There is a really toxic phenomenon for website traffic, which occurs on virtually all websites and that almost no one (except professional SEOs) knows about: keyword cannibalization.
Building an audience for a website takes a lot of work .
When, finally, after posting content after content, you start to see your Google ranking improve and your visits grow, it’s very rewarding. Finally, the first fruits arrive after so much work.
But this joy does not usually last too long, after a few years, almost all these websites usually reach a “traffic plateau” in which, despite publishing more and more content, the traffic stagnates . The web cannot grow any more, whatever the author does.
What’s more, the normal thing is that, after staying on that plateau for a while, web traffic even begins to drop little by little , despite continuing to publish new content, one after another.
How is this contradiction possible? For what is this?
Almost always the answer will be: a gradual increase in the degree of cannibalization of keywords in the content of your website.
So let’s take a look at what exactly keyword cannibalization is and how to fix it to get your traffic back up.
You might be in interested in What is Web Analytics and what is it for? guide from scratch
What is keyword cannibalization?
Keyword cannibalization occurs when different pages (URLs) of the same website compete for the same words in Google (or other search engines).
A typical example that produces cannibalization is the series of contents with the same title that do not differentiate search intentions .
- Tutorial de WordPress. Parte I
- WordPress Walkthrough. Part II
- Tutorial de WordPress. Parte III
It is easy to understand the problem that this generates if we see it through the eyes of Google: which of these pages should Google show to someone who searches for “WordPress tutorial” ?
We are making it difficult for Google to make the decision, even though the content speaks of clearly differentiated things, for example: the first part of the installation of WordPress in a hosting , the second of its configuration and the third of the content creation .
And it is that, apart from the fact that the title of a page is very decisive in the keywords to which Google will associate it, no matter how much the contents differ, in the end, this does not clearly resolve the doubt of which of them fits less with the search “WordPress tutorial” .
You might be interested in The Best SEO Agency & SEO Services in Karachi, Pakistan by Shaz
Consequently, Google’s reaction is usually to punish the three pages by not positioning any of them well for this search.
That is, we just caused a cannibalization of book keywords for “WordPress tutorial” and other searches with a similar search intent ( “WordPress course” , etc.).
The “plateau effect” and its disastrous consequences for websites
Now you might think: “Well, that’s it. I avoid doing series like this and solved “ .
Unfortunately it is not that simple: certainly, if you banish repetitive and undifferentiated structures in your search intent like the one in the example, you will have taken an important step , but not enough .
The problem is that if you don’t keep a close eye on cannibalization, over time it just happens naturally and inevitably , especially if your website is highly themed (which generally means less variety of searches). .
This is so because, even if you have published content with different themes, many of them will end up stepping on each other , at least partially for the same search intentions.
Let’s search, for example, “use web images” in Google within our website:
You will get these results:
As you can see, there are results with very similar content. The second and the third, for example, respond to practically the same search intention.
On a site like a blog, which periodically publishes new content around the same theme, you have to be very rigorous so that these things don’t end up happening.
In this case, “use web images” has been a single illustrative example, it is not a search for which we intend to position ourselves and it hardly has volume. Therefore, it is of no importance to us. But, technically, there is cannibalization in this case.
The crux of the matter is that what really matters to us here is that over the years this also occurs in many other contents and this is going to affect searches that do interest you.
This accumulation of many small (and large) cannibalizations is ultimately what ends up producing the “plateau effect” .
Why does it occur?
It happens simply because the new searches that you are gaining in Google with your new content are negatively compensated by the loss of positions in the current content that is increasingly “stepping on” each other, as in the example.
The effect can be so pronounced that you lose more search traffic than you gain, causing your website to go into “recession” , losing more and more traffic.
If you are not proactive, your web traffic will decrease yes or yes
In fact, this happened to us on this very blog after growing uninterrupted until 2016, in that year this traffic trend turned around. First it stagnated and shortly after it began to go down , and a lot, too.
