“Duplicate, Google chose different canonical than user”: what it means and how to fix it

What does this Search Console status actually mean?

Google’s Page indexing report documentation defines it in one line: this page is marked as canonical for a set of pages, but Google thinks another URL makes a better canonical. The URL you see in the report is the one you declared. It is not indexed under that address; Google has folded it into a cluster of duplicates and shows a different member of the cluster in search.

The status sits next to three relatives that are easy to confuse:

The reason the first one matters on a small site is concentration. A national publisher with 40,000 near-identical product URLs can lose a few to canonical overrides without noticing. A Denver plumber with 30 pages cannot afford Google indexing the Aurora page when someone searches for Denver, or indexing a tracking-parameter URL instead of the clean service page.

How does Google choose a canonical URL?

Google first groups pages it considers duplicates “based on similarity of content,” then picks one per group. Its canonicalization guide (last updated July 10, 2026) ranks the signals you control:

• Redirects are a strong signal that the redirect target should become canonical.
• rel="canonical" link annotations are a strong signal that the specified URL should become canonical.
• Sitemap inclusion is a weak signal.

The guide also says Google prefers HTTPS over HTTP unless there are problems such as an invalid certificate, and prefers URLs that are part of a reciprocal hreflang cluster. It tells you not to use robots.txt for canonicalization and not to use noindex to choose a canonical within one site, because noindex removes the page from Search entirely.

Two things follow. First, the canonical tag is a strong hint, not a command; when other signals disagree, they can outvote it. Second, the most common “fix” we see, adding a second canonical tag through a plugin, makes things worse, because conflicting tags are themselves a signal Google learns to ignore.

How common are canonical tags, and where do they go wrong?

The HTTP Archive measures millions of pages every year. Its Web Almanac 2025 SEO chapter (published January 15, 2026) found canonical tags on 68% of desktop and 67% of mobile pages, up from 65% on both in 2024; the 2024 chapter reported 59% desktop and 61% mobile in 2022.

Adoption is not the problem. Consistency is. The same chapter measured how often the canonical a crawler sees depends on how it looks at the page:

Three percent sounds small until you remember what it means: on roughly one mobile page in thirty-three, JavaScript changes the canonical after the HTML arrives. If Google reads the raw HTML and the rendered page and they disagree, you have handed it two answers. Our JavaScript SEO page covers why we keep canonicals in the server HTML and never let a script touch them.

What causes it on local business sites?

These are the causes we check first on service-business sites.

1. City pages that are the same page

The classic: /denver-roof-repair/, /aurora-roof-repair/ and /lakewood-roof-repair/ with the city name swapped and nothing else changed. Each declares itself canonical. Google, judging by content similarity, decides they are one page and picks one. Google’s documentation is blunt that a duplicate must be similar to the canonical, and the reverse is also true: pages that are nearly identical will be treated as duplicates whatever their tags say. The fix is not a tag. It is giving each location page content only that location has: the crew that works there, local projects, local reviews, the permits and HOA quirks of that city. Eye To Ad Media has a longer piece on whether location pages are doorway pages, with a similarity checker.

2. URL variants nobody consolidated

http and https, www and bare domain, /services and /services/, /Services/ and /services/, /index.html and /. If two versions both return 200, you have duplicates by construction. The fix is a site-wide 301 to one form, internal links that all use that form, and a canonical that matches it exactly.

3. Tracking parameters that leak into internal links

Adding UTM parameters to the website link in your Google Business Profile is a good habit for measurement. Copying that tagged URL into your own navigation, a footer, or an email signature that gets crawled is not. When a page is linked internally with ?utm_source=gbp more often than without it, the parameter version collects the signals. Keep UTMs on external links only, and make the canonical point at the clean URL.

4. The canonical points somewhere broken

A canonical that points to a URL which redirects, returns 404, is blocked, or carries noindex is a contradiction. Google cannot index the target, so it picks something else. This happens after redesigns, when templates keep the old URL pattern in the canonical tag.

