How semantic & traditional keyword research come together for better SEO

Author: Lazarina Stoy

Keyword research is the starting point of all organic marketing projects. It unveils how, why, and when your target audiences search for (and discover) businesses like yours, and what their needs are. 

For many years, this process focused on direct keyword targeting. Nowadays, keyword research aims to understand users’ content and information needs, and to pair that with brand content (in search results as well as on other platforms). 

As an industry, we're transitioning to a more holistic and integrated approach that marries three critical components: 

• A user-focused strategy for interpreting search context as well as explicit and implicit search intent

• An insightful analysis of queries, pinpointing the topic and other indications of how the user wants information served to them

• A strategic decision-making process on content creation, informed by pre-existing knowledge, topics, and entities

In this article, I’ll show you how semantic and traditional keyword research methods can come together to address these considerations so that you can create user-first content that also ranks well in today’s search landscape.

Table of contents:

• Traditional keyword research: Still useful, but outdated

• Semantic keyword research uses context cues to modernize traditional techniques

  • Traditional vs. semantic keyword research: What’s the difference?

• Semantic keyword research: Concepts and how to integrate them into your keyword process

  • Entities, entity attributes and entity attribute values

  • Search query sequences and query paths

  • Query augmentation

  • Query context and session context

  • User search behavior

  • Information gain

Traditional keyword research: Still useful, but outdated

With the introduction of semantic search, traditional keyword research needs to evolve.

Traditional keyword research generally begins with identifying seed keywords, which you run through SEO tools to discover related terms. From there, you apply different criteria to filter out irrelevant terms, laying the groundwork for a keyword universe. 

Of course, each of these steps can vary in complexity, influencing the quality of your keyword data. For example, to identify seed keywords, you can turn to sources like user surveys or first-party data, product or service research, competitor research, and so on, which can result in a more comprehensive data gathering process. Similarly, when sorting and filtering your keywords, you can incorporate additional data points, like keyword intent classification, SERP features, and topical and brand relevance to ultimately create a higher-quality keyword universe.

Even so, this approach is potentially heavily skewed by incomplete or flawed data and metrics, like search volume and traffic potential, keyword difficulty and competitiveness. 

Traditional keyword research is metrics-driven and backward-looking, when it now needs to center around the user, their search journey, and content needs. 

These practices worked for a long time, as many of Google’s ranking systems are based on text evaluation. In this context, keyword matching between on-page text elements (like the title or URL) and the search query was a pretty significant sign to search engines that the content would likely satisfy the user’s needs. 

You still need some of the basics of traditional keyword research, though, as they provide essential context for subsequent steps in your content strategy. For instance, breaking down your keyword universe by different query characteristics (e.g., search intent, brand mentions) enables you to identify patterns of search behavior to understand how they relate to metrics like competitiveness and search volume. This granular analysis allows for more precise content targeting and optimization, ultimately enhancing your user engagement and satisfaction.

Moreover, traditional keyword research excels at uncovering related keywords (semantically connected terms, synonyms, jargon, local expressions, or other language variations). This helps you better understand and align your website copy to the way your potential customers express themselves, which can also make your website more discoverable in the process. For example, if a region-specific term like ‘fizzy drink’ is more commonly used than of ‘soda’ in a particular market, incorporating this local expression can significantly improve search relevance for those audiences.

In short, while search engine algorithms have changed, the principles of traditional keyword research remain foundational. And, with the integration of semantic understanding in the content ranking process, SEO keyword research requirements have also changed. 

Semantic keyword research uses context cues to modernize traditional techniques

Incorporating semantics into SEO research has been a consistent theme in the work of researchers, like the late Bill Slawski, for more than a decade. These concepts are directly extracted from technology patents that Google and other search engines filed.

Semantic keyword research evolves our understanding of user search behavior and context—as well as how to leverage it. This approach goes beyond traditional keyword research by integrating concepts like entities, knowledge graphs, the context of searches, the intent behind search queries, and so on—rather than just relying on the keywords the user puts into the search bar.

Traditional vs. semantic keyword research: What’s the difference? 

Unlike traditional keyword research (which prioritizes metrics), semantic keyword research focuses on real-world user behavior, placing the searchers thoughts and actions at the center of your keyword research practices.

Semantic keyword research adds real-world context to the keyword process by emphasizing the user's search journey (i.e., the sequence of queries and platforms used) and the context surrounding searches. 

Traditional keyword research	Semantic keyword research
Keywords Keyword length  Keyword search volume  Keyword difficulty  Traffic potential  Competition	Entities, entity attributes, entity attribute values Query sequence Query path Query augmentation Query context Session context User search behavior Information gain Search intent  Knowledge graphs

Semantic keyword research: Concepts and how to integrate them into your keyword process

There are several concepts that inform how I approach semantic keyword research. They include:

• Entities, entity attributes, and entity attribute values

• Search query sequences and query paths

• Query augmentation

• Query context and session context

• User search behavior

• Information gain

Entities, entity attributes and entity attribute values

Keywords represent phrases or specific words that hold some value from an SEO perspective, while entities reflect things that exist in the real world. Sometimes these two things overlap in the context of keyword research. 

• Entities are distinct and well-defined concepts, such as people, places, things, or ideas. (E.g., Barack Obama, The Louvre, a smartphone, etc.)
  

• Entity attributes are characteristics or properties of entities. For example, for the entity ‘dog’, attributes include ‘breed’, ‘ fur color’, etc.
  

• Entity values represent the specific values of entity attributes, such as ‘Labrador’ for the attribute ‘breed’ for the entity ‘dog’.

Let’s contrast this understanding against how SEOs traditionally considered keywords. 

From the context of a user typing a search query into a search engine like Google, the query is composed of keywords of varying importance for SEO, some of which might also be entities. Let’s look at some examples.

Search query	Keywords and importance	Entities and type
[Shop online Nike Jordan Air force one]	Shop — Transactional intent Online — Navigate to website or store Nike — Branded search  Jordan air force one — Product search	Nike — Organization Air force one — Product (Michael) Jordan — Person
[Beginner guide to SEO research]	Beginner — Level of complexity need, introductory content Guide — Content type need, informational intent SEO — Main topic Research — Subtopic	Beginner - other SEO research — other Guide — other

Understanding entities is essential as it aligns your content with the way modern search engines interpret queries. Google uses entities in many of its ranking systems to:

• Interpret content and evaluate its relevance, prioritizing pages that comprehensively address the user/query’s context.
  

• Connect entities in its knowledge graph to deliver richer, more personalized, and contextual search results (as both are used to improve recommendations for search results and SERP features). 

When researching keywords, instead of just taking them at face value, incorporating entities can help you: 

• Understand different ways people refer to concepts.
  

• Identify entity attributes that people commonly search for and their most popular variants, indicating potential content and site structures. (E.g., when referring to ‘dog food’ people might often search for specific types like ‘kibble’ or ‘canned’, so for the entity ‘dog food’, an attribute would be ‘food type’ with variables ‘kibble’, ‘canned’, and so on.)
  

• Detect information gaps that you can address to create more comprehensive content. (E.g., If you identify that web content commonly discusses kibble and canned dog food, but not raw or semi-moist, researching and writing about these food types on the web can set you apart from competitors).
  

• Improve the contextual relevance of your content, enhancing user intent alignment.

It’s also worthwhile to consider entities, entity attributes, and values whenever your website can potentially help out users—even if such search terms are not widely popular. 

To start extracting entities from your keyword lists with Google’s own Natural Language API, use this free, no-code, beginner-friendly template in Google Sheets. 

Search query sequences and query paths

In semantic keyword research, you should also aim to understand the sequence in which users perform search queries. The aim here is to grasp (in the same way that search engines do) the relationships and sequences of queries performed in a single session, and use this information to enhance the content you produce. 

This involves analyzing how users refine their searches, the progression of their queries, the number of queries they conduct, and how their search behavior evolves during a session. 

Here are just a couple of ways to achieve this: 

• Run important keywords for your business through Google Autocomplete to see the searches that Google is recommending to users (related to their original search). 

The Pemavor Google Autocomplete tool for bulk keyword analysis and extraction.

