Natural Language & The Future of SEO
OnCrawl's Rebecca Berbel joins the SEO Rant Podcast to talk about the role of language processing on SEO both now and in the future:
The evolution of Google's ability to understand content naturally
How language processing will impact SEO going forward
The impact of updates like the Product Review Updates & the Helpful Content Updates on the Google ecosystem
The Role of Language Interpretation In Search
Google's ultimate goal is to understand language the way we as humans do... naturally. This is where Natural Language Processing meets Google Search and SEO. Leaving aside everything about SEO from ranking factors to crawlability, Google's most basic goal is to understand what a user is searching for and to understand web content so that it can give users what they are searching for.
Which is why, for example, keywords don't work the same way for SEO as they used to. The idea is for Google to move away from a linear "look up" and more towards understanding what people are actually searching for. A huge part of this depends on Google's ability to profile language to understand what the structure of the language on a page implies in terms of accuracy, relevancy, and quality.
This prevents SEOs, content creators, etc., from to abusing Google's algorithm. It also gives the search engine a more holistic understanding of what the corpus of a specific topic really is thereby allowing for a more layered set of results.
To that, the focus SEOs place on keywords has become overstated. What matters more than keywords is how the language, as a whole, is implemented on the page and across the site. To the latter, the domain comes into stronger focus both in terms of assessing quality and site identity (and therefore what is and what is not topically relevant to the site).
Part of this holistic topical "profiling" means that concepts take center stage, not keywords. Again, Google isn't looking to match the words in the query to the words on the page any longer. What they want to do is to take the latent concept embedded within the query and match that to content that best represents that concept. Meaning, showcasing that your site/page handles particular concepts well is of strong value. For a page or even a site to handle a concept well means that the concept needs to be contextualized. Contextualization is very often how we identify concepts and it's how machine learning properties are able to make "conceptual determinations."
For the record, as Rebecca points out, contextualization cannot exist without depth. That doesn't mean your content needs to be 500,000 words in order to rank. What it means is, in order to provide contextualization there have to be nuanced facets, and layers to the content. Otherwise, contextualization simply won't exist. If you can do that in 500 words - great. It's not about content length it's about results.
My advice? Think of content as a corpus. When you treat it as such contextualization is bound to occur.
For more of the SEO Rant Podcast check out our previous episode on why SEO does not need to always be complicated.