5 years ago

Assessing Link Signals in 2012 – Has Exact Anchor Text Had Its Day?

Well it’s been a whirlwind start to 2012, Google Panda continues to roll out and reshape the web and only a week ago Google announced 40 changes to search quality signals.

The SEO world was particularly interested in the turning off of a link evaluation signal Google have been using for the past 7 years. As usual Google didn’t give anything away and since then SEO’s have been digging around their data trying to understand what the change was and how they need to change their link building efforts accordingly.

For purposes of clarity, here is what Google have said:

we often use characteristics of links to help us figure out the topic of a linked page. We have changed the way in which we evaluate links; in particular, we are turning off a method of link analysis that we used for several years. We often rearchitect or turn off parts of our scoring in order to keep our system maintainable, clean and understandable.

We have seen some significant movements across the verticals we work in and I have had discussions with others in the industry concerning:

> Link networks being devalued
> Anchor text being devalued
> General link devaluation
> Link page relevance

I haven’t seen anything concrete yet, and probably won’t given the fact there are so many signals to account for, however I naturally wanted to check some correlation data across 20 of the verticals I currently work within, including SEO, finance, tech and fashion. I checked the SERP’s across 20 core keywords, trying to establish what these changes might have been.

Before I go into the data, let me stress that these were fairly competitive terms with multiple brands attempting to rank, also just to make it clear this is correlation, not causation. πŸ™‚

First let me apologise for not putting this in originally, it was a trend I spotted and wanted published asap. See below for full details of the test, some of the specifics have been left out due to client confidentiality.

The test was based on Data pulled from SEOmoz & Open Site Explorer, it was a comparison of the top ten sites across 20 core terms we currently track. The following metrics were compared;

Brand Anchor Text

We wanted to understand what percentage of brand anchor text the top ten sites had in their link profile, both in terms of the exact brand phrase and brand + keyword. e.g. SEOmoz + inbound marketing πŸ™‚

Keyword Anchor Text

Again the idea was to find out how the top sites for these keywords were using exact and variations of the exact ranking term within their link profiles. e.g. what percentage of their link profile was made up of variations of the ranking keyword and what percentage was made up of the exact keyword.

In this instance I didn’t look at the amount of variety, but I think this would be a good metric to check in another post.

Domain Authority/Page Authority

Again very simple piece of correlation, what is the average domain/page authority of sites in the top 10 rankings.

Unique Domains, Average Authority of Linking Domains

We all know the number of unique domains is heavily correlated with high rankings, however I also wanted to see how well the average domain authority of the top 10 links correlated. This was based on the fact that were seeing lots of sites ranking with very few links and almost no anchor text.

Social Signals

I didn’t think it would be complete without looking into social signals. How are tweets, facebook likes and google + shares affecting rankings.

I can’t share the keywords used as it is business sensitive, however they are all competitive and in the finance, tech, SEO and dating industries.

Brand Anchor Text

The main change I have noticed across the industries I work within, is that sites seem to be rankings with very few links, and almost no anchor text. Seems strange, but some massively competitive terms seem to be dominated by sites that historically would never have stood a chance of ranking.

brand anchor text signals

The chart shows the percentage of anchor text that includes a reference to the ‘Brand’, I believe building links with your brand name in the anchor text is great for future proofing your link building efforts and although not a direct factor, in my opinion, it does seem like a lot of top ranking sites have a good percentage of links including the brand name. Plus, using brand + keyword in your anchor text gives you the opportunity to optimise for your chosen keyword without risking any filters.

Keyword Rich Anchor Text

Something that I have personally seen happening over the past 12 months, is the devaluation of exact match anchor text.

anchor text signals

Both exact anchor text and variations correlate with rankings, however notice the significantly low percentages of exact match anchor text, 80% of the websites in question have less than 20% exact match anchor text. I ran a similar test a couple of years ago, albeit across different keywords, however the average exact anchor text result in the top 10 was around 38%.

Taking these results into account and the changes I have seen in the SERP’s, I would say now is the time to worry less about exact match anchor text, instead focus on natural signals including lots of variation and Brand based anchor text.

Whilst I was on with this Data I also decided to check out a few more factors:

Domain/Page Authority

It may seem obvious that page authority and domain authority are closely correlated with rankings, however the results we’re still interesting:

authority signals

As you can see there is a massive jump in order to get into the top 3 of Google, and ranking in the top 3 for highly competitive terms is what it’s all about.

When you look at unique domains, and the average page authority of the top 10 linking domains, we see a similar trend.

domain signals

It goes without saying that to get good domain/page authority you need to have links from authoritative pages. If you’re in extremely competitive verticals you have to source links from the best sites in order to rank highly, they seem to be the difference, bombing your profile with exact match anchor text doesn’t seem to work anymore.

Social Signals

This little test wouldn’t be complete without some social signals, after all Google could be counting them towards rankings.

social signals

Even though there seems to be some pretty good correlation, the signals are far too weak to be able to establish them as a ranking factor, multiple top ranking sites were doing so without a single tweet or Facebook share/like.

So the key take away from this reinforces what I preached in 2010, build a more natural blend of anchor text incorporating brand, and worry less about bombing your link profile with lots of exact anchor text.