To test redirects after a site update, start at the old address and verify every hop and the final page. A successful HTTP response tells you the destination loaded. The heading and content tell you whether the visitor ended up in the right place.
Imagine you moved your setup docs. A link from a customer onboarding email now opens the help homepage. The browser shows a working page, but the reader still has to find the instructions they came for.
Which old URLs should you test first?
Start with the redirect map from the release. Add addresses with recent visits or Search Console clicks, useful backlinks, and links from emails or product screens. Include unusual variants only when they were actually used or are part of the launch plan. Keep one row per old address with the expected final URL and the task it should support.
Google's redirect guidance recommends a server-side permanent redirect when a move is permanent. It treats permanent redirects as a signal about which URL should be canonical. For a temporary move, choose the temporary behavior that matches the change.
What should a passing check look like?
- Open browser developer tools and enable Preserve log in Network. Load the old URL in a clean session.
- Inspect the document requests and record the redirect status and destination at each step. Check for a loop, a stale intermediate page or an unexpected hostname.
- Open the final URL directly. Confirm it returns the expected page, uses the right canonical if applicable, and still contains the section, download or action people need.
- Update editable links to point directly to the final destination. Retest a sample of real paths from navigation, a help article or an email.
In a fictional example, an old export guide redirects to a retired documentation hostname before reaching the new export guide. The final content is right, but the extra dependency can break later. Update the rule to point straight to the new guide. If a second old link goes to the generic help homepage, map it to the relevant guide only when that guide actually answers the same task.
What if nothing replaces the old page?
Ask what a visitor expects from the URL. If the resource was intentionally removed and no equivalent exists, a clear missing-page response can be more honest than silently redirecting everyone to an unrelated page. Record the decision so someone can revisit it if an equivalent resource becomes available.
Retest your priority rows after deployment, then use the visibility audit checklist to log lingering broken paths and assign follow-up.
How do you turn the redirect map into a release check?
Keep the expected result distinct from what you actually observe. That lets a second person retest the same path without interpreting the original migration notes.
| Record | Example from a fictional docs move |
|---|---|
| Old path and expected final path | /docs/old-export should reach /help/export. |
| Observed response chain | The old path returns 301, then the replacement returns 200. |
| Reader check | The page opens the export guide and still links to the downloadable CSV example. |
| Disposition | Pass, or assign the exact rule, link or page that needs repair. |
Before launch, test the redirect rules in a safe preview environment if your host supports them. After launch, rerun the priority rows on the public hostname. A preview can behave differently from the production CDN or routing layer. Test as a new visitor if you've previously loaded the same redirects in your browser.
If a chain ends at the expected page but drops a needed fragment, update links you control and check that the target section still has a stable anchor. Browser fragments are handled on the client. Verify the visible landing position in an actual browser, alongside the network checks. That gives readers the right section as well as the right page.



