"Duplicate without user-selected canonical": you counted pages, Google counted addresses

There is a row in Search Console’s Page indexing report that owners don’t so much fear as contradict. Duplicate without user-selected canonical, it says — and the owner reading it knows better. Every page on the site was written once. Nothing was ever copied. Whatever Google found, it cannot be duplicates.

An owner asked Google’s own help community, in 2024:

An owner in Google’s help community, as reported

For daysknob.com, Search Console shows 60 instances of “Duplicate without user-selected canonical” as a reason for not indexing. Where did this come from? I am not aware of any duplicate pages on this website.

(“Indexing”, if the word is new: to index a page is to add it to the catalogue Google answers searches from.) This article’s answer is an odd one: everyone in that thread was right. The owner was right that no page was ever copied — an answerer who went and looked found the site answering identically with www. and without, no preference stated, both spellings listed. The duplicates were real; they were just never pages. You counted pages — things you wrote, one each. Google counts addresses — every distinct spelling that answers with content. When the same content answers at two addresses, that is what Google’s vocabulary calls a duplicate; no copying is involved. Its Search Console glossary files the word under the entry “Duplicate URL”:

Google’s Search Console glossary, defining a duplicate URL

When Google finds multiple URLs on the same site that show essentially the same page contents, it considers these URLs to be duplicates.

URLs — addresses — not pages. Two more translations. The canonical is the main address: the one spelling chosen to stand for the content in results. A page can name its own main address, and the standard way is a tag — one short line in the page’s code, invisible to visitors, shown in full later. And the “user” is you: the person running the site, who could have named the main address in advance — and, says this row, didn’t. So Google chose one by itself. That choice, not any copying, is the event this row records.

What the row reports, word for word

The reason’s full definition sits on Google’s help page for the Page indexing report:

Google’s help page, defining the reason

This page is a duplicate of another page, although it doesn’t indicate a preferred canonical page. Google has chosen the other page as the canonical for this page, and so will not serve this page in Search. You can Inspect this URL to see which URL Google considers canonical for this page.

Notice what the definition never claims: that your content is missing from Google. It is on Google’s list — under the address Google chose. What stays out is the runner-up spelling. The definition’s second paragraph, on the same help page, acquits you in its first sentence:

Google’s help page, continuing the definition

This is not an error, but is working as intended, because Google does not serve duplicate pages. However, if you think that Google has chosen the wrong URL as canonical, you can explicitly mark the canonical for this page. Alternately, if you think that this page is not a duplicate of the Google-chosen canonical, you should ensure that the content differs substantially between the two pages.

Not an error, working as intended. So the row leaves you exactly one question, smaller than the row looks: is the address Google chose the one you would want people landing on? Usually it is — the clean, ordinary spelling, the one on your business card — and nothing further waits behind the row. Sometimes it isn’t — the sections below take that case seriously, including the one version with no switch to flip. The question is settled by reading two lines on one screen, and this article walks there.

Is this why your traffic fell?

Many owners meet this row while hunting for the cause of a drop. My judgement, plainly: if your traffic fell, this row is very unlikely to be the reason. What it files are addresses whose content is already listed under another address — and declining to list a runner-up spelling takes away nothing a visitor could have landed on. In a thread from 2019, an owner watched four hundred addresses land in this row — “google has removed 400 of my pages”, as he put it — then checked the part that actually matters: “I checked them in search and they are still all listed in google with the same rankings.” Removed, and still findable — because what left Google’s list was extra spellings, not pages.

Two things follow. If a page that matters to you seems to have genuinely left Google, this row is not where that story is told: paste that page’s own address into the inspection tool below, and read the reason the report gives for it. And the count on the row — the 60 that opened this article, the 400 above — is a count of extra spellings, not a severity scale.

