authoritative sources prioritised for replacements (.gov.au, industry associations, recognised publications)
  • Partner and affiliate links escalated to account manager rather than changed unilaterally
  • Government and regulatory citations confirmed with client before updating
  • Pre-fix export saved to client Drive folder before any changes were made
  • Zoho comment posted with counts, escalations noted, and time logged

Common Mistakes

Mistakes to Avoid

Mistake Why It's a Problem What to Do Instead
Acting on Screaming Frog 4xx without browser validation Some sites block crawlers (Cloudflare, bot protection). You'd remove or replace links that are actually fine and serving real content to users. Open every flagged URL in a browser before deciding what to do. Only act on links that genuinely fail in-browser.
Replacing a broken citation with any available page Linking to a low-quality or irrelevant source is worse than removing the link. It damages editorial credibility and signals poor judgement to Google. Only replace with authoritative sources: .gov.au, industry associations, recognised publications. If none available, remove the link.
Changing partner or affiliate links without checking These are commercial relationships. Changing them without notice can break revenue-share tracking or damage a business relationship — potentially a financial liability for the agency. Flag these to the account manager. Never change commercial links unilaterally.
Running external link check during the standard crawl Doubles or triples crawl time on large sites and can cause timeouts. Mixes two different data sets into one export, making it harder to act on either. External link checking is a separate configuration and a separate crawl run. Keep them distinct.