Home / Blog / / How We Increased Traffic 110% By Reducing Content Cannibalization

How We Increased Traffic 110% By Reducing Content Cannibalization

Many people use our clustering tool to create content. That is, they carry out their keyword research, upload their keywords into Keyword Insights, and then identify content opportunities by creating content based on keyword clusters (we’ve written a guide on how to do this here).

However, we’ve recently done the opposite and used Keyword Clusters to identify content to get rid of and the results have been incredible; an uplift of 110% in organic traffic almost immediately. When the cluster insight is combined with the ranking insight you’re able to find possible content cannibalization issues very quickly and, hopefully, make some really quick organic wins for your website.

What is Content Cannibalization in SEO?

Content cannibalization happens when you have multiple pages targeting the same keyword intent on the same website. There are multiple potential problems when this happens:

  • Search engines are unsure as to which page to rank. They’re likely to constantly swap one out for another meaning you never consistently rank well for one.
  • Link equity may be shared between two very similar pages where there would be a better net gain combining them and doubling your equity to one page.
  • Depending on the site and how the pages are being created, larger sites may needlessly be bloating their site and reducing crawl efficiency/crawl budget. Please note – this is only a problem if the site is REALLY big (like – millions of pages).It’s this last point we capitalized on and reduced a website’s URL count by around 15 million which increased our organic traffic by 110%. Yes, you read that right – 15 million pages that were cannibalzing themselves.

The Background

In this case, our client was a large property website in the USA. Imagine a competitor of Rightmove (UK) or Zillow (USA).

The types of queries they wanted to rank for include:

  • Homes for sale in California
  • Homes for sale in Las Vegas
  • Homes for sale in Nevada
  • 2 Bedroom homes for sale in Florida
  • Log cabins for sale in Boulder
  • Mansions for sale in Los Angles
  • Colonial homes for sale in Orange country

And so on.

There were a number of issues including a constant struggle to rank many of their town based queries beyond the second page, having hundreds of thousands of URLs “crawled but not indexed” in search console and often having the wrong page rank altogether.

We started trying to diagnose the issue by crawling the site… but two weeks later our crawler had crashed and had only completed 50% of the crawl (it tapped out around 20 million URLs).

The problem was, they created a “property type” for every single keyword in their keyword research meaning they had around 400+ property types including:

  • Homes for sale in …
  • Houses for sale in…
  • Properties for sale in…
  • Log cabins for sale in…
  • Wood cabins for sale in…
  • Lakeside houses for sale in…
  • Waterfront houses for sale in….
  • 9 bedroom houses for sale in…
  • Mansions for sale in…

And so on.

These keyword, property type pages would be created for every state, town, neighbourhood and postcode in the USA leading to MILLIONS of pages.

Hopefully, you can see a potential problem here. Are search results actually different for homes vs houses vs properties? Probably not – they seem too similar.

How about lakeside houses and waterfront houses? Maybe, these could be different.

9 bedroom houses vs mansions? Surely they’re different enough that they should have their own page? Not sure though as a 9 bedroom house is pretty big.

Our Hypothesis

Our hypothesis was that we were badly cannibalzing ourselves as well as leading to a hugely unnecessary amount of URLs for search engines to crawl.

We couldn’t imagine “properties for sale”, “houses for sale” and “homes for sale” were different enough to warrant each of their own categories, for example. There were other obvious ones like “colonial homes for sale” and “Spanish colonial homes for sale” but there were also many potentially ambiguous ones (like 9 bedroom houses vs mansions).

Our Problem

We had over 400 “property categories” to check. Checking each one in the Search engine result pages wasn’t an option; it’s not scalable and would have been so boring.

 

Our Solution

We downloaded a list of all the property types our client had and then appended “for sale in California” to them. We appended a state to the queries because we found just typing in properties on their own sometimes led to blog related content rather than property listing pages so wouldn’t have given us the information we needed (for example, “farmhouses” on it’s own leads to a number of articles around what they are rather than a category page listing them).

