Use a canonical tag when two or more live URLs show the same content and you want one of them to be the main one. Delete the copy or redirect it when it has no job of its own. Fix the page itself when the copy points to a real content problem.
What is a canonical tag, in plain words?
A canonical tag is a short line of code in the head of a web page. It names one URL as the main copy of that content. The tag uses rel="canonical". Search engines read it as a hint when they find two or more pages that look the same. It is a hint, not an order. A search engine can pick a different URL if its own signals point that way.
Think of it like a label on a file. You have three copies of one file in three folders. You put a note on two of them: the main copy lives in the first folder. People can still open the other copies. It tells the reader which copy counts as the main one.
A canonical tag does not move a reader. A person who lands on a copy stays on that copy. A canonical tag does not delete a page. It does not hide a page from users. It does not fix weak copy. It does one job: it states your pick for the main URL among look-alike pages.
The tag looks like this in the head of a page: <link rel="canonical" href="https://seoelinks.com/page/">. The href holds the full URL of the main page. Use the full URL, with https, not a short relative path. The main page should point to itself with the same kind of tag. That is called a self-canonical, and it is the normal setup for a page that has no copies.

When should I canonicalize instead of deleting the duplicate?
Canonicalize when the copy has to stay live and the content is the same. That is the core test. Two parts, both must hold. The copy stays live for a real reason, and the content matches the main page in substance. If one part fails, a tag is the wrong tool.
Print views are a clean case. A print view exists so a reader can print the article. It shows the same text with no sidebars and no ads. It has to stay live, or the print button breaks. Point its canonical tag at the main article. The same logic fits a PDF copy of a page and a share link that adds tracking marks to the URL. The copy serves a use. The main page keeps the search value.
Tracking marks are the most common case on a working site. A link from an email or an ad adds a string like ?utm_source=newsletter to the URL. The page is the same. The mark only tells you where the click came from. Point the marked URL back at the clean URL with a canonical tag. Do not delete the marked URL. It is not a page. It is the same page with a note attached.
Sort and filter views fit the same shape when the filter adds no new buyer value. A shop page sorted by price shows the same products in a new order. A list filtered by one colour shows a subset of the same list. The base page is the main copy. The sorted and filtered views point at it. But a filter page can grow into a real landing page that people search for and buy from. Give that page its own content and let it stand on its own.
Host and path variants are the quietest case. The same page loads at http and https, with www and without, with a trailing slash and without. Your server should settle on one form and send the rest to it with a redirect. Fix the server rule first, then add the tag as a second signal.
The test stays the same in each case. Ask two questions. Must this copy stay live for a real use? Is the content the same as the main page in substance? Yes to both means canonicalize. No to one of them means you need a different move, and the next section covers those moves.

When should I fix the duplication instead of adding a tag?
Add a tag when the pages are the same. Fix the pages when the pages are not the same but look alike because the content is weak. A tag cannot turn two thin pages into one strong page. It can only hide the fact that you have two thin pages.
City pages are the classic trap. A site builds one page per city with the same text and a swapped city name. No office in the city. No staff in the city. No local proof of any kind. A canonical tag pointed at one city page does not fix that. The pages are not duplicates of one strong page. They are weak pages with no reason to exist. The fix is editorial: keep one honest service area page, or build real local pages with real local facts, or take the weak pages down.
Near-copy service pages fail the same test. A site has a page for SEO, a page for search engine optimization, and a page for Google ranking services. All three say the same thing in slightly different words. Each page has no fact, no proof, and no next step of its own. A canonical tag would pick one and starve the other two. But the picked page is still thin. The right move is to merge the three into one strong service page that says the thing once, says it well, and carries the proof.
