The Continued Demotion Of Keywords In Search

As we launch headlong into the new year, I’ve been thinking a lot about where search in general, and paid search specifically is headed. People have been saying for a couple of years now that keywords are becoming less and less important as a focus for search strategy. I have always been reflexively irked by this, probably because I came up through search at the peak of keywords’ power and influence.

But, much as I might not want to admit it, I think the evidence has been mounting for a while now and it has been starting to compile faster of late.

Exhibit A – The End Of Exact Match PPC Keywords

In late 2014, AdWords removed the ability to opt out of “close variants” of keywords, thereby effectively eliminating exact match keywords. Much was debated and lamented at the time about this move, but one thing is for sure – it has meant a lot more work to be sure that your campaigns are only firing for certain term variations if you don’t want all of possible variations and permutations. Looking back, this certainly seems like the first little push down a path where our reliance on distinct keywords was being tested.

Exhibit B – The End Of Detailed Keyword Data In The AdWords Keyword Planner Tool

In the fall of 2016 AdWords stopped making detailed keyword volume data available in its Keyword Planner tool. At first it seemed like a glitch (that was what Google said when it was first spotted) but eventually, it became clear that Google simply was not going to provide search volume figures for specific keywords any more. Keyword data is now shown in ranges and related terms are all lumped together, rather than showing traffic volume for each variation. This shows two things, in my opinion, (1) Google is trying to make advertisers think less about keywords and more about concepts and (2) we were not really ready for this data to just go away overnight.

Exhibit C – Growing Percentage Of Voice Searches Across Devices

By mid-2016 Google was reporting that 20% of all searches it completed were voice queries. The number and type of voice queries vary by demographics and geography, but even it we just take this percentage on its face, that is a significant portion of search traffic. Voice search emphasizes longer, conversational queries versus the succinct queries we have all been using and targeting. More keyword fuzziness (as I like to call it) fits in directly with voice queries, which tend to be made in the form of questions. Consider the difference between these two queries:

  • Traditional/typed: sushi philadelphia
  • Voice: what is the best rated sushi restaurant in philadelphia?

People tend to include more context in their voice searches than their typed searches. It’s probably at least in part because they feel like they are talking to “someone” a la Siri or Cortana rather than just asking a search engine to return a result. Either way, the value of interpretation of longer voice queries and matching them to variations of the core terms is expanding rapidly now and will continue to evolve as technology improves. Which brings me to…

Exhibit D – Machine Learning Applied To Search Queries

Dr. Pete had a great piece on Moz recently about how machine learning is impacting search. It is a long article, but definitely worth the read! In the piece, he talks about how machine learning is helping search engines to get smarter about associating words, phrases and concepts together. In other words, as machine learning matures and is applied in even more instances, the results that are returned to searchers will continue to get more sophisticated in their makeup.

In the early days of SEO, if you wanted to rank for variations on keyword terms, you really needed to make sure that you included the different variations within your content. Otherwise, you might get indexed for one phrase and not a closely related other. This lead to the inevitable abuse of keyword terms and I’m sure frustrated many a Google developer. Google has always not wanted web sites to try to optimize themselves for better positioning within Google. Now, with machine learning, they should start getting more of their wish, as old practices just don’t work and can even hurt now that engines are becoming more capable of equating queries across true variations and related concepts.

Exhibit E – Google Going “Mobile First” In Its Indexing

Mobile sites generally need to be more succinct than their desktop equivalents, both for speed of page loading and because people behave and want different things on a mobile versus desktop device. As more and more web time is spent on mobile devices, it only makes sense that Google would move in this direction as well. Mobile sites can’t afford to have too many words, so this also fits nicely into the new paradigm of quality, useful content rather than trying to cover all the permutations on a term or topic.

Search Engine Land has a great write up on this if you want more details.

So Now What?

In search, as in any industry, those who want to continue to thrive must be willing to adapt to the environment that is, but also the environment that is developing. Stubbornly clinging to things as you’ve always done them will ultimately not end well for providers or our clients. Social media, both usage and as an advertising platform, further demotes keywords from their once dominant position. It’s time to start thinking about audiences and messaging first and keywords second, as a means to capture traffic where they are still enjoying a relevant role. One thing is for sure – these changes are not going to slow down or stop!

What do you think? As always, love to hear your take on this topic. Sound off in the comments or hit me up on Twitter (@NeptuneMoon).


  1. […] The Continued Demotion Of Keywords In Search: Are keywords less important in Search now than they were in the past? […]

Speak Your Mind