So our list looked something like this:

  • Homes for sale in California
  • Houses for sale in California
  • Properties for sale in California
  • 2 bedroom homes for sale in California
  • 3 bedroom homes for sale in California
  • 4 bedroom homes for sale in California
  • Castles for sale in California
  • Mansions for sale in California
  • 9 bedroom houses for sale in California
  • Log cabins for sale in California
  • Colonial homes for sale in California

And so on + another 500 odd (we added some of our own keyword research to the list to also make sure we weren’t missing out on any. 2 birds and one stone…)

Keywords merged for cannibilzation

We then uploaded these to Keyword Insights and selected the clustering and ranking insight. If you’re unfamiliar with how our clustering tool works, it analyses the search results for every query uploaded and “groups them together” based on whether they can (and usually should) be targeted on the same page. Basically, if “properties for sale in California”, “houses for sale in California” and “homes for sale in California” appeared in the same cluster, it means we should only have one page targeting them and not 3 like they currently did.

We also pulled through the rank and ranking URL. By doing this, we can quickly see if a cluster is being ranked for by multiple URLs which is another clear sign we’re cannibalizing ourselves.

What Did It Show?

As we suspected, we had too many categories and there was a lot of cannibalization between them. Some were obvious (homes, houses and property pages for example) but we also found many not so obvious ones that would not have been found quickly (such as cabins, chalets and log home property pages).

Below is a screenshot showing the first 30ish property type pages. You can see in the right column the property types Keyword Insights has suggested to roll into the property type in the left column:

clustered keywords for cannibalization

In total, we’ve narrowed their property type pages from 413 to 85.

Across all their different states, postcodes, towns and cities this has resulted in reduction of around 15 million URLs.

The Result

An incredible 110% rise in organic traffic:

content merging

It’s also much easier to crawl now as there are a lot less URLs.

Other Examples

This method of finding content cannibalization opportunities can be applied to other large sites which lend themselves to having loads of keyword driven categories. Think of the likes of Ebay, Amazon or Etsy where there may be many overlapping product categories that could be rolled together.

In fact, we mapped this for another similar retail e-commerce client and found many areas of optimisation. In the example below, you can see a range of different URLs all ranking for one single cluster:

dress merge

Remember, if keywords have been grouped together within a cluster it means that they could and, generally, should be targeted on the same page. The fact different URLs for the same cluster are showing up in the Keyword Insights report means they’re likely cannibalzing themselves and they may consider rolling some of these category pages together.

Here’s another example for the same site where you can see a number of different URLs popping up in a single cluster again:

wedding dress merge

And two more examples:

other dress merge

In fact, we worked out we could have reduced this large retail e-commerce site but at least a couple of thousand pages by merging categories that our clustering tool had shown mean the same thing to Google at a keyword level.

So there you have it, you can also use Keyword Clustering to get rid of pages as well as creating content. Try it yourself for free with 400 credits when you sign up (no credit card required).

Andy Chadwick

Andy Chadwick

Andy Chadwick is a digital marketing consultant, specializing in SEO. He has been in the industry since 2013 and worked with start-up companies (he grew his own start-up to a turnover of £2.5 million in 3 years) as well as international organizations. He’s also worked in-house as well as agency side. Andy runs a successful SEO consulting business in the UK as well as Snippet Digital SEO consultancy with Suganthan.

Start your trial today for only $1

Sign up today for a $1 trial and enjoy access to 6000 keyword clustering credits, 3 Keyword discovery searches, 1 Content Brief and Pro versions of SERP Similarity, SERP Explorer.

Table of contents

    Start your trial today for only $1

    Sign up today for a $1 trial and enjoy access to 6000 keyword clustering credits, 3 Keyword discovery searches, 1 Content Brief and Pro versions of SERP Similarity, SERP Explorer.

    Subscribe to our newsletter

    Subscribe to get our latest news, offers, insights, and any updates.