Anyone who’s heard Dave speaking over the last year or so (and believe me I get my fill every single day!) will have heard him mention the ‘link landscape.’ It’s an interesting concept that might be familiar to you under different names such as ‘backlink profile’ but I just thought I’d share with you how we here at Bronco go about one branch of our research.
Now don’t think this is one of our entirely out-there kind of ideas; after all, most SEOs have trailed their fingers through the slurry of other people’s links looking for a pattern. But perhaps a majority of people take as their starting point: “what are the competition doing?” We start from a slightly different point: “what signals are Google telling us are important.”
In effect, rather than starting out by looking at what particular links an individual site has, we look for patterns that show up across a vertical (or ‘market’ if you’re old fashioned like what I am).
Firstly, we draw down a big stump of data for all of the sites currently ranking in the top 10 for a given keyword. Immediately, we get a useful pointers as to what’s happening – if there are lots of links then you can be sure it’s a keenly fought market and you’ll have to spend big to compete.
But obviously the brute force of numbers only takes us so far in understanding what Google is looking for. Here’s some of the other metrics we look at:
Distribution of Anchor Text
For each dataset we download, we attribute what we think would class as ‘brand links’ – either URLs, brand names, company names etc. That means we can split out the anchor text in various ways (this example is just one way you might like to slice the data).
It’s also instructive to look at the number of referring domains against the absolute volume of links. Another thing we’re seeing is that sitewides are having little effect in some markets – and maybe even harmful in others.
Comparing these two graphs, for example, we can see that while 16% of the volume of links is anchor text, only 3% of those links are from unique domains – so we can see that there are lot of sitewide anchor text links supporting the top 10.
Our gut feeling at the moment – supported by recent behaviour in the SERPs – is that Google is twiddling the knobs downward when it comes to anchor text. Keyword anchor text is, logically, something that is rarely seen outside of links built for SEO. It stands to reason that as Google continues its long, grinding war of attrition on SEOs it would start to dampen the importance of anchor text.
There’s been a lot of talk over the last couple of years (not least on here) about the importance of geotargetting.
Most SEOs accept that the locality of links plays a role somewhere down the line – especially when looking at local flavours of Google, be it the .co.uk, the .de or whatever. With the data on tap, you can actually figure out whether its an important factor in your market or not.
It also raises side issues. Large volumes of links from particular countries like India can indicate that there is mass, offshore linkbuilding going on. Not necessarily so, but it can give you a hunch as to what kind of market you’re looking at.
Site vs. Market Comparisons
Of course, it’s when you look at individual sites against the market average that you start to really get a feel for things…
Looking at individual sites is very enlightening in a lot of cases. Often you can spot which sites are strong and which are weak just by looking over a heads-up graphic display like this one.
This is particularly useful when looking at your own client: you can see if you’re overcooking on anchor text, short on volume or not paying enough attention to the brand.
Types of Domain
We also slice the referring domains to see what kinds of sites are linking in. We’ve still got a long way to go to being able to automatically categorise domains, but it’s fairly simple to start building up a ‘whitelist’ of importance domains (newspaper sites, education/government, Wikipedia etc) and look for blog footprints to help whittle down the numbers.
In some markets, we see that news coverage is more important than others, for example. We try to break down links into categories:
- Whitelist “power sites”
- News coverage
- Government / education sites
- Affiliate sites (look for patterns in the structure of the referring links)
- Social media
- PR sites
- Article sites
- Forums and messageboards
In the interests of honesty, it’s worth noting that the biggest chunk of any such analysis is (for us) ‘other’ 🙂
Again, it’s worth knowing what kind of linking is going on across the market so you can plan a strategy that reflects what Google apparently think is ‘natural’.
Nofollow vs. Follow
This near-religious debate in SEO often misses the point. The real question is: “what does Google think a natural backlink profile looks like?” The follow/nofollow pattern is part of that mix and worth exploring.
A side effect of doing a big, market-wide exercise like this is that you can start to pull in the common links. While not being a big believer in the value of “just getting the links the other guys have” myself, it is instructive to see where 5 or 6 sites might be getting links from… 😉
Reading the Runes….
SEO is not a science and the truth can’t be found in numbers. It is all to easy to be beguiled into thinking “just one more link…” when you’re pushing for a certain slot. Things that don’t show up in the numbers – like brand, trust, regional variations, personalisation and more – can be equally if not more important.
So whenever you undertake big number-crunching exercises like these – keep some perspective about you. SEO isn’t just about coming up with a formula. It is about understanding the rhythms and patterns in the SERPs, applying a little prognostication and imagination and developing a strategy to outmatch the competition by hook or by crook.
Oh… A Note on Data
Obviously, we can’t download Google’s own index (how nice would that be?) You can choose your own dataset, but we currently use Majestic SEO‘s index.