For over a decade, the phrase “keywords are dead” has echoed across SEO. Yet, despite advancements in semantic search and Google’s ability to understand relationships between words, keywords remain essential to effective SEO.
Nicolas Garfinkel, the founder of Kixely & MindfulConversion, explores why.
Google has undeniably evolved. With innovations like Hummingbird (2013) and BERT (2018), its ability to process natural language and measure relationships between words has improved significantly. However, Garfinkel emphasizes:
“If Google is so good at understanding the relationship between words and phrases… why are the SERPs so different for similar words?”
Analyzing 500 keyword pairs, such as “best coffee makers” vs. “top coffee makers,” revealed that 71% of the time, the top results were completely different. Garfinkle suggests that Google’s semantic understanding may not fully align with our expectations.
However, Kimberly Brown, our Onsite Content Manager, had more to say about this revelation:
“I think those are two different searches, so I expect the SERPs to be different. To me, “top” means the most popular, which can vary year to year based on sales or reviews. “Best,” on the other hand, means “of all time,” based on various categories selected by the reviewer. So to me, they aren't serving the same searcher intent, and I would actually expect the results displayed to be different. I would not expect the SERPs to be the exact same, nor would I consider it a misstep on Google's behalf if they didn't exactly match."
Essentially, the searches are similar but are not expected to be the exact same.
Additionally, Danica Barnack, our Content Outreach Manager, stated these two queries could support two different intents, particularly in local search:
“‘Top coffee maker’ could mean a local coffee shop versus looking for a machine.”
Does this mean that Garfinkel’s research has no standing? Not necessarily. Even though his example may have some oversight, it is still a good reminder that SEOs must consider searcher intent as they evaluate search queries.
Over the years, the fear of relying too heavily on keywords led SEOs to focus on broader strategies like content clusters and pillar pages. Yet, Garfinkel highlights:
“We need to stop ignoring keyword research because keywords are dead… and start questioning everything we do.”
Instead of blindly producing content, SEOs should return to their roots: identifying and filling content gaps with intent-driven, keyword-informed pieces.
Garfinkel argues:
“Keyword research continues to become more and more important. It's doubled in volume in the last five years alone, depending on how well you trust the data from Google Trends. We think keywords are dead, but we keep wanting to do keyword research.”
While we believe in the power of keyword research, we also believe that content should be helpful to the user. While a topic may not have keyword value, it could help guide the user through the buyer’s journey.
The content could also reach a broader audience and help bring traffic to your site. It could also support thought leadership or social media initiatives. It all depends on the goal of the content.
Many SEOs lean on keyword difficulty scores to prioritize content. However, the data doesn’t always back this up:
“We found there was no relationship between how difficult the keyword was… and the position of where this content was ranking.”
This disconnect reinforces the need for smarter, data-driven approaches to evaluating opportunities.
At Page One Power, we’ve always taken third-party metrics with a grain of salt. Garfinkel’s insights into the keyword difficulty metric underscore the importance of understanding how third-party metrics work and why they should be used as a guide instead of a “North star.”
Unless the tool has access to data directly from Google, the metrics will be an estimation at best.
In a crowded content landscape where 97% of content gets no Google traffic (according to Garfinkel), SEOs must combine intuition with data science.
“We need to move from educated guesses to informed predictions… and spend time on things that actually matter.”
Garfinkel calls for leveraging tools like Python and machine learning to predict what will rank before content creation.
“We really need to start looking at things like machine learning to help us find patterns for our content opportunities that we can't just do as using our own human judgment.”
Garfinkel suggests creating a data pipeline and training it against Google rankings to start looking at predictions of what may rank in the SERPs. He recommends using data like backlinks, topical authority, and content relevance to train your model.
However, like third-party metrics, there’s no guarantee that there will ever be a platform or tool that can accurately predict what will rank in Google. You may get close, but unless you know all 200+ ranking factors, this type of tool will always be a “best guess.”
So, SEOs still need their intuition to determine the best content to fulfill a search query. While machines can do the heavy lifting, humans understand humans best.
Garfinkel closes his presentation by saying SEO isn’t about abandoning keywords but evolving our approach. By blending human judgment with advanced tools, SEOs can create content that fills meaningful gaps in the SERPs.
“Don’t waste your time creating content for the sake of content. Create content that you know fills meaningful gaps in Google and that's the way you're going to drive traffic.”