The plateau effect is also seen very well in this graph of the web rankings for our target words:
As you can see above, around March 2016 our ranking was slowly, albeit very subtly, getting worse, despite constantly creating new content targeting new searches (which should improve our ranking).
What, however, was not subtle, was the drop in traffic as a result. We lost around 25-30% of our traffic.
We went around in circles with this for more than a year without knowing what was really going on until after the summer of 2017 we took the issue seriously and dedicated the time to analyze the problem in depth.
This quickly led us to see that the main problem was cannibalization , mainly caused by three SEO errors :
- Not being rigorous in the “compartmentalization” of search intentions . That is, instead of responding to each search intent (an affinity group of search intents) with a single content, we allow overlaps.
- Do not monitor the cannibalization of our content . This allowed him to silently climb higher and higher.
- Giving too much weight to the amount of published content and too little to the return that each of them should produce.
In this, SEMrush was essential to us , the SEO tool from which the previous graph comes, since it allowed us to detect the global positioning drop and go to analyze the fine thread at the level of specific keywords over time.
“In passing” and thanks to the SEMrush Keyword Magic Tool , we also realized that we were not really taking advantage of the potential of a good analysis of search intentions and keywords.
All this, ultimately, led us to make the following decisions:
- Prioritize the weight of SEO (and with it the visibility of the publications) against the number of publications.
- Consequently, we reduced our posting frequency (from once a week to once every two weeks ) to gain the additional time needed for SEO.
- Be very rigorous, both in the depth of the SEO analysis of each new content and in the “compartmentalization” between them (separate them as cleanly as possible from each other at the level of answered search intentions).
- Never publish content “arbitrarily” again (I mean, without analyzing it at the SEO level). This applies especially to the content of the invited authors, which was where this was most sinned.
- As a consequence of the above, in the case of guest authors, we first perform an SEO analysis of the content and pass them a content template with an “SEO On Page” content index (tentative title and subtitles based on keywords).
You can see the result above, when the graph rises sharply again from the end of 2017 with a spectacular improvement in our positioning and, consequently, in the visits received.
Right now, we generally have cannibalization pretty much under control, except for very minor searches (like the example above) that typically aren’t really ranking targets.
In the video that you can see above this paragraph you can see a real demo of this process that we are following now . It is an SEO On Page webinar that I did at the time with SiteGround on SEO On Page for blogs in which I break down a real example of our website.
How to know when cannibalizations really happen
Before seeing the specific techniques to locate the pages that are cannibalized, it is essential that you know how to fine-tune when a cannibalization actually occurs because if we analyze them “mechanically” we are going to identify channels that, in reality, are not.
Remember that we have to detect the following situation: for the same search there are several pages with different URLs that compete for the same search .
The point is that if we want to do things right, this is not as simple as it may seem at first glance.
Do they respond to the same search intent?
The first thing to be crystal clear about is the concept of search intent .
Google has become very intelligent and now it no longer works simply based on the words you enter in the search engine (which are the keywords ) but rather tries to guess what you really want to search for (your search intention ).
For example: would you be able to say for sure what a user who searches for “octopus a feira” on Google wants to find?
I would not be sure because it could be that I am looking for recipes, restaurants or fairs in Galician towns where octopus a feira is eaten. Namely…
The crux of the matter is that Google does know and that allows you to find out for yourself by simply doing the search in question:
As you can see, the search intent (of the majority , not 100% of people), according to Google, would be recipes .
If we search for “octopus a feira” and “octopus a feira recipe” we will get practically identical results. Therefore, despite being different keywords, they compete for the same search intent and there could potentially be cannibalization if we have more than one page ranking for them.
There are many more examples.
A very different example from the previous one would be the following: I would have thought that “web visits” and “blog visits” respond basically to the same search intention, giving the second variant a more specific weight to content about blogs.
