What is cannibalisation and semantic flux?

We have a pretty good idea that duplicate theming can create keyword cannibalisation within a site (see chart below) and should be dealt with, but we tend to hope that Google will know the difference.

But what about similar theming across sub-domains; across international sites, or sites that have entirely different domains owned by the same group?

This impact, which I call ‘semantic flux’ is down to more than duplication, this is domain conflict.

Why is this an issue?

This conflict could be happening to your site. The challange is, whether your site structure and the content offering are compounding the issue.

Businesses need to manage multiple assets, but if these are conflicting in Google it poses a great challenge for multiple departments from content developers, to webmasters to digital management.

Here I will show examples of this flux within sites that are semantically related, not just through a domain but also through a higher layer of association.

The elements behind for semantic flux and cannibalisation

  1. Duplication of content or theming – this could even be very general theming.
  2. No clear canonical page.
  3. A perceived relationship (on domain/cross domain/sub-domain etc).
  4. Google identifies the relationship and penalises one.
  5. It is visible in the SERPS.

Four types of semantic flux and cannibalisation

Below are examples of the different types of cannibalisation and semantic flux using Pi Datametrics.

The different coloured lines represent the performance of separate URLs in Google UK for particular search terms.

1. Internal conflict: keyword cannibalisation

When two pages within a site vie for the same search term it is generally referred to as keyword cannibalisation which is an internal conflict.

This conflict can be remedied with appropriate redirects, re-theming, landing pages and a better internal linking structure.

Cannibalisation often occurs when a product page is competing with a landing page and Google cannot decipher between the two. This result could be due to lack of content, the strength of the links or overall theming duplication.

Below is an example of RyanAir where Google can not decide which page within the site should be returned for the search term ‘flights to Hungary’.

As you can see the light blue line was doing performing in position two for a while, then suddenly dropped. After this drop Google then throws up three other URLs in the original page’s place, which never regain page one positions.

Chart 1: Internal keyword cannibalisation. Site: Ryanair.com. Search term: Flights to Hungary

keyword cannibalisation

2. Sub-domain conflict

Caused by duplicate theming (even at a very high level) across sub-domains, an example of this is a news page competing with a landing page elsewhere on the site.

Google nowadays generally only features one sub-domain in the SERPs even if both domains offer different content for a general term.

This will mean that in many instances sub-domains are competing with each other this could be difficult to sort out if different sub-domains have separate P&Ls and business models eg the gambling sector.

In the example below Asos (yellow) is competing directly with its own marketplace site (pink). Here we see that every few days Google swaps the sub-domain it is returning for the search term.

At no point does either site make it onto page one of the SERPs for this term.

Chart 2.1: Sub-domain cannibalisation. Site: Asos.com. Search term: Wool cardigan

This sub-domain cannibalisation can happen across an unlimited number of sub-domains within one domain. Remember www. is also regarded as a subdomain.

The more general the term, the more likely there will be conflict across the sub-domains.

Below is an example of three sub-domains vying for the same search term. In this instance the search term ‘grand national betting’ is extremely competitive and newsworthy.

Coral below seems to be pushing a lot of content out in the run-up to the race, but to no avail as each sub-domain brings the other down. Notice how when a new one appears, it seems to bring down the performance of the previous one.

Chart 2.2: 3-way Sub-domain cannibalisation. Site: Coral.co.uk Search term: Grand National Betting

When creating sub-domains with similar theming a real understanding of the semantic relationships need to be taken into account.

However, within some industries the managers of separate subdomains will have separate P&Ls and may rarely even collaborate on their content strategies.

So if a gambling site has a ‘poker’ subdomain and a ‘casino’ sub-domain which one should appear for the search term ‘texas hold’em’?

3. International site conflict

Conflict occurs most frequently between same language sites such as US and UK this is usually down to duplicate theming or duplicate content.

Any duplicate theming will have a negative impact in the search engines for both sites. Telling Google through relevant rel=alt tags and webmaster tools can remedy this.

Below we see conflict between a search made in the UK, but the USA site is conflicting with the UK site. Both pages on the hotels website are extremely similar, therefore Google can not separate them and is probably penalising one for duplicate content.

This page set up and content strategy therefore isn’t good for the user who could find themselves on an American site when they are after a UK one, or the search bot.

Chart 3.1: International site conflict. Site: hotelscombined.com and hotelscombined.co.uk Search term: Hotels in Texas

4. Family site conflict

This is perhaps the version of conflct that impacts SEO strategy most. Here we see sites within a family of sites, which offer similar themed content (or in some cases duplicate content) that leads to impact.

Sometimes sites that offer entirely different services, but may allude to a subject matter could see a semantic flux. (We saw a major brand that has a finance offering impact with their trans-Atlantic airline offering).

This is more difficult to remedy and will impact entire online strategies.

Chart 4.1: Family of sites conflict. Site: rbs.co.uk and natwest.com Search term: Secured loan

With this family impact for some search terms we see one site doing well, and the other simply going up and down, unable to penetrate the other’s performance.

Above we see a direct synchronisation of performance between the two sites, this is clearly due to the similarities in content.

The image below shows RBs’s and Natwest’s ‘Offset Mortgages’ page.

Duplicate content and site structure can harm your performance

Family site conflict

But not all semantic flux is down to duplicate content. Here for Currys and PCWorld we see flux for the term ‘gaming computer’.

There is obviously a relationship between the two businesses but the pages on offer here are different enough not to be affected by duplicate content.

In fact many sites have similar content to these ecommerce type pages across the SERPs but it is due to the relationship that these brands have that they are being penalised. 

Chart 4.2: Family of sites conflict. Site: currys.co.uk and pcworld.co.uk Search term: Gaming computer

Below are the pages returning for ‘gaming computer’, obviously similar (snippet) content but not identical as with the example above.

Again in the gaming sector we see Currys and PC World impact eachother’s search positions in the below chart despite the content being different enough.

Chart 4.3: Family of sites conflict. Site: currys.co.uk and pcworld.co.uk Search term: Gaming computer

The impact on content strategy and structure

When setting in place a URL strategy the semantic relationships that sites have across sub-domains, family sites or even stable-mate sites need to be considered.

These examples of cannibalisation show that simply creating a new sub-domain and hoping for multiple positions in the SERPs is simply not a solution.

Furthermore, if your holding company purchases a new brand within your sector, having the different sites but with similar offerings will not mean double the search traffic.

Brands in travel, gambling, fashion, groceries and more need to seriously look at their duplicate theming across their domains to eradicate cross-domain cannibalisation and semantic flux.