In a recent post, I talked about why internal links and hub pages are a vital factor for SEO, as this helps avoid the problem of pages on the same site 'cannibalising' each other's search rankings. 

In this article, i'll look at how cannibalisation and semantic flux are affecting search rankings for separate, but related sites. 

The examples I use here are PC World/Currys and Tescobank/Tescocompare. Each brand is cannibalising its own rankings in different ways. 

Currys and PC World 

These two sites are owned by Dixons Carphone and there is an overlap between the products stocked by the two retailers. 

Offline, many of the stores seem to have blended together, but online they are still two separate websites, with exactly the same design.

Currys leans more towards kitchen appliances etc, and PC World towards computing, but both sell laptops, PCs, tablets and TVs. And they are cannibalising each other in Google for these product terms. 

This chart, provided by PI Datametrics, shows the two sites' Google rankings for the term 'gaming computer' between March and July this year. 

(click for a larger version)

In this example the two are obviously swapping places. Google can't decide which site should perform for this term and gives credit to one site for a few days, then flips to the other.

According to Sam Silverwood-Cope from PI:

We call this The Semantic Flux. Presumably one site is just looking at its own positions in isolation, and therefore can not see this relationship that Google has placed on it. 

Here's the same chart just showing Currys' stats. Its SEO team must be wondering what's going on. Indeed such fluctuations may set alarms bells ringing over possible penalties. 

You can see the same pattern on PC World. It ranks on page two or three, then vanishes from the SERPs for weeks at a time.

And here's the cause of the problem. The sites have virtually identical pages for gaming PCs. Here's the Currys page: 

And PC World's gaming PC page. The only difference seems to be that there's an ad for personal loans on this one. The products, descriptions, page layouts etc are all the same. 

So, we have two separate sites, owned by the same company but selling the same products on pages which are almost exactly the same. No wonder Google doesn't know which one should rank. 

The solution? 

In a nutshell, Dixons Carphone needs to favour one site over the other as the current situation isn't helping. 

As Jon Earnshaw, CTO at PI Datametrics explains: 

Decide which doorway you want to present in the SERPS for key products and services and optimise these pages uniquely and appropriately. If there are two sites involved the same applies.

Additionally, if flux is likely then provide clear migratory pathways between domains for searchers so if they have come in through, in this case, Dixons Carphone’s 'wrong door’ due to SERPS changes then they can easily migrate to a different part of that world.

These two sites need to think of the bigger picture, i.e. Dixons Carphone’s bottom line. Semantic flux can, as we have seen numerous times with other forms of cannibalisation, lead to both domains losing out to competitors. So, decide a SERPS winner by category or service, theme, optimise, connect and monitor both position and searcher behaviour.

So, the answer would seem to be either merging the two sites or to avoid duplication of products between the two retailers. Maybe Currys should leave computing to PC World. 

Tescobank and Tescocompare 

This isn't quite the same as the Currys/PC World example, as the sites seem to be sharing the same rankings. However, they're both down on page four and below, so perhaps they are bringing each other down. 

The chart shows ranking changes for the search term 'travel insurance over 70s'.

As Sam explains:

There are different types of Semantic Flux. Here we see almost exactly the same performance from both sites, where they have come together in the SERPs and dance together, they seem to be inseparable.  

This could be quite an interesting case study, as Tesco closed its Tescocompare site on August 29, leaving the field open for Tescobank. 

You might think that, without the confusion caused by its 'sister' site, Tescobank might rank higher, or at least more consistently.

Perhaps it's early days yet, as it is currently at the top of page six for the term. 

The updated chart shows that Tescocompare has now dropped out, and Tescobank is currently a little higher than its average ranking over the year.

Perhaps it might take time to benefit from the removal of the conflict between the sites. 

Still, Tesco may have found the solution to its ranking problem here, even if this wasn't the reason it closed the comparison site. 

Comparison sites are very much an SEO play and, though Tesco would have been able to promote it through other channels, the fact that it was failing to catch enough search traffic may have led to its closure. 

Whatever the reasons, the lesson here is that Google is able to spot a relationship between the brands, whether manually or not, and companies with multiple sites cannot expect to duplicate content, site structure etc and retain search rankings.

In the next post I'll look at the relationship between RBS and Natwest, two brands owned by RBS Group, and how their are damaging each other's SEO strategies. 

Graham Charlton

Published 17 September, 2014 by Graham Charlton

Graham Charlton is the former Editor-in-Chief at Econsultancy. Follow him on Twitter or connect via Linkedin or Google+

2565 more posts from this author

You might be interested in

Comments (3)


Andrew Cock-Starkey, Product Marketing Manager at Cambridge English Language Assessment

This is one issue I follow with interest. We have more than 20 language version sites. Whilst some of this may overcome many issues ("homepage" becomes "page d'accueil" in French!) there are others where, broadly, the content of the pages is the same... and repeated by us over 20 times.

We cannot cut off those other sites (as Tesco did in your example) but how we optimise them and flag up the correlation for Google (geo tagging; lang tags; canonical tags etc.) is a constant headache.

over 3 years ago


Gordon McNevin

I've seen this many times since the Panda 4.0 update. In my research I'm led to believe that the final call comes down to the dwell time the user has on the page before returning to the serps. I'm also a firm believer in the theory that Google has 'try outs' for certain keywords and uses dwell time to determine if they should be higher in the serps than the others.

What do you think about this? Do you think dwell time is a factor for ranking everyday queries? Or just that with high bounce rates?

over 3 years ago


nora mcgrady, Consultant at Freelance

If two sites retained the same products in parts, do you think a cross domain canonical tag and perhaps a link in the body content would be enough to eradicate the issue? Or must you remove a particular product or service from one of the sites for you to rank?

almost 3 years ago

Save or Cancel

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Digital Pulse newsletter. You will receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.