5. The sitemap and internal links argue with the tag

The page says canonical is /water-heater-repair/. The sitemap lists /water-heater-repair-denver/. The menu links to /services/water-heater/. Each of those is a vote. Make them all the same URL.

6. Your content lives on someone else’s domain too

Press releases, directory profiles with your full service descriptions, and franchisor sites that copy local pages can all win the cluster. If the copy is on a stronger domain, Google may choose it. Publish the original first, ask partners to link to it or use a cross-domain canonical, and do not hand directories your best paragraphs verbatim.

7. A staging or old domain is still reachable

A staging copy at dev.example.com or an old domain that still serves 200s is a perfect duplicate. Password-protect staging, and 301 old domains page-to-page.

How do I fix it, step by step?

1. Inspect, do not guess. In Search Console, open URL Inspection for the affected page. Under Page indexing, note the user-declared canonical and the Google-selected canonical. Open both in a browser.
2. Decide who is right. Sometimes Google is. If it picked the https, trailing-slash version of your own page, your tag is the thing that is wrong; change the tag. If it picked the Aurora page for the Denver page, or a parameter URL, your site is sending mixed signals.
3. Pick one URL per piece of content and write it down: protocol, host, path, case, trailing slash.
4. Make the strong signals agree. 301 every variant to that URL. Put a single, absolute, self-referencing canonical in the server-rendered head. Remove duplicate canonical tags from plugins or themes.
5. Make the weak signals agree. Update internal links in menus, footers, breadcrumbs and body copy. List only canonical URLs in the sitemap. Strip parameters from internal links.
6. If the pages are genuinely different, prove it. Add the content that makes the page about its own topic or place. Two pages with the same 400 words and a different H1 will keep merging.
7. Validate and wait. Use Validate Fix in the report where it is offered, or request indexing on the main URLs. Canonical decisions update as Google recrawls, which takes days to weeks; do not change the tags again while you wait.

Our crawled, currently not indexed note covers the neighboring status, and the technical SEO audit checklist includes these canonical checks among its 40.

What about Spanish pages and other language versions?

Plenty of Denver businesses publish a Spanish version of their key pages, and language versions are a frequent source of this status. If /es/reparacion-de-techos/ is mostly the English page with a translated headline, Google may see the two as duplicates and pick one. Google’s canonicalization guide says it prefers URLs that are part of a reciprocal hreflang cluster, which gives you the right setup:

• Each language version declares itself as canonical. Never point the Spanish page’s canonical at the English page; that tells Google the Spanish page is a duplicate to be dropped.
• Each version lists every language version, including itself, in hreflang annotations, and each of those versions lists it back. One-way hreflang is ignored.
• The content is actually translated, including the body, the calls to action and the form labels. A page that is 80% English with a Spanish title is not a Spanish page.

The same thinking applies to the website link in each Google Business Profile. If you have two locations, link each profile to that location’s page, not both to the home page, and keep any UTM tags on the profile link only. Then the location page you want to rank is the one receiving the profile’s visits, and its canonical matches the clean URL your own site links to.

When should I leave it alone?

When Google is right and nothing is lost. If the declared URL is a printer view, a sorted listing, or a filtered category whose content really is the same as the main page, Google folding it into the main page is the outcome you wanted. Check that the URL Google chose is the one you would want in search results and that it gets the clicks in the Performance report. If so, update your tag to match and move on. The status is information about how Google sees your site, not a grade.

How do I check a whole site, not just one URL?

Export the affected URLs from the Page indexing report. For each, you want four values in one row: the URL, its canonical tag, the URL it redirects to (if any), and whether it appears in the sitemap. Any crawler that renders JavaScript can produce the first three, and comparing raw and rendered canonicals catches the scripts that rewrite them. Where the declared and chosen URLs differ only by a parameter or a slash, fix the pattern once in the template or server config rather than page by page. Our free internal link graph is useful here to see which version of a URL your own links favor.

If you would rather hand it off, our technical SEO work starts with exactly this kind of signal audit, and the fixes ship as code, not recommendations.