Search engine optimization, or SEO, has changed significantly in the past few years. Thanks to the power of machine learning and artificial intelligence, the only way to build a sustainable, long-term SEO strategy is to create content people want to share. To combat these new trends, we need to employ our own machine learning technology to find what works and perform modern SEO at scale.
In this post, we’ll look at step 1: inventory.
Inventory of Performing Content
The first step in our modern SEO process is to inventory our content and understand what we have. Using any standard SEO tool such as Moz, AHREFs, SpyFu, SEMRush, etc., we start by scanning our website to find our best-ranking content. What are our most valuable URLs? What are the URLs most in need of help?
Below, I’ve segmented all the URLs on my blog in four clusters using k-means clustering (a form of unsupervised machine learning) on two metrics from Moz:
This analysis shows all the URLs on my blog by MozRank and MozTrust, two metrics I find very helpful. MozRank is effectively a measure of our ability to generate inbound links and other valuable external SEO outcomes. MozTrust is effectively a measure of our ability to generate high-quality, high-authority content and other valuable internal SEO outcomes.
What we see are four categories:
- Pages with high rank and high trust in dark blue (doing great!)
- Pages with high rank and low trust in red (fix up the content)
- Pages with low rank and high trust in orange (pitch for more links)
- Pages with low rank and low trust in teal (consider scrapping or rewriting)
We’ve now got a solid start for what’s in inventory. Our next step is to determine what our different clusters of content have in common. We want to know if the top 25% or so of top-performing pages have similar or different content than the other clusters. By exporting each cluster as a separate file, we will be able to analyze, using topic modeling, what we write most about and how well these topics perform.
In addition to topic models, I also recommend doing simple bigram and trigram analysis to see what multi-word phrases crop up frequently in each cluster.
What do we do with these topic models? We use them to inform our overall content marketing efforts, from building or modifying keyword lists to knowing what we should write more or less about.
Is this level of analysis necessary for every website? No. If your website has only 5 pages, then this level of classification and analysis is something you can do by hand. However, for larger websites, machine-learning powered classification and clustering is mandatory if we want results sooner rather than later.
Once inventory is complete, we should have a good idea of what’s working and what’s not on our website. We should know our own internal keywords, content, and performance. Next, we want to answer:
- What else is out there?
- Where do we have gaps in our content?
- What topics are hot?
In the next post, we’ll look at enriching our content inventory.
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of Marketing Blue Belt!