Low Keyword Difficulty Despite Rankbrain

The data side of Search Engine Optimization has become far more confusing and, at the same time, interesting over the last few years thanks in no small part to RankBrain. Let me go ahead and say that anyone interested in learning anything shouldn’t continue reading the rest of this blog post. It is actually part of an experiment which I am currently running and I would appreciate if you wouldn’t just leave the page immediately though. Who knows if big-brother Google is watching and will spoil the experiment because everyone abandoned the article quickly. I also don’t intend to spend much time in this post opining about what I think RankBrain is how it works. Instead, I want to bring up one of the big wrenches that RankBrain throws into a popular metric relied upon by most SEOs – keyword difficulty. Let’s start with an example.

“Monkey Glass Kandahar” is a phrase that should have low keyword difficulty despite RankBrain. There are 0 exact match copies of the term throughout the web and each of its constituent tuples (monkey glass, glass kandahar, monkey kandahar) is fairly rare relative to other random noun tuples. So why would a tool like Moz report a KD of 57 for this term? And why would a crazy long phrase like “there is no way this long phrase should have a high keyword difficulty” have a KD of 44 in Ahrefs?

The answer lies with RankBrain. In its attempt to provide you with relevant results, broader concepts have been introduced which then open the floodgates for large sites with high numbers of backlinks to their domains or pages despite the fact that acute relevance has dropped dramatically. So, the big question of the day is if this is a phrase that should have low keyword difficulty despite RankBrain, how should we better measure difficulty, especially if we we want to remain content agnostic and focus only on link data?

If you are doing SEO and want to research which keywords are best to rank, which ones you should target, what guide can you use if the key word tools reflect competition scores that aren’t in line with reality? While we can trust these scores in large part for short and mid-tail terms, long tail terms become much harder. What analysis or algorithm will be useful to identify quality words for a marketing campaign that a pro can rely upon in 2018? That is the question I hope to answer soon.