Setting a canonical strategy for repurposed content across several event sites

An editor at a six show portfolio ships the same 1,100 word guide to planning a first exhibition stand on four brand sites, swaps the venue name in the intro, and asks a developer to sort out the canonical strategy for repurposed content so that search does not punish anybody. Three rel=canonical tags go on, pointing at the fourth. Everyone moves on to the next thing.

Six weeks later two of the three have been honoured and the third has not. Google picked its own canonical for that one, and the version it picked is on the brand that had the smallest audience.

What a canonical actually claims

Google's documentation on consolidating duplicate URLs, current in 2026, lists the ways you can influence which URL gets indexed "in order of how strongly they can influence canonicalization". A redirect comes first, described as "A strong signal that the target of the redirect should become canonical". The rel=canonical annotation comes second, also a strong signal. Inclusion in a sitemap comes third and is "A weak signal".

None of the three settles the question. Where you declare nothing, the same documentation says Google "will identify which version of the URL is objectively the best version to show to users in Search", and that is the same procedure it runs when it decides your declaration was wrong.

So the tag is an argument you are making, and it only carries when the two pages really are copies of each other. Everything that goes wrong with cross brand canonicals goes wrong because the pages were never copies.

It is worth being precise about what a successful canonical buys. The signals that point at the copies get attributed to the chosen URL, and one page competes in search instead of four. That is the whole benefit. The copies stay live, stay linkable, stay in your navigation and stay in your newsletters, so a reader can still land on three versions of the same guide in a fortnight and form her own view of how much original work your titles do.

Where the tag gets used on pages that are not duplicates

Joshua Hardwick's Ahrefs guide to keyword cannibalization, updated in January 2024, is blunt about this: "Canonicalization is not a way to fix keyword cannibalization." He reserves the tag for near duplicate content. His companion guide to canonical tags, updated in October 2024, describes the case the tag was built for as "the same or similar content available at multiple URLs", meaning tracking parameters, trailing slashes and the same article served twice.

The failure in an event portfolio looks different from a parameter problem and gets the same treatment anyway. Two shows in adjacent verticals both publish a piece on packaging compliance. The pieces share a source, share four paragraphs, and diverge on the examples because the audiences differ. Somebody notices they look similar and canonicalises one to the other.

If Google honours it, you have removed from the index a page that was answering a question the surviving page does not answer, and the brand that lost the page has no way of knowing why its traffic fell. If Google overrules it, the tag sits there making a claim your own analytics contradicts. Neither outcome is a consolidation.

There is a quieter version of the same defect that comes from the publishing system. Several event CMS templates emit a canonical automatically, usually a self-referencing one, and a syndication module then overwrites it with the source URL on every imported article. Nobody wrote that policy, and it applies to every piece the module has ever handled. Before you design anything, view source on ten reused pages and read what the templates are already declaring, because a strategy written on top of an unexamined default is a second policy running against a first one.

Three routes, decided per pair

The strategy worth writing down has three destinations and a rule for choosing between them.

• Near identical copies. Same body text, different venue and date. If both must stay reachable because each brand insists on hosting it, put a rel=canonical on the copies pointing at one home and a self-referencing canonical on the home. If one copy can disappear, redirect instead, because Google ranks redirects as the stronger signal.
• Overlapping pages. Shared material, genuinely different framing or audience. These want a merge, and picking the winner and folding the loser into it is a different piece of work with its own arithmetic.
• Distinct pages. Same topic, different question. Leave them alone, give each a self-referencing canonical, and stop worrying about them.

The routing needs a threshold and every threshold here is arbitrary, so pick one and hold it. Overlap above roughly 80 per cent of the shorter page reads as a copy. Below about 30 per cent the pages are doing different jobs. The band between is where the merge decision lives.

Name the denominator before anybody runs the numbers. On a pair where 940 words of body text appear in both a 1,180 word page and a 1,240 word one, the overlap is 80 per cent measured against the shorter page and 63 per cent measured against the union of the two, because the union is 1,480 words. One pair, two honest figures, two different routes. Fix the denominator in the policy and the argument goes away.

