Start with the journey, not the error count
Collect broken URLs from analytics landing-page reports, server or CDN logs, Search Console, a site crawler, campaign destination reports, email templates and customer-service complaints. Deduplicate by requested URL and record where each link originated. A single broken checkout link used in every cart email can matter more than hundreds of obsolete article links with no current traffic.
Classify errors by journey: paid landing page, organic discovery, product selection, form completion, booking, account access or informational browsing. Record approximate affected sessions, conversions at risk, active campaign exposure, last known working destination and business owner. Avoid claiming exact lost revenue from a 404 count alone; intent and conversion probability are uncertain.
Separate legitimate 404s from broken experiences
A 404 is appropriate when content is genuinely unavailable and has no equivalent replacement. Google explicitly advises using the correct HTTP status, rather than returning a success response with a message that says the page does not exist. A page displaying an error while returning HTTP 200 can be treated as a soft 404 and can confuse monitoring.
Check the actual response code, redirect chain and final rendered page. A client-side router may display a branded not-found screen while the server returns 200. An expired product can also redirect to an unrelated homepage and technically avoid a 404 while leaving the visitor without an answer. The user's outcome, not merely a green status check, determines whether the journey is repaired.
Prioritize failures using a simple impact matrix
Score each broken path by current traffic, intent, business criticality and repair confidence. Mark active paid destinations and conversion-step links as urgent. Next handle heavily linked category pages, high-intent service pages and links from important email sequences. Low-traffic historical URLs with no useful replacement can remain correct 404s rather than becoming unnecessary redirects.
For example, a service company discovers that a popular 'request a quote' CTA points to an old form route. Repair the CTA immediately, confirm submission delivery and check whether advertising campaigns use the same URL. By contrast, a five-year-old event announcement that no longer exists may simply need a useful not-found response.
Choose the right recovery mechanism
When a page has moved to an equivalent destination, use a permanent redirect and update the internal links to the final URL. When the page is temporarily unavailable, communicate the status accurately and avoid indefinite temporary redirects. When there is no relevant replacement, keep the 404 or 410 and provide clear navigation and search options. Do not automatically redirect all missing pages to the homepage.
Make the recovery page useful without pretending the original content exists. Explain that the page cannot be found, offer the site's primary navigation, search if available, a relevant category and a way to report a broken link. Preserve the visitor's dignity with plain language; avoid jokes that obscure the problem or aggressively push a contact form when the user wanted a product detail.
Test the actual path across devices and sources
Follow links from search results, email, social posts, paid ads, navigation and PDFs. Check URL encoding, capitalization, trailing slashes, query parameters and mobile deep links. Test browser back behavior, the header and footer, and whether the suggested recovery link itself works. Verify that the replacement page loads and contains the promised offer or information.
Use synthetic test submissions when a broken path leads to a form or booking flow. A successful redirect to the form is not enough if the form no longer sends email or CRM records. For ecommerce, verify that the replacement product is genuinely equivalent and that availability is current. Include accessibility: the not-found heading should be understandable and keyboard navigation should remain functional.
Monitor after the fix and prevent regressions
Create an error dashboard showing requested URL, referring path or campaign, first seen, last seen, affected sessions and owner. Add automated checks for critical destination URLs and crawl internal links after every major release. Set an alert for spikes in 404s on active conversion paths rather than for every ordinary crawler request to a missing URL.
Review external backlinks and campaign links that still point to retired URLs. Update owned links at the source where possible, even when a redirect exists. Recheck redirect chains and Search Console trends after release. Measure restored journey completion and form delivery, not just the reduction in reported 404 responses.
Turn recovery into a release acceptance test
Keep a small list of business-critical URLs and run an end-to-end smoke test before publishing a website redesign, CMS migration or campaign. Include canonical destinations, CTA links, forms, thank-you states and analytics events. Store the expected HTTP status and owner beside each URL. This turns broken-link prevention into a shared operational responsibility.
Pair this guide with the website form reliability and post-conversion experience articles. A recovered link only creates business value if the rest of the journey works, the visitor can complete the task and the resulting enquiry or order is actually received.