How to fix broken links and redirects on a website
Fix broken links and redirects with a repeatable audit. Confirm what fails, find the source of each issue, and verify the exact replacement.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What a broken link and redirect audit is
To fix broken links and redirects, trace what a visitor clicks from the original page to the content that should open. A link can end on a deleted page, pass through several forwarding addresses or land on a page without the section it promises. A broken link checker can collect those paths. The repair work begins when you confirm which ones fail and work out what the visitor ought to find.
A useful audit links each confirmed problem to a source, an exact change and a check an editor can repeat. It groups repeated issues so the team can fix one shared menu item or server rule in one place. It also gives you a record of what you could test, which links still need a decision and which destinations you could not reach. That makes the result useful both to the people who edit the site and to the client deciding which work comes first.
What it does for the business
- Repair paths visitors already useStart with links that affect a common task, such as booking a service, opening product instructions or reaching a contact page. Restoring those routes can reduce the need for people to backtrack or ask for help.
- Fix a shared cause onceA broken item in a footer can appear on every page that uses the template. Grouping those appearances sends the developer to one edit and gives the reviewer sample pages to check afterward.
- Keep published references usefulA verified new address or an appropriate archived copy lets editors preserve the source a reader expects to find. When the resource no longer exists, the editor can decide whether the surrounding passage still holds up.
- Make a migration easier to maintainA list of old paths and their intended successors gives the agency a record of what moved. Use it to clean up links in new copy and spot a redirect rule that needs attention before the next release.
How to fix broken links and redirects, step by step
Collect the links, their sources and the crawl limits
A repair starts with knowing where the visitor meets the link.
What to do
- Set the site boundary and seed the website crawl with its home page and sitemaps. If Google Search Console supplies missing addresses or page clicks, keep that information alongside the crawl, with the source and date.
- Collect links and images from page content and shared templates. Inspect browser-rendered pages when scripts add links, and record which templates or addresses could not be rendered.
- Save the source URL, where the link appears, the link text or image reference and the destination. Record each occurrence even if the same destination only needs one status check.
- Run at a rate the site can handle and keep a list of destinations the crawl cannot reach. Distinguish checks of your own site from requests to other websites so you can report coverage for each.
Why
A destination list tells you what to test, while a source list tells you what to change. Together they can show that many failures come from one shared footer or that a problem appears only after a script loads.
Confirm what each destination does
A failed request needs evidence before it becomes a repair.
What to do
- Read the HTTP status codes, the numbers a server uses to describe its response. Save each redirect hop, its destination and the final response, then flag loops, unnecessary chains and paths that end at a missing page.
- Retry missing responses and nonexistent hosts at a different time. Record refusals, rate limits and timeouts as unconfirmed, and use a browser when a server may show ordinary visitors a different response.
- Read suspicious success responses to identify a soft 404, an error page that answers as if the request succeeded. For a broken anchor, inspect the destination's section ID, the name the link uses to jump to a heading or element.
- Check whether a redirect supports an intended action such as sign-in or language selection. Record its purpose before asking a developer to replace or remove it.
Why
An apparent failure may be a temporary connection problem or a redirect that serves a real function. Retaining the response trail and the result of the second check helps the client inspect the evidence before changing a working path.
Group failures and choose a repair that fits the link
A replacement should deliver what the original link promised.
What to do
- Group issues caused by a shared template, redirect rule, sitemap setting or dead outside host. Record the affected placements and prioritize problems that interrupt important visitor tasks.
- Read the anchor and its surrounding sentence to identify the intended destination. Choose a closest equivalent page for a moved internal resource and a direct final address where redirect chain repair is appropriate.
- For a citation, look for the same document at the publisher's new address. If you cannot find it, consider an archived copy that supports the passage or remove the link when the citation is unnecessary.
- When no page serves the original purpose, decide whether to restore the content, remove the link or leave a retired address missing. When you need an owner decision, state the options and what each would let readers do.
Why
The wording around a link sets an expectation. A generic category page or a different source may answer successfully while leaving the reader without the original instructions or evidence. Deciding at the level of the shared cause also makes related changes easier to review.
Verify the replacement and hand over the exact edit
A working address must open the intended content and be ready for an editor to use.
What to do
- Request each replacement and check for a direct success response. Open the page and confirm the content fits, the section works when specified and the page can be indexed if it is meant to appear in search.