It seems obvious, right? Well, strangely enough, that’s not the case at all.
If you search for “web visits” in Google, the results that come up all refer to tools to measure web traffic statistics . Until then, good.
However, with “blog hits” the content turns out to refer to how to get more hits on a blog , which is a completely different question .
Therefore, in this case, two pages each positioned in one of these searches would not be in a cannibalization situation because we are talking about very different search intentions.
We can no longer see Google searches in a merely mechanical way as before . Now Google also has an important semantic understanding (of the meaning of the content).
Or what is the same: do not think in terms of keywords, always think in terms of search intent which will translate into one or more equivalent searches (keywords) that will be the ones that will show us the tools with which we will work on this .
Which pages are competing for the same keywords in similar positions?
Building on the above, let’s now look at the basic idea of ??how to identify cannibalizations.
To do this, let’s remember, again, the definition of SEO keyword cannibalization from above:
But with what we have just seen about search intent , we have to qualify this “classic” definition of cannibalization (which is what they usually tell you on all sites) .
As many times a specific search intention will be translated into a set of several keywords (several specific searches) and SEO tools work with keywords, we will have to analyze this entire set of keywords that correspond to this search intention.
Going back to the recipe example:
We want to analyze a possible cannibalization of the search intent that we have called “octopus a feira recipes” . It turns out that for this search intent we found with the tools three searches (keyword) with a relevant search volume:
- “pulpo the fair”
- “octopus a feira recipe”
- “cook octopus a feira”
Well, what I am saying is as simple as the process that we will see next, you have to repeat it with each of these searches.
Of course, when it is a large set of keywords, you will see that analyzing two or three at most is usually enough, the result is usually identical for the rest .
How to find the contents that suffer cannibalization
Now let’s move on to the practical part: how to identify when and where cannibalization situations occur on your site .
Let’s first see the process to follow for this, which consists of the following steps:
- Choose a set of keywords (a specific search) to analyze as in the examples above. For example, “octopus a fair” .
- Locate which pages (URLs) are positioned for that search. If there is no more than one page, perfect, there is no cannibalization for this search and everything ends here, we repeat the process with the next search to analyze.
- If there is more than one page, you have to see if there is one of them that clearly dominates (much better positioned and more clicks) or if there are several that compete with each other. That is, it has a similar position and number of clicks (or no clicks at all).
From this third point, several scenarios can occur:
- If there is one page that clearly dominates and the others don’t respond to the same search intent, all in order, there is just normal residual traffic on other pages for that search.
- If there is no clear dominance of either and the search intentions are similar, there is cannibalization . Then touch what I comment in the section on how to correct a cannibalization situation
Free tools to locate cannibalizations
Now that you have the process clear, let’s look at the tools you can use to go through this process, starting with free tools that are available to everyone.
Locate cannibalizations with “site:domain”
The first option to detect potential cannibalizations is what I already did above in the first example; the “site:domain” operator in Google searches:
I recommend you start here for a quick survey because reviewing Google’s own results at a glance usually gives a relatively clear first idea of ??whether we are in a search with or without cannibalization potential.
Now, it must be clear that it is still a fast, very agile and useful survey method to take that first global look, but without the metrics you need for an accurate and definitive diagnosis of the situation .
Locate cannibalizations with the Google Search Console
To obtain this information, we need a more advanced tool.
And the best free option for this is, today, the Google Search Console . So much so that it even outperforms many paid SEO tools in the field of cannibalization.
If you do not have your site integrated with this tool, I advise you to do it right now , not only because of the cannibalization issue, but also because of the many other features that it incorporates and that are essential for any other website.
Here is a video that explains how to integrate it with a WordPress site:
In this tool we are going to use the “Performance in Search Results” section (Performance/Search Result menu):
Here you can filter the Google searches that lead to visits to your website by a bunch of criteria.