Old posts that repeat a new guide are a third case. A site wrote a short post in 2021, a longer post in 2023, and a full guide in 2025, all on the same topic. The old posts still rank a little. The new guide is the best copy. A reader who lands on a 2021 post gets stale facts and a tag they cannot see. The honest move is to update the old post, merge it into the guide with a redirect, or rewrite it so it covers a slice the guide does not. The content refresh workflow walks through that call page by page, and it is the right tool for this job.

What does a canonical tag do, and what can it not do?
A canonical tag does three things. It tells a search engine which URL you pick as the main copy. It helps the search engine group look-alike URLs together instead of treating each one as a separate page. And it helps link signals point at one URL instead of scattering across five copies. That is the full list. It is useful, and it is narrow.
It cannot force a search engine to pick your URL. The tag is a hint. The search engine weighs it with its own signals: which URL has the links, which URL sits in your sitemap, which URL you link to from your own pages, which URL people visit. If those signals point at a different URL, the search engine will pick that one. Your tag loses. That is normal. A tag is a vote, not a veto.
It cannot delete a page from the index. A page with a canonical tag can still show up in search results. A search engine may index the copy if it thinks the copy is the better match for a query, or if the main page is weak, or if the tag is wrong. If you need a page out of search results, a tag is the wrong tool. Use a noindex, or take the page down. Each of those is a separate call with its own trade-offs.
It cannot move a reader. A person who clicks a link to a copy lands on the copy. The tag is invisible to them. A redirect moves a reader. A tag does not. When the old page should no longer be read at all, redirect it. Do not tag it and hope the reader finds the new page on their own.
How do I pick the main URL?
Pick the URL you would give to a customer. That is the test that cuts through most debate. If a customer asks for the page, which link do you send? That link is the main URL. It is usually the shortest clean URL, on https, with no tracking marks, no sort string, and no session ID.
First, the main URL is the one your own site links to. Open your main menu, your footer, and your in-text links. They should point at the main URL, not at a copy. If your own links point at the copy, fix the links. A tag cannot outvote your own navigation. The internal linking guide covers how to point your links at the pages that should earn.
Second, the main URL is the one in your sitemap. Your sitemap is a list of the URLs you want found. It should hold the main URL and leave out the copies. A sitemap that lists a copy while the tag points elsewhere sends a mixed message. The sitemap governance guide covers how to keep that list clean as a site grows.
Third, the main URL is the one with the links. When other sites link to you, they link to one form of the URL more than the rest. That form has the link value. Picking a different form as the main URL throws that value away. Check which form has the links before you pick.
Fourth, the main URL is the one that works best for the reader. It loads fast, it shows the full content, it has the working form and the working next step. A main URL that loads slowly or shows a stripped-down copy is a poor pick, even if it is the shortest address.
Once you pick, make the pick visible in all the places that count. The main page carries a self-canonical tag that points at itself. The copies point at the main page. Your internal links point at the main page. Your sitemap lists the main page. Four places, one URL, no mixed message.
Should I redirect the duplicate or canonicalize it?
Redirect when the old page is done and a new page does the same job. Canonicalize when both URLs must stay live. That is the split. A redirect moves the reader. A canonical tag does not. Pick the tool that matches what the reader should feel.
Redirect fits a page that moved. The old service page has a new address. The old blog post merged into a new guide. The old product page has a direct replacement with the same stock and the same price. A permanent redirect, a 301, is the way to show a page has moved for good. The redirect documentation covers how permanent redirects pass the old page’s standing to the new one.
Canonicalize fits a page that must stay reachable. The reader needs that address to work. The search engine needs to know which copy is main. A tag does that job without breaking the live address.
Do not redirect to the home page as a catch-all. A visitor who wanted a guide and lands on a home page has been sent to the wrong place. If no true replacement exists, leave the old URL to return a proper not-found page, or rebuild the content. A redirect is a promise that the new page continues the old page’s job. Keep that promise or do not redirect.
Use the table below to make the call in under a minute. Find the row that matches your case and use the move it names.
| Your case | The move | Why |
|---|---|---|
| Same content, copy must stay live | Canonical tag | The copy serves a use, and the main page keeps the search value. |
| Old page is done, new page does the same job | Permanent redirect | The reader lands on the true replacement, and the old address retires. |
| Same promise, weak content on all copies | Rewrite or merge | A tag cannot fix thin copy. One strong page beats three weak ones. |
| Page has no job at all | Remove or noindex | A page with no reader job should not sit in search results. |
| Link between pages is still unclear | Defer and note it | A guess baked into a redirect or a tag is hard to undo later. |
When a whole site moves, a single page call is not enough. A domain change, a host change, a platform move, or a large URL restructure needs a map, a staging check, and a watch list. The migration checklist covers that larger job, and it is the right guide when the move touches more than a few URLs.

What breaks a canonical tag most often?
Most broken canonical tags are not broken code. They are mixed messages. The site says one thing in the tag, another thing in the sitemap, and a third thing in its own links.
Mixed signals are the top cause. The tag points at URL A. The sitemap lists URL B. The main menu links to URL C. Each one is a vote. Three votes for three different URLs means no URL wins clean. Fix it by picking one URL and lining up the tag, the sitemap, and the links behind it. The canonical URL guide lists the signals that should agree, and it warns against naming different main URLs through different methods.
A tag in the wrong place is the second cause. The tag belongs in the head of the page. A tag in the body of the page is easy to miss, and some crawlers will not read it there. A tag added by JavaScript after the page loads is worse: a crawler that reads the raw HTML never sees it. Put the tag in the server-sent HTML, in the head, on each page that needs one. Check the raw source, not only the rendered page in your browser.
A relative URL is the third cause. A tag that points at /page/ instead of the full https://seoelinks.com/page/ can resolve to the wrong address when the page is reached through a copy, a feed, or a cached copy. Use the full absolute URL with https. It costs nothing and it removes a whole class of errors.
A tag that points at a dead end is the fourth cause. The main URL redirects to another page, returns a 404, or is blocked from crawling. A tag that points at a redirect chain is a pick for a page that does not exist at that address. A tag that points at a blocked page is a pick no crawler can follow. Check the main URL first: it loads, it returns a 200, and it is not blocked.
Two tags on one page is the fifth cause. Two SEO plugins, a theme and a plugin, or a plugin and a hand-added tag can each add their own canonical element. The page then names two main URLs. Pick one source for the tag and turn the other off. Check the raw source to see how many tags the page sends.
How do I check that the tag works?
Start with the raw source. Open the live page, view the page source, and search for canonical. You should find one tag, in the head, with the full URL of the main page. If you find none, the tag is not live. If you find two, two sources are adding tags. If you find one in the body, it is in the wrong place. The raw source is the test because it is what a crawler reads first.
Next, follow the href. Open the URL named in the tag. It should load, return a 200 status, show the full content, and carry a self-canonical tag that points at itself. If it redirects, the tag points at a moving target. If it 404s, the tag points at a dead page. If it shows a stripped copy, the tag points at the wrong page. A tag is only as good as the page it names.
Next, check the sitemap. The main URL should be listed. The copies should not be listed as main entries. A sitemap is a list of the URLs you want found, and it should agree with the tag. If the sitemap lists the copy, take the copy out or fix the tag, so the two agree.
Next, check the index state. Search Console shows which URL a search engine picked as the main copy and whether the copy was left out of the index. Give the change time to be crawled, then read the report. The indexing triage guide walks through how to read those states and what each one means for your next move.
Keep a short record of the check. The canonicalization documentation describes how a search engine picks a main URL from a set of look-alike pages, and it is worth a read before you lean on a tag for a large set of pages. Your record should hold the URL set, the main URL, the tag state, the sitemap state, the link state, and the date of the check. Recheck after each change that touches the URLs. Tags break quietly.
What about faceted pages, print pages, and URL parameters?
