Network analysis for SEO
Network analysis is defined, simply, as the analysis of the relationships between objects. It’s a field based on both graph theory (a branch of mathematics) and sociology. The final output of network analysis is a network graph, which visually maps the relationships that exist between objects. These relationships can then be expressed mathematically based on an object’s position on the graph and the number of relationships they have.
When applying network analysis to the field of SEO, the objects are websites and the relationships between them are links. By graphing these relationships, insights can be gained about how a website is linked to and exists on the web.
There’s a number of ways that network analysis can be applied to SEO:
– Visualizing internal links on a website
– Visualizing interlinking between various domains
– Visualizing backlinks to a domain or domains
– Competitive analysis of backlinks
As an example of how network analysis can be used to find backlinks, below is a network graph showing how six different websites are linked to. The dots represent websites and the lines represent links. Our client’s site is highlighted in blue, while five of their major competitors’ websites are highlighted in red. External sites that link to either our client or their competitors are represented by black dots.
As you can see in the graph, each of the six core websites has their own particular backlink profile – sites that link only to them. Some links are shared between the different sites, and these lie at the center of the graph.
What we’ve highlighted in this next network graph are websites that link to at least two of our client’s competitors, but who do not link to our client.
These websites are prime linking targets. As they link to at least two competitors, they demonstrate that they have an interest in that particular industry and would be likely have an interest in our client’s website.
Network analysis can be applied broadly to the field of SEO in analyzing how websites link to each other, and in conducting competitive analysis. The field also has obvious applications for social media in understanding how users are connected to each other in online networks. Any online social network – whether it be on Twitter, Facebook, Youtube, Flickr, etc. – can be easily modeled using a network graph. A graph could focus on how one user is linked to other users, which is called an ego network. Or a social network graph can focus on a particular topic – comments about a Youtube video, a Twitter hashtag, or a Facebook group – and look at how that group of users is interconnected. The search engines have stated that the social web has a small influence on rankings so there is also an SEO implication for this type of analysis.
As an example of social network analysis, here is a network graph of the Twitter network of 4FRNT skis, a ski manufacturer. This graph has been analyzed to 1.5 degrees, meaning that we’ve not just looked at who follows 4FRNT on Twitter, but how those people are connected to each other as well.
While 4FRNT has about 1,200 followers, those followers are highly interconnected with approximately 30,000 connections between them. When you have a dataset this complex it can be hard to make sense of what is going on. For 4FRNT, what we’ve done is break down their network into clusters – groups of highly connected Twitter users – and discovering who is central in those clusters.
Dots in orange are a group of users focused around ESPN Skis. Dots in green are users focused around Armada Skis, another ski manufacturer serving a similar niche market as 4FRNT. The dots in pink focus around a Twitter user called The_Canyons, which is a ski resort in Utah (based in the same location as 4FRNT). The dots in blue focus mainly around 4FRNT itself. Instead of just viewing the graph as a mass of dots and lines, we’ve been able to get some sense about who on the 4FRNT network of users is serving as sub-areas of attention.
If you’re looking to do your own network analysis, there are a number of software programs you can use. Several of the software programs require knowledge of programming languages such as Python or the statistical programming language R. An easy to use, free and open source plugin for Excel is NodeXL. If you are looking for something more advanced that does not require programming skills, UCINet is a nice option.
Network analysis is in use in areas including the sciences, social sciences, and business. Here are a few TED talks showing network analysis in action in other fields: