What a Cluster F!

Why clustering is …. Well, just a cluster F!_!CK.

Why Clustering is Important (It is the differentiator between really good and mediocre)

The rise of keyword clustering (5) as a technique is a response to the shift away from the one keyword per page targeting approach to the more topical and thematic targeting approach.

This in itself is a response to the changing nature of search and the advancement in search technology.

It is possible to rank for a keyword you don’t explicitly mention on the page. This is possible because search technology is advanced enough to understand how words are topically related (6).

NOTE: We don’t recommend not mentioning important keywords but the point is search is smart enough to understand how words and topics are related.

So, this is why we cluster words together and topically align them.

NLP Only Approach

Doesn’t understand how search engines see it
Fails to understand nuances
Doesn’t get the context
Doesn’t understand the complex buyer journey
Will cluster keywords together with very different intents

Natural Language Processing (NLP) (1) is a fantastic technology and has been one of the enabling technologies advancing search in recent times.

However, just blindly using NLP to group words or content together misses so much detail.

One of the main issues is that NLP doesn’t look at the SERP, so it doesn’t see things how search engines do.

You need to be able to understand the buyer journey and the subtle nuances of keyword variations that can mean very different things even if the phrases used for search are very similar.

Perhaps the biggest issue with the NLP only approach is that it does not understand intent at all, and that is very important for classifying keywords together (2) in a group. Having mismatched and disparate intents within a keyword cluster group will significantly diminish the effectiveness of the content produced or aligned to that meet the needs of that cluster.

SERP Similarity Approach

Doesn’t understand what the cluster is about
Doesn’t get context
Doesn’t understand the complex buyer journey

Search Engine Results Page similarity (SERP) (3) is another method for clustering and its one that I like a lot. If you consider there is no better arbitrator on what is relevant in Google than Google, then you can use SERP Similarity to your advantage in clustering (4).

Further than that though, you can use the SERP Features as well as the organic results to get a better picture of how related different types of content and SERPs are.

The challenge with this approach is that it doesn’t really know or understand anything about the results and the searches, just that they are similar results to other searches.

Furthermore, this is still essentially a static one-shot approach that doesn’t get context or the complexity of a buyer journey. The SERPs may be quite similar but the intent can be quite different.

Proximity & Priority (P&P)

Perhaps the biggest detail missed by the two standard approaches is which word takes priority.

Example;

Best foldable walking treadmill for weight loss.

Which cluster does this go in?

Best (comparative)
Foldable (feature)
Walking treadmill (product)
Weight loss

We must rank the priority and consider the proximity.

Imagine the misplacement or absence of a comma.

That can change the whole meaning of a sentence.

Take for example:

"Let's eat, grandma!" vs "Let's eat grandma!"

Oh what a difference a comma makes!

So how do we do this analytically?

We must assign a priority score to certain important terms. We must also consider the proximity of keywords.

So how do you do this?

Well …. That would be telling.

What if a keyword fits in more than one cluster?

Another challenge with most keyword clustering techniques is that a keyword search must fit in one cluster and only one cluster.

But what if a keyword search is relevant and applicable to more than one cluster.

We can’t just count that keyword twice in different clusters as that will affect the total MSV (Monthly Search Volume) calculations for individual clusters and will disproportionately report the opportunity.

So how do we do this?

Remember our P&P score?

We use this technique to assign the appropriate fraction of MSV to words in their respective clusters therefore giving a fair and proportionate MSV for each cluster.

The VibeLogic Difference - Our Approach to Clustering

The VibeLogoc difference is taking both advanced NLP & SERP Similarity layered on top of detailed niche-specific and business-relevant categorization conducted by humans.

Our secret sauce is combining deep human insights tailored to your business with advanced processing techniques, augmented with additional factors.

Our approach;

Understand how search engines see it
Understand nuances
Gets context
Understand the buyer journey
Understand the market

By using a proprietary algorithm we are able to combine these different clustering approaches and conduct industry-leading keyword clustering and intent classification.