Where the second address comes from

You probably never made a second address, any more than you made a second page. They get made for you — by your platform, your links, your site’s own history. An answerer in Google’s help community, in 2024, listed the usual suspects in one breath:

An answerer in Google’s help community, as reported

This may not need “fixing” It means Google found two pages with substantially similar content and decided to index one of them. Frequently the differences are very subtle such as http://www.example.com and https://www.example.com or https://www.example.com and https://www.example.com/index.htm or https://www.example.com/about-us and https://www.example.com/about-us/

Very subtle is the load-bearing phrase: each pair in that list can show a visitor the identical screen, which is why you can click through your site all evening and truthfully find no duplicates — they were never on the screen. Beside those spelling pairs, three more shapes worth recognising:

• Nothing to do Addresses with a ? tail — a parameter: a piece of the address that adjusts what the page shows, like a sort order or a tracking label; Google’s help page names sites using “parameters to filter or sort a common collection” among the common reasons whole groups of pages go unlisted.
• Nothing to do Machine editions published beside the page — a feed (the machine-readable edition of your posts that apps subscribe to), or the PDF copy of a product sheet. In one thread an answerer read the owner’s screenshot: “we can see many URLs are /feed/ or PDF file only.”
• The one to read closely A second front door — your site reachable under a builder’s address and your own domain. Google Sites owners meet this one; an answerer explained one such case in 2025: Google met the site at its sites.google.com/view/... address first, a later custom domain left it reachable both ways — and the pick landed on the builder’s address. The hardest case here; its verdict is near the end.

Ask Google to name the pair

An honest note first: this article carries no worked example of my own — on the sites available to me, this reason has never held an address I could open and examine. What follows replaces that example on the site that matters more anyway: yours.

First, open the row’s list. Click the reason’s row in the Page indexing report; the affected addresses sit in the Examples table — a sample, not a census; Google’s help page says of it: “The list does not necessarily show all URLs with that issue, and is limited to 1,000 rows.” Read them as spellings, with the list above in hand — the www., the ? tail, the /feed/, the builder’s domain; usually they identify themselves at a glance.

Second, ask Google to name the pair. Copy one affected address and paste it into URL Inspection — the search bar at the top of any Search Console screen — and press Enter. Expect the verdict URL is not on Google — not new bad news; it is the row restated. The information you came for is lower down: the Page indexing section, which arrives closed — click its bar. Inside are the two lines this whole walk is for — the help page for the report points at the exact spot, “under Page indexing > Google-selected canonical” and “under Page indexing > User-declared canonical”.

• User-declared canonical is the preference your page states — for this row expect it to state none: the “without” in the reason’s name, confirmed on screen. The empty field is not a fault; it is the row’s cause, located. A page filed under this row is one of a set of alternate spellings, and for exactly that case Google’s note on the field recommends declaring the main address explicitly. Filling it is the next section.
• Google-selected canonical is the pick — the same page defines it as “The page that Google selected as the canonical (authoritative) URL when it found similar pages on your site”: the other half of your pair. One limit before you trust the screen: the same page says the tool provides this information “if the canonical version is in a property that you own” — a property being a site you have registered in Search Console. For the usual pair — two spellings of your own domain — the line simply names the pick; when the pick landed on a builder’s domain instead, a screen naming no pick is this limit showing. One way to see the pick anyway costs one search: look up your page’s exact title on Google — Google’s documentation on canonicalization notes that “A Google Search result usually points to the canonical page.”

Third, judge the pick. Open the Google-selected canonical in a browser.

• It is your normal, clean address — the one you would print on paper. The row is settled: your content is on Google at the right address; the “duplicate” is a runner-up spelling nobody will miss.
• It is an address you would never hand out — a tail, an old spelling, a domain not yours. This is the rare version of the row that deserves your evening; the next section is the fix, and for the builder’s-domain case it ends with the straightest answer the evidence allows.
• The “pair” is two pages you meant to be different. Then no address is at fault — the content is too similar for Google to tell apart, and the definition’s own advice applies: make the two pages differ substantially, or accept that Google will keep listing only one.

Last, know what you cannot check today. URL Inspection also has a live mode, usually the fastest way to confirm a fix. Not for this one — Google’s help page rules it out in as many words: “Most notably, duplicate or canonical conditions are not tested in the live test.” Whatever you change, the answer arrives on Google’s schedule, not yours.

Naming the main address yourself

What Google says about doing nothing is unusually permissive — from its page on stating a canonical preference:

Google’s documentation, on whether you must state a preference at all

While we encourage you to use these methods, none of them are required; your site will likely do just fine without specifying a canonical preference.

