First, split the expected redirects from the ones that point at a problem
The crawling and indexing guide, citing Google's redirects documentation, sets out what a redirect does: a permanent redirectA permanent server redirect telling browsers and search engines a URL has moved for good. passes the old URL's signals to the destination, which is indexed instead, while a temporary one may keep the original indexed for a while. The expired pages guide adds what that looks like in Search Console (GSCGoogle's search performance and indexing tool.): redirected URLs appear in the Page Indexing report as pages with redirects. So an old address in that list is, by default, a redirect doing its job. The exact conditions behind each reason live in the Search Console help centre; the Search Console tutorial explains how to read the report by trend and by whether the listed pages should be indexed.
Export the examples and ask three questions of each. Is this an address that should redirect: an old URL, a merged page, a host, protocol or trailing-slash variant? If yes, does anything the site controls still point at it? And does the redirect reach a working page in one step? Then glance at the redirect error reason as well, which the SEO checks for staging and releases name among the reasons to read after every release: those entries are redirects that do not resolve, and they always need fixing.
| What the example is | What it means | Action |
|---|---|---|
| An old or variant address, nothing on the site names it, one hop to a live page | The redirect is working | Leave it; log it as expected |
| An old address still linked, listed in the sitemap or named by a canonical or hreflang | The site contradicts its own redirect | Update the reference to the final address |
| A page that should be live | A rule is catching it | Fix the rule |
| An address whose redirect hops several times, loops or ends on an error | The redirect never resolves | Point it straight at a working destination |
Before removing any redirect that looks pointless, apply Chesterton's fence: find out why it was added. The SOP for changing a page's URL treats a redirect for a moved page as permanent, because links on other sites and old emails go on using the old address.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Old URLs from a move, or protocol, host and slash variants, redirect in one hop | Redirect map; request a sample | Old addresses and host or protocol variants that redirect as intended | Nothing; record as expected and keep the redirects |
| A crawl finds links to redirected addresses in menus, templates or copy | Crawl inlinks to the listed URLs | Internal links and navigation pointing at redirected URLs | Repoint links to the final address at the source |
| Listed URLs are in the submitted sitemap | Compare the sitemap with the examples | A sitemap listing redirected URLs | Regenerate the sitemap from final, indexable addresses |
| Live pages declare a canonical, or hreflang alternates, that redirect | Read the canonical and hreflang on templates; request each target | Canonical tags or hreflang naming a URL that redirects | Point every annotation at the final address |
| Pages that should be live redirect elsewhere | Request them; release log against first detected date | A redirect rule catching pages that should be live | Narrow or correct the rule; recheck the pages |
| Redirect errors; a trace shows several hops, a loop or a failing destination | Trace the full redirect path for each example | Redirects that never resolve: chains, loops or broken destinations | Send every old address to a working final page in one hop |
Fix, verify and prevent each cause
Old addresses and variants that redirect as intended
After a move, a merger of pages or a change of protocol, the old addresses are supposed to redirect, and the site migrations guide plans them in a Redirect MapThe table of every old URL and its new destination, one row per address, from which a site migration or URL change is built and then checked. for exactly that reason. Host, protocol and trailing-slash variants are redirected to one preferred form for the same purpose, as the page on duplicate without user-selected canonical explains.
Fix: none; confirm each sampled entry appears in the map or follows the site's variant rules, and log the group as expected.
Verify: a sample of old addresses returns one permanent redirect each to the page that serves the same purpose.
Prevent: keep the redirects in place. The migrations guide asks for old redirects to stay alive because links on other sites never update, and removing them turns these entries into not-found errors.
Internal links and navigation pointing at redirected URLs
Every Internal LinkA link from one page to another page on the same website. to a redirected address makes visitors and crawlers take a detour. The internal linking guide asks for links to the canonical URL, because a link to a redirecting address sends signals through a detour, and lists links to redirects among the faults fixed at the source. After a move, menus, footers and body copy often keep the old addresses.
Fix: update the links in templates and content to the final address.
Verify: a fresh crawl finds no internal links to the listed URLs.
Prevent: build links from the same source of final addresses as the rest of the site, and include internal link updates in every URL change, as the URL change SOP requires.
A sitemap listing redirected URLs
The crawling guide's controls table says an XML SitemapA file listing the URLs a site wants search engines to discover and crawl. should hold canonical, indexable URLs that return success only, and that a sitemap containing redirects sends contradictory signals. A generator that reads an old list produces exactly that.
Fix: regenerate the sitemap from the final addresses and confirm the generator reads the same source as the live site.
Verify: every sitemap URL returns success when requested, and none of the listed examples remains in it.
Prevent: check the sitemap among the release checks, which already ask for it to list only canonical, indexable URLs. One deliberate exception exists: during a move with URL changes, the migrations guide keeps the old sitemap available for a period so old URLs are recrawled and their redirects discovered.