Faceted navigation is where most large sites build their duplicate sets without noticing. A shop category with filters for size, colour, price, and brand can make thousands of URLs from one list of products. Each filter mix gets its own address. Most of those addresses have no search job. They exist so a shopper can narrow a list, not so a search engine can index a new page.
The rule for a filter page is the same as for any copy: does it have a job of its own? A filter mix that people search for, link to, and buy from is a real page. Give it a title, a short intro, and links, and let it stand on its own. A filter mix that no one searches for and no one links to is a view of the base list. Point it at the base category with a canonical tag. The base category is the main copy. The filter view is a way to look at it.
Tracking marks deserve a firm rule. A mark like ?utm_source=newsletter changes nothing on the page. It only notes where the click came from. The marked URL should point at the clean URL with a canonical tag. Never point the clean URL at a marked copy. Never let a marked copy into your sitemap. And never build links on your own site that carry tracking marks. Your own links should use the clean address.
Pagination needs a different call. Page 2 of a list is not a copy of page 1. It shows different items. Each page in a set should point at itself with a self-canonical tag. Do not point page 2, page 3, and page 4 at page 1. That tells the search engine that the later pages are copies of the first, and the items on those pages lose their main copy. Let each page stand as the main copy of its own items.

Common questions
Does each similar page need a canonical tag?
No. Likeness alone does not prove that one page stands for the other. First work out whether the pages are the same in substance, whether they serve different reader jobs, or whether one page should retire into a direct replacement. A canonical tag fits copies that must stay live. It is not a swap for that call.
Should each old page redirect to the home page?
No. A permanent redirect should send a visitor to a direct replacement that does the same job. If no true replacement exists, a home page redirect sends the visitor to the wrong place. Leave the old URL to return a proper not-found page, rebuild the content, or pick a page that does the same job.
Can a self-canonical tag force a search engine to pick my URL?
No. A canonical tag is a hint, not an order. A self-canonical tag is a clear signal from the site owner when the page is the main copy, but the search engine weighs it with links, sitemap entries, and internal links. It does not promise a pick, a crawl date, an index entry, or a ranking.
Can a canonical tag make thin city pages safe?
No. A tag does not create a real local office, a distinct reader benefit, or a supportable public claim. Decide whether the pages should exist and whether the business facts support them before you touch any tag. Weak pages need editorial work, not a technical label.
When should a team hold off on a canonical change?
Hold off when no one can name the current owner, the link between the pages, a direct replacement, or the right reader path. Write down what is missing, keep the current state under watch, and make a narrow change only when the link between the pages is plain. A guess baked into a live tag is hard to undo later.
Make the link plain before you make it technical
A strong canonical call starts with a plain sentence a reader could follow. One page either continues a retiring page’s job, stands as the main copy of the same content, or gives a different person a different useful answer. Once that link is real, the technical move is easy to pick and easy to check. When the link is weak, a tag or a redirect can hide the gap for a while, but it cannot close it.
Work through the set one pair at a time. List the URLs. Write one sentence for what each page helps a visitor do. Sort the pairs into copies, distinct pages, retiring pages, and unclear cases. Redirect the retiring page to its true replacement. Tag the copy that must stay live. Rewrite or merge the weak pages. Leave the unclear cases alone until the proof exists.
If the set is large, or the site is about to move, get a second pair of eyes on it before you change a live URL. A wrong redirect or a wrong tag applied across hundreds of pages takes far longer to undo than a single page call. Our SEO services cover technical audits, migrations, and indexation work, and the free audit below is the place to start.
Not sure which pages are copies and which ones earn their keep?
Get a free audit: a 120-point crawl of your site, a competitor gap map, and a 90-day roadmap, delivered within 48 hours. Free, with no obligation.
Get a free SEO audit of your website
A 120 point crawl, a competitor gap map and a 90 day roadmap, ready in 48 hours. Free, with no obligation.
