Inspect the URL in Search Console and read the Google-selected canonical line. That is the address Google kept, and it points to the fix. If it is the same page at a different address, make your redirects, internal links and sitemap all use one URL. If it is a different page, Google thinks the two are duplicates: make them genuinely different, or keep one and redirect the other.
Why does Google overrule my canonical tag?
Because the tag is a preference, not an instruction. Google says stating a canonical is “a hint, not a rule.” It weighs your tag against everything else the site says. A permanent redirect is at least as strong: Google ranks redirects first and calls both redirects and the tag “a strong signal.” A sitemap is a weak one. Google also prefers HTTPS to HTTP, and notices which URL your own links use. When all of these agree, they reinforce each other. When they conflict, Google picks.
There is a second route into this status. Google can decide your page duplicates another page and group the two, whatever the tag says. Google’s help page describes it as a page “marked as canonical for a set of pages, but Google thinks another URL makes a better canonical.”
How do I see which URL Google chose?
In the Page indexing report, open this status and inspect one of the example URLs. Two lines under Page indexing tell you what happened. User-declared canonical is the address your page names. Google-selected canonical is the one Google indexed.
Go by the indexed result, not the live test. Google chooses the canonical at indexing time, and its help page says “the live test cannot predict” whether the tested version will be treated as canonical. So it cannot confirm a fix either.
Then look at the URL Google kept. It will be one of three kinds, and each has its own fix.
| Google kept | What that tells you | Fix |
|---|---|---|
| The same page at another address: www or bare domain, http, a trailing slash, a ? on the end | Your signals disagree about which address is the original | Make every signal name one URL. Or leave it, if Google kept the one you prefer |
| An address you would rather nobody used: an old path, a staging site | Something still sends Google there: a redirect, a link, the sitemap | Point every signal at your chosen URL |
| A different page: another location page, the first page of a list | Google thinks the two pages are the same page | Make them clearly different, or merge them |
The third kind is the one that frustrates service businesses most. A set of location pages such as IT support in Birmingham, IT support in Coventry and IT support in Leicester, built from one template with only the town swapped, can look like one page to Google. It groups pages whose main content is nearly identical and indexes one of them for the whole group.
What sends Google mixed signals about one page?
Usually something small and site-wide. These are the set-ups we look for first on a service business site:
- A tag that names a different addressThe page loads at www.example.co.uk/services/ but its canonical names the bare domain, or drops the trailing slash. If that other address then redirects back, the redirect and the tag contradict each other and Google has to overrule one of two strong signals.
- A second host that names itselfThe site loads on both www and the bare domain (or on http as well as https), and each copy’s tag points to itself. That is the classic route into this status.
- Two canonical tags on one pageOften a theme and an SEO plugin each print one. Google’s advice is that it “will likely ignore all the rel=canonical hints” when there is more than one, so the page ends up naming nothing.
- A plugin pointing the wrong wayGoogle warns that some CMS plugins “can make incorrect use of canonicalization techniques.” Check what yours writes on office, location and service templates.
What a second copy’s tag says decides where it appears in the report, if Google finds it. A copy whose tag points back to the original becomes an alternate page with proper canonical tag, which needs no action. A copy with no tag at all goes under Duplicate without user-selected canonical, where Google picks for it.
Does it matter which URL Google kept?
Sometimes not at all. If Google indexed the address you would want a customer to land on, people still find the page. Google’s troubleshooting guide asks you to consider whether its choice “makes more sense than your preferred canonical URL” for people arriving from Search.
When it has kept the wrong one, the cost is more than an untidy address. Google “uses the canonical page as the main source to evaluate content and quality,” so the copy you did not want is the one it judges and shows. Only that URL can appear as a link in AI Overviews or AI Mode, which need a page that is “indexed and eligible to be shown in Google Search with a snippet.” And if Google has folded your Coventry IT support page into the Birmingham one, the Coventry page cannot rank for Coventry at all.
How do you get Google to use your URL?
Start with the strongest signal you control and make everything else agree with it.
- Redirect copies nobody needsThe other host, plain http, an old path: send each to the original with a 301. Once that is done, Search Console lists the old addresses as Page with redirect, and there is nothing left for Google to decide.
- Tag the copies that must existA quote link with ?utm_source=linkedin still loads the page. That copy’s tag should name the original as a full URL, in the head of the HTML. Google ignores the tag anywhere else.
- List only originals in the sitemapIt is the weakest signal, but it should never name a different URL from the one the tag names.
- Link to the original everywhereMenu, footer and links in your copy should all use the same address. Every internal link to the other version is a small vote for it.
- Make look-alike pages different, or merge themIf Google kept a different page, no tag will fix it. The pages have to differ. Give each location page what only it has: the team that covers it, response times, the sectors you work with there, the questions clients there ask. If two pages really do say the same thing, keep one and 301 the other to it.
Do not add noindex to a copy, or block it in robots.txt, to force Google’s hand. Google says noindex “will completely block the page from Search,” and tells you not to use robots.txt for canonicalisation.
Then give Google time to look again. Request indexing for the few URLs that matter most (there is a quota), or press Validate fix in the report. Google says it “might hold pages in a duplicate cluster for up to two weeks” after a fix, and validation “typically takes up to about two weeks.” Check the indexed result in URL Inspection after the next crawl. If the URL turns up under crawled - currently not indexed instead, that is a separate problem with its own guide.
Google and AI tools need to know which URL is yours before they can recommend it. Lining up a site’s redirects, tags, sitemap and links is the kind of change we make every working day for service businesses, on one plan at £1,750 a month. See what the plan includes.
Questions
Is duplicate content bad for SEO?
Some of it is normal. Google says “some duplicate content on a site is normal and it’s not a violation of Google’s spam policies.” It groups pages that look alike and indexes one. The cost is that Google, not you, decides which address appears.
How is this different from duplicate without user-selected canonical?
It comes down to whether your page named a canonical. In duplicate without user-selected canonical, the page “doesn’t indicate a preferred canonical page,” so Google chose one for it. In this status, you named one and Google picked a different URL.
Can two canonical tags on one page cause this?
Not this status. Google’s 2013 guidance is that it will likely ignore every canonical tag on a page that has more than one, so the page effectively names none. Remove the extra tag.