- Write a repair table with the current address, the source placement, the link text, the proposed edit and the evidence. Provide a sample affected page and enough detail to find every shared occurrence.
- Read the first rows as an editor would. For a section jump, give the corrected section ID. For a server rule, give the old route and intended final address. For a shared template, say which element to change.
- Record decisions and coverage limits in the handoff. If a replacement fails verification, leave it for another decision and state what failed.
Why
An editor needs enough detail to apply the intended fix without guessing which link or rule you meant. Verification also catches candidates that forward again, point to an error page or no longer contain the promised section.
Check what changed after the repairs go live
A repair is complete when the live link works in its original context.
What to do
- Reopen the original affected pages after release and locate the edited element by its text, template location or old address. For a shared component, test examples from the templates that use it.
- Follow the final link as a visitor would. For a redirect rule, test the old address as well as the final destination. Confirm both the response and the content shown.
- Record which fixes were made as proposed, which were made differently and which still need work. Keep pages you cannot open in a separate group until you can check them.
- Schedule a broader crawl when the release touched more areas than the repair list covers. Keep the earlier table to make it easier to see which problems returned.
Why
A release can contain a sound fix at one source while leaving a duplicate in another template. Rechecking the original placements verifies the repair you agreed on and gives the client a remaining list of issues with evidence.
Common mistakes
- Calling a timeout a dead linkKeep temporary failures and access refusals apart from confirmed missing resources.
- Treating every redirect as a faultRead the full path and retain shortcuts or user flows that intentionally redirect.
- Verifying only the status codeRead the destination so a success response cannot hide a missing page or the wrong document.
- Editing repeated links one page at a timeFind the shared component and check where else it appears before opening individual tickets.
- Closing a repair without a live checkReturn to the original page and verify the exact element and its destination after deployment.
What a good result looks like
- A dated crawl with the scope, pages checked and unconfirmed destinations visible.
- A source and test record for every confirmed failure.
- Repeated issues grouped under their causes with affected locations available.
- An exact, verified replacement or a written decision for every action row.
- A recheck plan that names the original placements and the expected response.
And then what? Using the result
- Assign shared template or redirect repairs to the developer and page-level replacements to the content editor.
- Decide which missing resources are valuable enough to restore and give each one an owner.
- Review external citations that lack a good replacement and update their surrounding text if needed.
- After deployment, recheck the planned edits and use a broader crawl to review any untouched sections.
When to run it
- After a migration or URL changeTrace old routes and their successors and check that menus and in-body links lead to the final addresses.
- When customers hit errors in a key taskStart from the page they were on and follow the same steps. Group any repeated failures you find across other pages with the same component.
- Before handing over a client siteUse the verified table to document what the agency checked, what requires an owner decision and what to confirm following launch.
- When refreshing a library of articlesCheck prior citations in batches and group missing sources by publisher so editors can look for moved documents together.
Questions people ask
How do you fix broken links and redirects?
Find the source of each issue, confirm what the destination does, then update the link or redirect to a verified address that serves the intended purpose. Group repeated issues by shared template or rule. After release, follow the edited links on the live site and repeat the status checks.
What counts as a confirmed broken link?
A link is confirmed broken when repeated checks at different times show that the destination is missing, gone or on a host that no longer exists. A redirect that repeatedly ends on a missing page belongs in that group. Record timeouts, access refusals and rate limits separately, with their evidence.
How do you repair a redirect chain?
Update internal links to the correct final address and shorten any forwarding path you control so the old address can send a visitor directly there. Check the purpose of each rule first. Test the old and new addresses and other URLs that may share the rule.
What is a soft 404?
A soft 404 is a missing-page response that looks like an error to the visitor while the server sends a success status. Read the page and compare it with a clearly nonexistent path. Update the link or restore the resource if the promised content is missing.
Should you redirect every missing page to the home page?
No. Redirect a retired address when a successor serves the same purpose. When no equivalent exists, restore material people still need, remove links that no longer help, or leave a genuinely retired address returning a proper missing-page response.
Can a broken link checker verify the repairs after launch?
Yes, if it reopens the original source pages and checks the actual edited elements and their destinations. A targeted recheck tests the fixes you made. Run a broader crawl if you also need to find problems introduced elsewhere.
More jobs in SEO agencies
Fix broken links and redirects, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac