Beyond Traditional Keyword Research: How to Adapt to Google’s Ever-Changing Algorithm
Keyword research lies at the core of organic search engine optimization strategy: You need to know which words your target customer types into the search box to be able to rank high enough to get discovered.
Yet, what many digital marketers fail to realize these days: Google’s algorithm in general and its understanding of search queries and text content in particular is changing and maturing and so should our keyword research strategy.
Google and Keywords: What Has Changed?
Back in the days when I was starting in SEO, it was pretty straightforward. All you needed to do was to check one of a few available SEO tools (Google Analytics and SEMrush were the two top choices) and extend your core keyword in order to find longer, and thus less competitive phrases to focus on.
We did have our own challenges: Mainly, we had to find the way to optimize for every little variation of each search queries (including singular and plural forms of the same keyword). In many cases, that meant creating a separate landing page for each and every slightly different query.
Apart from the obvious extra work this brought about, there was a bigger-scale problem with this approach: The amount of less-than-satisfactory content ranking on top of Google.
Google realized the problem and came out with a few updates that aimed at fixing the quality of search results. Those included
1. Google Panda
Google Panda Updates started rolling out at the beginning of 2011. The purpose of those updates was to reward high-quality content and get rid of “thin” content pages. While it sounds complicated, at that point Google was using pretty straightforward signals to distinguish higher-quality content.
For example, large amount of content while few to no backlinks pointing to deep pages was a signal of lower quality. Or the amount of non-original (duplicate, dynamically generated, etc.) content throughout the site was also a signal of low quality.
With Panda, most website owners - that had been enjoying search visibility by having a separate landing page for each tiny variation of each searchable keyword - got affected. At that point, we started “consolidating”, i.e. merging multiple landing pages into one.
Google Hummingbird (announced in 2013) was a big step forward from Panda because instead of using a set of signals (which were still quite easily manipulated), Hummingbird focused on a deeper understanding of the search query and the user’s intent behind it. The easiest way to better understand what Hummingbird is about is to explain it like this: Post-Hummingbird Google was serving results for “things instead of strings”.
If before Hummingbird you had been searching for [hiking area], for example, you’d have seen this exact phrase in the title of each top result in Google. With Hummingbird, you’ll see all kinds of related phrases that perfectly match your search intent, i.e. finding interesting places nearby to go hiking.
Hummingbird relies on semantic search, i.e. taking into account the broader context of each search query.
Rankbrain (announced in 2015) is Google’s algorithm component which is built using machine learning.
Machine learning is basically teaching the machine to analyze and understand data to come up with patterns and solutions a human brain wouldn’t be able to grasp. With machine learning integrated into Google’s core algorithm, Google has become even better at understanding a user’s true search intent.
Rankbrain is believed Google attempt to get ready for voice search, including better natural (spoken) language processing and answering the searcher’s questions immediately.
With Rankbrain in place, we started to see more and more of those featured snippets, i.e. selected search results that appear above all organic search results. Featured snippets are designed to give quick answers to users’ queries (especially to mobile and voice users which are always in a hurry).
How to Research and Optimize for Keywords Now?
Google has worked hard to make keyword matching tactic obsolete. Has the SEO community been keeping up?
While there’s still a lot of confusion, keyword research has been maturing too. One of the main reasons I love being an SEO is because it’s pretty excited to watch. We have been able to keep up with Google’s algorithm pretty darn well.
With that in mind, the key to optimizing your content these days is finding new tools that evolved in response to Google’s changes.
Here are three keyword research tactics to embrace now:
1. Apply Semantic Search Analysis with TextOptimizer
As I have mentioned above, Post-Hummingbird Google relies on semantic analysis in identifying search intent and taking into account the search context. This is exactly the approach we need to adopt as well.
Luckily there are tools to help and the one I’ve been actively using recently is Text Optimizer. The tool works as follows:
- It runs your target query in Google and extracts the search snippets
- It implements semantic analysis to identify related concepts and terms you need to include into your content to create a better-satisfying copy
The idea is genius: Google generates search snippets based on what it thinks satisfies the search query best. This way, search snippets represent the best possible summary of each given search query:
Now, take those snippets and use semantic analysis to extract underlying concepts Google uses to understand the query context:
TextOptimizer also extract action terms to help you create more actionable content:
Use Keyword Clustering with Serpstat
Remember how I mentioned above that optimizing for every little variation of your keyword doesn’t make sense because Google now serves results for “things instead of (keyword) strings”? But how to technically do that if we have to deal with hundreds (if not thousands) of keywords in any niche?
Keyword clustering is the answer. Keyword clustering is basically grouping keywords by meaning. It’s not a new tactic. We were clustering keywords 10 years ago. It’s the methodology that has changed.
Several years ago we were grouping keywords by a common modifier. Going back to our hiking example, the traditional keyword clustering tactic would result in a group of keywords that would contain [hikes] and [easy] in one phrase, as [easy] is a common modifier in this case:
- [easy hikes near me]
- [easy hiking trails near me]
- [easy mountains to hike near me]
As you may have guessed (from reading above), this method of grouping doesn’t make sense any more because it still relies on exact matching.
Serpstat is the only tool in the industry that uses a different approach to clustering. It identified relevancy based on the number of overlapping URLs for the researched queries. The idea is simple: The more identical URLs rank on top of Google for two queries, the more related those queries are bound to be. This way you come up with keyword groups that have may have no common modifiers but are still closely related:
Optimize for Questions with Featured Snippet Tool
With voice search on the rise, we’ve seen Google focusing more and more on answering questions from within its SERPs. Google’s “People Also Ask” results create both challenges and opportunities to search marketers:
- On one hand, they steal some clicks from search results (as search users may click questions and see answers right away)
- On the other hand, they let us understand Google’s interpretation of the query like no other search feature.
While I cannot help with the first point, the latter one is actually gold. Try clicking through “People Also Ask” results and you’ll see Google add more and more questions. But those are not simply related questions. They are follow-up questions: They are dynamically generated on the fly, depending on the question you clicked. For example, if you click “What is the best state for hiking?”, Google will add the following questions:
- Which state has the most outdoor activities?
- What state has the best mountains?
But if you click, “What is considered a hike?” in the same box, the questions that will pop up will be absolutely different:
- How long is a hike?
- Is Hiking better than running?
- What’s the difference between hiking and walking?
There is also a tool that helps you aggregate “People Also Ask” results from multiple target queries of yours, called Featured Snippet Tool. Use the tool to create content that actually solves problems:
More ways to research questions include:
- Use Buzzsumo’s question analyzer
- Answer The Public
- Twitter comments
- Your own internal site search queries (This is the best way! If you are not collecting this data yet, start doing so by using of these plugins)
Google is getting better at satisfying its users’ needs in the best possible way. The key to search visibility today is to try and create content that solves problems, matches search intent and includes related terms and concepts. Hopefully the tools above will help!
Have you recently discovered a keyword research tool you felt excited about? Please share it in the comments!