How do you tell whether a canonical is being honoured?

Read the page indexing report rather than the tag.

Two statuses matter. "Duplicate, Google chose different canonical than user" means, in the wording of Google's page indexing report documentation, current in 2026, that "This page is marked as canonical for a set of pages, but Google thinks another URL makes a better canonical." Your declaration was overruled. The second is "Duplicate without user-selected canonical", which covers pages where you declared nothing and Google chose for you.

Both are worth counting because they measure different failures. The first says your strategy exists and is wrong. The second says no strategy reached those pages at all.

What an audit of thirty four cross brand canonicals turns up

Take a portfolio of six show sites that has been reusing content for two years. Export the indexing report from every property and filter for pages whose declared canonical sits on a different brand's domain. Suppose that comes to 34 pages.

Of those 34, twelve appear under the overruled status. That is 12 divided by 34, or 35 per cent of every cross brand declaration in the portfolio, doing nothing. A further 61 pages show up as duplicates with no user-selected canonical, which means the reuse happened and nobody declared anything, so Google resolved 61 pairs on its own preferences.

Now read the twelve. In an audit of this shape the pattern is usually the same: the overruled declarations sit on pairs where the two pages share a subject and diverge in substance, exactly the case Hardwick's guidance says the tag was never for. The twelve are not a tagging bug. They are twelve merge decisions somebody avoided by reaching for a tag.

Sort the twelve by the impressions the losing page earned in the twelve months before the tag went on, and work down. In a portfolio of this size the top three will account for most of the loss and the bottom five will be pages nobody would defend, which means the real remediation list is short even when the count looks alarming.

The 61 undeclared pairs are the bigger number and the cheaper fix, because most of them will route to the third destination and need nothing except a self-referencing canonical and a note in the commissioning record. Getting there is a two afternoon job for one person with a spreadsheet, and it converts an argument about search hygiene into a count that a publishing director can act on.

Who loses a page when the canonical points at a sibling brand?

This is where a technical decision becomes a commercial one, and it is the part that gets skipped in the ticket.

A cross brand canonical asks Google to stop serving one brand's page in favour of another brand's. If it works, the losing brand's article keeps existing for anybody who has the URL and stops arriving from search. Six months later that brand's editor is looking at a traffic decline with no obvious cause, because nobody told her that a developer resolved a duplication problem by pointing her page at a sibling.

Settle it before publication. The decision belongs in the same conversation as the test that decides whether a reused asset earns a page at all, and where the second brand does want the material in front of its readers, an excerpt with a link back to the source gives the reader the article and gives search one URL to rank, without either brand quietly losing something.

Write the owning brand into the commissioning record at the point of reuse. One field, filled in by the person who authorised the second copy.

Where this stops

Google can overrule any canonical you declare, and there is no appeal and no notification. That makes the tag a request with a good hit rate, which is fine for the case it was designed for and useless as a control you can promise a board.

The tag also does nothing about the reader. Both URLs stay live and reachable, so a subscriber who follows a newsletter link to the sibling copy reads the same guide twice and the canonical is invisible to her. Canonicalisation is an indexing instrument and it has never been a content decision.

There is a case it handles badly, which is the annual edition page. Two editions of the same show preview are 80 per cent identical by design, and canonicalising last year's to this year's throws away a page that still earns links and traffic from people searching for what happened. Those want an archive policy of their own.

The last limit is measurement. You can count how many declarations were overruled, and you cannot measure what a correctly consolidated pair earned you, because the counterfactual page no longer exists to compare against. Anybody promising a percentage lift from a canonical clean up is guessing.

Export the page indexing report for every brand property you run, filter for the two duplicate statuses, and count how many of your cross domain canonicals Google is currently ignoring. That number tells you whether you have a tagging problem or a content operation problem, and they need different people to fix.