Look at the menu highlighted above in the image, here you have to do several things:
- Filter by the query you want to analyze . In this example, “create blog” . This is done by adding this new filter with “+ New” . I recommend you use the option “The query is exactly” that appears when entering the search to obtain maximum precision in the analysis.
- Add with a new page filter with the option “URLs that do not contain” page exclusion criteria . For example, I always add that the pages do not contain the hash mark (“#”) because it is in the URLs of the internal links of the page (from the table of contents) that here only annoys.
- Make sure you have the boxes under the filter criteria above activated (the activated ones are in color, the deactivated ones, in black and white). This adds the information we need to the results (clicks, impressions, average position).
- Choose the “Pages” tab (highlighted in the central part of the image). This makes the results list focus on the URLs of the pages that appear in the searches, which is what we are interested in.
Once this is done, you will see a result like the one in the image.
In this case, we are seeing two pages that have appeared in Google results.
In fact, the result is somewhat inconsistent because the second page, for a few impressions, has come out in the first position.
This is a small anomaly that may be due to various reasons that are not currently relevant. What matters is that, if you compare the figures, despite appearing two pages in the results, there is clearly no cannibalization here .
The first page (how to create a blog) clearly dominates and makes a lot of sense because the second, “ How to create a blog for free in WordPress” , although it is a more specific variant of “How to create a blog”, has a sufficiently differentiated search intention from the first.
Here, the example has been very simple, but you can find cases with even tens of pages. Also notice the highlighted part above this list of pages. Here you can download the list in an Excel and work with them from there, it can make things easier when there are many pages.
Let us now see a counterexample where cannibalization does occur.
I must add that this case is invented (these contents do not exist on the blog). Since this blog hardly suffers from cannibalization, I needed to create it “artificially” to have a really clear case:
And for this, I have gone back to the beginning, to the initial example of the WordPress tutorial series.
In the image you can see a very clear example of cannibalization : in a huge niche, with a lot of searches, the three posts hardly receive any traffic because all of them have positioned themselves very badly.
What has been said: a case of book cannibalization , with the typical dire effects that occur in these cases.
Let’s see, finally, one last example, this one, again, real and, moreover, quite peculiar.
If we repeat the cannibalization analysis process, but now for the search for “ migrate wordpress ” , we find this result on our site:
As you can see, here, to “migrate wordpress” , we have two contents that respond to search intentions very similar to the “create blog” example :
- A post answers a more generic way to the question of how to migrate WordPress.
- The other answers this question in a more specialized way by focusing on the specific case of migrating from one hosting to another (ie, it does not answer, for example, how to migrate WordPress.com to a hosted WordPress site).
Both posts are positioned in a similar way for the same search, therefore, we would have, in principle, a case of cannibalization .
However, in this case, both are positioned very high , middle positions between second and third.
And this is the strange part because in a cannibalization situation, it is normal that there are hardly any clicks on both (due to the poor positioning of both) or that there is one that dominates the others, leaving them “dry” in clicks. However, sometimes, atypical situations like this also occur.
In any case, do not trust either, these pages will surely end up going down over time if we do not take action on the matter before. I’ve already seen it many times.
The lesson is that not everything is black and white in cannibalization . Keep this in mind, you always have to analyze everything with an open mind and common sense.
Why is this happening in this case?
I can’t tell you for sure because the ways of Mr. Google are inscrutable, but I see sense that it is due, more or less, to these factors:
- Probably, to “migrate wordpress” , the two search intentions (general case of migrating vs. case of migrating with hosting) are quite spread out . This suggests that, among the first 10 results, 4 respond to the general and 6 to the most specific. So Google decides to satisfy both intentions. Perhaps one cannot conclusively prevail over the other.
- However, the predominant intention seems to be the most specific (of the first 5 results, 4 respond to the specific one). Therefore, in our case, the specific one, in principle, would have the upper hand. It turns out, however, that the generic one has been published for a long time and has a lot more authority (links) and that pushes it up.