If you want the choice back — because the pick was wrong, or because you would rather it never be open — the standard instrument is the tag from this article’s beginning: one line in the page’s code, invisible to visitors, naming the main address. Google’s example of the line (the address in it is Google’s illustration, not a real shop):

<link rel="canonical" href="https://example.com/dresses/green-dresses" />

Read as a sentence: whatever address you arrived by, this content officially lives at this one. The likeliest twist: your platform probably writes that line already.

• Nothing to do WordPress writes it by itself on individual posts and pages — its developer reference files the responsible function under “Outputs rel=canonical for singular queries.”
• Nothing to do Yoast SEO, per its feature page: “Yoast SEO adds a canonical URL tag to every page and post by default, based on that page’s own URL.”
• Nothing to do Rank Math, per its guide to this exact row: “By default, Rank Math uses the current post/page URL as the canonical URL”.
• Nothing to do Shopify lists it among built-in features on its SEO overview: “Auto-generated canonical tags are added to pages to prevent duplicate content from appearing in search results”.
• Nothing to do Wix, per its help page: “Wix automatically creates a default canonical URL for all of your pages.”
• The exception Google Sites gives you no way to write it. An answerer told an owner so, in the Sites community in 2025: “The new Google Sites does not provide a canonical link element for the pages or any other way to explicitly declare what you want the canonical version of a page/site to be.”

That owner — and you, if the pick on your pages landed on sites.google.com — deserves this article’s plainest paragraph, promises removed. I could not establish a fix for this case. The platform offers no canonical setting, and none of the threads this article rests on ends with a Sites owner’s own domain becoming the pick. What it costs, and what it does not: your content is on Google, and people who search can find it — but the address they land on reads sites.google.com/view/... rather than the domain you bought. Google’s canonicalization documentation calls a stated preference “a hint, not a rule”, and picks can change — but nothing in the record I can show you promises this one will, so I won’t. If the address in the results matters enough to you, the realistic route is hosting the site somewhere that lets you write the one line this section is about.

If platforms write the line anyway, why did your row appear? Usually because the listed addresses are the ones no platform reaches — machine editions like feeds and PDFs, or a second domain your platform doesn’t know it answers under. The check costs under a minute at a computer: open a page, view its source code — the right-click menu offers it — and search for canonical. A line like Google’s example, with a real address inside it, means your platform is already answering the question page by page. Finding no such line, on a page filed in this row, is the “without” made visible (a preference can also travel outside the page, in a technical header or a sitemap). And when the runner-up address shouldn’t exist at all — an abandoned spelling, a retired domain — the cleaner tool is a forward from the old address to the main one, which has its own article in this series.

And if none of this is your world — someone else built the site, and “view source” was already one step too far — the job you would be handing over is small, routine work, not a project. The whole request fits in one sentence: Search Console files some of our addresses under “Duplicate without user-selected canonical” — please make each page declare the address we want listed, with a canonical tag.

What happens after you answer

Don’t watch for the row to empty. Watch for it to change shelves. An answerer in Google’s help community told an owner what to expect once the line is in place:

An answerer in Google’s help community, as reported

Set the canonical to the version of the URL you want indexed. It will move these to “Alternative page with proper canonical tag” over time, so no real big difference, but you are not making Google guess at to which version should be indexed.

The shelf that reply names — the help page’s spelling is Alternate page with proper canonical tag — is the neighbouring reason where labelled spares are filed, and it has its own article too: the row you get to close on sight. Success looks like the same addresses reappearing one shelf over, wearing your label instead of Google’s guess. Which answers the button on the row’s page: VALIDATE FIX is not the instrument for this reason in either direction — if you agreed with the pick, there is no fix to confirm; if you named the address yourself, you are waiting for the addresses to re-file next door, at the pace of Google’s re-reading.

Two timing facts, both Google’s own. Its troubleshooting page for canonical issues says of re-checks: “Even after fixing content issues, Google might hold pages in a duplicate cluster for up to two weeks.” And your named preference is “a hint, not a rule” — the sentence quoted above — so Google may still overrule it; the report keeps a separate shelf for that outcome, Duplicate, Google chose different canonical than user on the same help page. For this row, the loop closes where the seeing started: the two fields in URL Inspection. When User-declared canonical shows the address you named and Google-selected canonical agrees with it, the word in the reason’s name that no longer applies is without. The row has nothing left to say about it.

Written by Oleh Koval at Specia1ne, who reads these reports for a living.