Canonical tags or hreflang naming a URL that redirects
A Canonical TagTells search engines the preferred URL. is a hint that Google weighs with redirects, sitemaps and internal links, as Google's page on consolidating duplicate URLs explains and the comparison of canonical tags and 301 redirects sets out. A canonical that names an address which then redirects tells Google two different things, and the crawling guide lists canonicals pointing at redirects among its common mistakes. Hreflang has the same weakness: the hreflang guide notes that Google documents hreflang values as needing the page's actual address, and that after a migration the set is regenerated, not redirected.
Fix: point every canonical and every hreflang alternate at the final address.
Verify: requesting each declared target returns success directly, and URL Inspection shows the declared canonical as the final address.
Prevent: generate canonicals and alternates from a configured list of final addresses, never from whatever address served the request.
A redirect rule catching pages that should be live
A broad rule for trailing slashes, case, old folders or parameters can sweep up pages it was never meant to touch. The Search Console error triage SOP uses exactly this as its worked example: a rule introduced in a release that sent category pages to a search results page. The release checks guide lists a replaced or reordered rules file among the ways a release breaks search.
Fix: narrow or correct the rule so it matches only the addresses it was written for, then run a live test in URL Inspection on the affected pages and request indexing for the important ones.
Verify: the pages return success and appear as indexed again over the following recrawls.
Prevent: include live pages from every template among the release checks' golden URLs, so any new redirect on a page that should answer success raises an alert.
Redirects that never resolve: chains, loops or broken destinations
Layered rules turn one redirect into several hops, conflicting rules loop, and a destination that was later removed leaves the redirect pointing at nothing. The page on a traffic drop after migration describes the effect: each extra hop slows visitors and crawlers, and a loop never resolves, so the destination is never reached.
Fix: flatten every chain so each old address goes straight to its final page, break loops by removing the conflicting rule, and repoint redirects whose destination is gone to a genuine equivalent. Where no equivalent exists, let the old address return not found or gone rather than redirecting it somewhere unrelated, which the soft not-found page explains is treated as an error.
Verify: a trace of each example shows one hop to a page that answers success; the SEO page checker confirms the status of the destination.
Prevent: when a page that receives redirects moves or is removed, update every row in the redirect map that points at it. If the old addresses now answer not found, the page on not-found errors in Search Console sorts which of them to redirect.
Prevention: one source for every final address
Almost every problem on this page is the site naming an address it has already retired. The cure is a single source of final addresses that every output reads: navigation and internal links, the sitemap, canonical tags, hreflang sets and the redirect map itself. When an address changes, it changes there once, and every output follows. Google's page on consolidating duplicate URLs treats these as signals that are weighed together, so keeping them consistent is what lets the redirect and the canonical say the same thing.
| Output | Reads final addresses from | Checked when |
|---|---|---|
| Navigation, templates and body links | The content system's current address for each page | Every crawl and after every URL change |
| XML sitemap | The same list of live, indexable pages | Every release |
| Canonical tags and hreflang | A configured list of final addresses, never the request | Every release, on golden URLs |
| Redirect map | Old address to final address, one hop | Every URL change and every removal of a destination |
Route single-page changes through the URL change SOP and site-wide changes through the migrations guide, both of which write the redirect map row before the page moves. Read the redirect reasons in the Page Indexing report in the week after every release. And keep a record of why each redirect rule exists, so the next person who finds it in the report can see the fence was put there on purpose. Server errors on the destination pages belong to a separate diagnosis, the page on server errors in Search Console; a new site that never appeared at all starts with a website not showing on Google.
A walking holiday company reads its redirect list (illustrative)
Hoxfeldra Walking Holidays
Hoxfeldra Walking Holidays, an invented company running guided and self-guided walking tours, moves its 85 tour pages under new addresses grouped by region. Six weeks later the page with redirect count has grown from 20 to 640 example URLs, and 2 redirect errors have appeared. Sorting the export shows 555 old date and variant URLs redirecting in one hop as planned, with nothing on the site naming them, which are logged as expected. The other 85 are the old tour URLs, and they are contradictions: the sitemap generator still reads the old address list, so all 85 are submitted, and the tour template builds its canonical tag from a stored field that kept the old address. The 2 redirect errors are live booking pages that a trailing-slash rule added in the same release sends into a loop. The team rebuilds the sitemap and the canonicals from the final addresses, narrows the trailing-slash rule, tests the booking pages live and requests indexing. Over the next 4 weeks the redirect errors clear in validation, the booking pages are indexed again, and the planned redirects stay in place.
Common mistakes
- Deleting redirects to shrink the report. The entries become not-found errors and every outside link breaks.
- Fixing the redirect instead of the reference. When the sitemap or a canonical names an old address, change the sitemap or the canonical.
- Leaving canonicals on the old addresses after a move. The tag then contradicts the redirect it sits behind.
- Writing broad rules without testing live pages. A slash or case rule can redirect pages that should answer success.
- Chaining new redirects onto old ones. Each move adds a hop unless earlier rows are updated to the final address.
- Redirecting to anything when the destination disappears. Without a genuine equivalent, not found or gone is the honest answer.