Remember: this explanation is a simple hypothesis of mine. The same thing we have caught Google simply “red-handed” in a problem of its algorithm and nothing more ?
Locate cannibalizations with payment tools
Interestingly, the big SEO tools ( SEMrush , Ahrefs, etc.) do not currently have specific tools dedicated to cannibalization that automate this analysis process.
I have only found specific functionalities relevant to this in Sistrix and, curiously, in a very young Spanish tool that is DinoRank , the SEO tool that Dean Romero has launched and that expressly has a cannibalization detector .
How to fix keyword cannibalization
We already know how to detect cannibalizations on our website, we have got down to work and have detected a few. What to do now?
Differentiate search intentions more (long tail strategy)
When you have multiple pieces of content that are cannibalized, the general principle is going to be to reposition those pieces of content into different search intents. This, in some cases, will be very obvious to do and others much more complicated.
Sometimes, it will be as easy as digging deeper into SEO analysis and better differentiating search intent .
Let’s take, again, the example of the WordPress tutorial. If we analyze the three contents at the SEO level, we will discover that there are indeed high-volume specific searches for each topic covered.
Therefore, we could rename them like this
- “WordPress Tutorial. Part I” becomes “How to install WordPress on hosting” .
- “WordPress Tutorial. Part II” becomes “How to set up WordPress correctly” .
- “WordPress Tutorial. Part III” becomes “How to create and edit content in WordPress” .
And now you may tell me (with good reason), “but with this I have given up positioning myself for [wordpress tutorial], which is a search with a lot of volume” .
True, but luckily I also have a solution for this:
Why don’t we just create a fourth piece of content that is an introduction that leads through the other three? Such content does respond well to the search intent of “wordpress tutorial” and therefore allows the use of a title such as, for example, “WordPress tutorial from scratch” .
As you can see, this has been a very easy case to fix.
Also remember that, if you are going to change the URL as well (which would be the most normal), you must use 301 redirects, an essential concept in what we have in hand and that I explain here:
Here we could choose to make a redirection of the three old posts to the introduction post (WordPress Tutorial) or to each of the new ones. I don’t see 100% clear which would be the better option, but probably more the second one for consistency with the title and possible texts in the existing links.
Other times, the strategy will not be as clear as in the previous example and you will have to choose to consolidate content , basically you have these options:
- Merge them into one big one.
- Integrate them totally or partially in others.
- Just delete them.
It is very curious the results that you can achieve only with a good content consolidation in which, incidentally, you eliminate or recycle poor quality content, update content that is already very old, etc.
The most spectacular case I know of is this case where Pat Flynn of Smart Passive Income interviews the author of FinancialMentor.com who managed to triple his traffic by removing two-thirds of the content.
How to avoid SEO cannibalization in new content
As in everything, it is best to prevent the problem of cannibalization to prevent it from happening.
With what we have seen here, you should already know what to do exactly so that it does not even occur, although even so a certain minimum degree of cannibalization will be impossible to avoid, which is why it is convenient to check your site every X time.
Let us remember what are the essential guidelines to prevent the problem:
- Avoid posting arbitrary content . Try to always look for Google searches to target them to. Do it in depth, there are almost always many searches that fit and that are not obvious at first.
- Organize your SEO well , take an Excel sheet (or use another formula) to be clear about what searches each content responds to.
- Occasionally, it will not be possible to match your content with a search intent . Nothing happens, but be careful because these contents are more likely to be cannibalized or cannibalized than those that do have a search intent assigned.
- Periodically “upkeep” your cannibalization levels . Review your website periodically, at least, in the most important search intentions to detect possible cannibalization.
And a final trick : many times you will be overwhelmed by finding new searches with which to fit your content. Consider regularly re-editing and republishing old content as new posts .
You will see how, if 2-3 years have passed since the publication of a content, its republication will be something most natural to your audience.