Canonicals – The Good, The Bad, & The Ugly

Canonical URLs can be marked up as a link element within the head of a page, or by rel=”canonical” HTTP headers to signal to search engines which page is the “preferred” URL to rank – usually in situations where content is duplicated or shared across many pages. The use of a canonical tag can help prevent duplicate content, and ensure all indexing and link signals from a set of pages are delivered to the preferred landing page, which should in theory then rank in the SERPs.

In practice things are different, as this tag won’t always be accepted by search engines – much to our dismay! A canonical is a directive to search engines, which means that it will not always be followed. Below we’ve run through a few examples of when experimental canonical tags have worked, failed, or simply been ignored in the good, the bad, and the ugly of canonical link elements.

Example #1 – The Good

A client in the cosmetic surgery industry was experiencing ranking fluctuations surrounding the term “breast implants”, with the site ranking on the first page or as low as the 5^th page of Google interchangeably.

Using Sistrix keyword history we identified that these fluctuations were occurring between 3 different pages: the homepage; a breast enlargement page; and a breast implants page, as shown in the image below. At this stage, there was no clear-cut difference in ranking positions to identify which page should be ranking, or was the most relevant in the eyes of Google – so it was up to us to do some more digging.

Deciding which page should rank came a large part down to an analysis of the Google SERPs for the “breast implants” term. What we determined was that the vast majority of competitor landing pages ranking on the first page were breast enlargement procedure pages. Therefore, it seems that Google determine the breast enlargement procedure to be the most relevant page for the search term. It was clear from reviewing the SERPs that the intent behind a search for “breast implants” was commercial by nature, which is why Google were saving breast enlargement procedure pages for the top spots.

With this in mind we added a canonical link element pointing to our breast enlargement procedure page from the more informational breast implants page. What we subsequently found was a stabilisation and steady growth in rankings as the canonical indicated to Google which page was the most relevant to rank and authority was passed across from the breast implants page, as can be seen below:

Point A in the image above represents the rollout of Penguin 4.0, which also may have had an impact upon the ranking of this keyword (but we’ll conveniently believe confirmation bias and hold the canonical responsible for now!).

There were certain elements of the page content in question that made the differentiation difficult for Google without a canonical; both the breast enlargement and breast implants page contained similar content and the two terms are used extensively on both pages.

Whilst we have since worked to clarify the purpose of each page on the site for users, the canonical remains in place and has achieved exactly what it set out to do.

Takeaway – pages currently performing well in the SERPs for a keyword can give strong clues as to what Google wants to see.

Example #2 – The Bad

Fresh from our success implementing the canonical form this first example, we sought to find similar situations that a canonical may improve content overlap and ranking fluctuations on the same site. Our confidence turned out to be misplaced, the pride before the canoni-fall if you will.

Two pages were identified that we deemed to have overlapping content, discussed procedures that were very similar, often shared keywords and contained information that the team logically identified as very similar. After analysing current rankings, we added a canonical and promptly saw the site drop out for keywords specific for the page. Whilst previously Google agreed that the topics were the same, in this instance they did not.

It was worth taking the risk to potentially improve rankings for this term by canonicalising it to a very similar page that was performing well. However, steps were not taken to optimise it more effectively to the target key terms it was supposed to help to improve. These keywords were present in the page copy, but not page titles or <h1>, which definitely will not have helped matters. This represents an important warning to heed – be very careful of the pages you choose to canonicalize if the topics could potentially be distinct for user or search engine and make sure they are suitably optimised!

This canonical has recently been removed, and after a couple of weeks has returned for the phrases that it dropped out for. As with many things in SEO, issues do not exist in solitude, and the page in question still requires content quality improvements – but it certainly would be an interesting conclusion if we see it return to previously held positions in the SERPs on the basis of what Google considers to be searcher intent.

The dots in the image above represent when the page was ranking in the SERPs and show the canonical caused this site to drop out (apart from one rogue week in 80th place), with a return when the canonical was removed.

Takeaway – Just because you deem pages to be similar, doesn’t mean Google feels the same – and Google has all the data to back it up!

Example #3 – The Ugly

A final example in the weird and wonderful world of canonicals looks at an ecommerce site selling greenhouses. The site has two landing pages for a particular brand of greenhouse: one is the e-commerce product page; the other is a fairly comprehensive informational page on the brand in question. The informational page is the one that has historically ranked well for related terms – and continues to rank well despite the canonical being added (Google please don’t pull a Giphy on us – we’re not boasting!):

 Why would you ever want to add a canonical to a page ranking so well? It’s a legitimate question and one that comes down to perceived searcher intent. We noticed that despite this page getting a lot of traffic from high rankings; the page, which contains vast amounts of information about the brand, does not convert well. Of nearly 1000 sessions over the course of 9 months, the page resulted in only one conversion. We wanted to use the canonical to test whether ranking the e-commerce page would result in a higher conversion rate and better experience for the users, who we logically thought would have their intent answered more successfully when directed to the e-commerce landing page for the greenhouse brand.

However, over a month after implementing the canonical, we are still seeing the informational page ranking. Google is overriding the canonical we have put in place and continues to return the information page. Not canoni-cool Google.

“Searcher Intent Daniel!” I hear you scream. Is Google returning the informational page for certain queries because it believes this to be the most relevant for the searcher intent? Indeed, this is something that we have taken into consideration and it would make perfect sense if the queries were perhaps for the greenhouse brand alone. But what we are seeing instead, is the informational page ranking consistently for typically transactional terms such as “buy [greenhouse brand]”. For this category of searches, we would expect to see the ecommerce page returned, since it is far more closely related to the intent of purchasing this greenhouse brand. When I search “buy” I want to buy, by the by.

We took a canoni-crawl of the page to determine that the canonical was properly implemented – this confirms that Google has decided to ignore our hint!

Takeaway – Adding a canonical doesn’t mean that Google will listen!

 TIP – A good way to check if Google is following the canonical directive you have put in place is by using the info: search operator for the canonicalized URL and seeing what it returns.

Example #4 – A Tale of 2 Canonicals

In another instance of canonical fun, a client updated their canonical tag via the Google Tag Manager Javascript function, but when checking the source code of the page, they were seeing the old self-referencing canonical in place – not the updated canonical URL. However, when checking in the Screaming Frog SEO Spider with Javascript rendering enabled, or indeed using the Inspect Element function in Chrome Dev Tools, they saw that the canonical had indeed been updated correctly, which caused some confusion!

The reason behind this apparent discrepancy can be explained because Google caches the HTML source code of a page, rather than taking a snapshot of the HTML after Javascript has been executed. Therefore, what is shown in the source code is not the same as what is seen by Google once Javascript has been executed. Easy to be canoni-fooled!

Whilst this has been a recent change, what we are currently seeing in the SERPs for a major keyword is both pages ranking, when previously only one of the URLs was present in the results. It remains to be seen what happens later down the line when the change has had time to settle. Using the info: operator as described above, Google appears to have accepted the canonical, however, it will be interesting to see how the SERPs change from here on out.

Canonicals – No Steadfast Canoni-Rule

There are some situations in which canonicals will always make sense to use, in examples where content is copied across pages (but both serve a use), however, what is clear from our experience is that there is no cut and dry rule for when canonicals will work for more experimental purposes. Google can agree, disagree, or ignore your canonical suggestion completely. The element is just a hint, and at the end of the day Google has the final say!

It pays to always be wary when adding a canonical, but don’t be afraid to implement them entirely – as it can serve to help in situations in which search engines themselves are not sure what page to rank!