• Scrape data from the Google SERP for your important keywords, specifically the modules (e.g., Related Searches, People Also Ask, People Also Search For, People Search Next), as these indicate queries that are semantically related in terms of topic, intent, and session context. Tools like Semrush can help you do this for individual searches, while DataforSEO can help you do it at scale.

Query augmentation

Augmented search queries are queries where the user incorporates additional information derived from entity references within a query. The concept of augmented search queries involves recognizing specific entities, such as people, places, or things (or a combination of them), mentioned in a user's search query. 

For instance, if someone searches for Blake Lively (Person) and Ryan Reynolds (Person), the search results would recognize the entity references in the query, and would suggest augmentations to the query to introduce new information to the user, like movies they starred in together, or whether they have kids.

The search refinement bubbles are based on the entities, in this case, Blake Lively and Ryan Reynolds.

Google can then generate additional queries based on these entities and their attributes, integrating the results with the original search query outcomes. These fused results ensure users receive a richer and more detailed set of information.

To incorporate augmented search queries into your keyword research process:

• Ensure that your content clearly identifies and references significant entities relevant to your topic. Understand how different entities can be logically combined with one another in searches. For instance, the entity Kamala Harris (Person) might be logically combined with Douglas Emhoff (her husband; Person), but it might also be combined with ‘2024 US elections’ (Event).
  

• Pay attention to search queries that reference entities that appear in your Google Search Console data. Analyze all queries that contain the same entity and ensure that there are different pages on your website that reference said entities, and that they are linked.
  

• Understand and incorporate thematic searches. For instance, around Coachella (music festival), people might search for festival wear, sparkly tops, boho shorts, or simply search for [Coachella outfit]. With that same logic, you can also improve your product category structure, adding pages based on interests, not just product characteristics.

Query context and session context

Contextual search leverages natural language processing to optimize search results based on the context the user provides in their query or in their overall search session. 

For example, if a user asks Google something along the lines of [What’s the capital of France?] followed by [What are some popular tourist attractions there?], the search engine uses the context from the first query (query context) and the sequence of interactions (session context) to answer the second query accurately. 

In this example, even though the second query does not contain the query context (i.e., in France), Google suggests query refinement and incorporates the query context in the search results to better address the searcher’s intent.

Query 1: [capital of france]
Query 2: [popular attractions] Note the suggestion at the top: “This search may be relevant to recent activity: popular attractions in france”

To incorporate query and session context into your keyword research process: 

• Use Search Console to understand the co-occurring queries related to your website’s key entities. Especially if you have similar types of pages, like tutorials or guides, researching the query patterns on how these are found by users can help you adjust the on-page content to better reflect user search patterns. For instance, a tutorial might rank for a query containing the term [how to fix ‘problem X’], but it might also rank for a query containing the term [how to ‘desired outcome’]. 

Session query 1	Session query 2

• Analyze the SERP and the top-ranking pages to find co-occurring entities and keywords that might help those pages rank for more terms.

• Understand subtopics of the content that ranks well in your niche.

This can help you incorporate context in your content (in places like headings, anchor text, and on-page text) and understand the context you need to add when discussing certain entities.

User search behavior

User behavior data like click-through rate (CTR), dwell time, and bounce rate can impact search rankings. High CTR and long dwell time can signal quality content, while low bounce rates and reduced pogo-sticking can lead to improved rankings over time.

Google may also use SERP interactions and user feedback to re-rank pages. Positive interactions enhance credibility—this can be anything like the actions a user takes on the SERP or the queries typed in after the page visit.

To incorporate this data into keyword research: 

• Merge data in Looker Studio, combining ranking queries from Google Search Console, entity data from those queries, and GA4 metrics like bounce rate, exit rates, and other user engagement metrics to monitor for patterns or any glaring issues. For instance, you might notice that all pages that rank for queries containing a certain entity are failing to engage users.
  

• Utilize tools like Microsoft Clarity to identify which page elements or content sections cause negative user interactions. Then, compare these findings with the elements and sections of better-performing competitor content to gain insights and make improvements.

Information gain

Information gain refers to how much additional useful information a new document (web page) provides to  users who may have already seen other pages on the same topic. As you may have already gathered, Google has a patent on information gain scoring and ranking pages based on this concept.

Put simply, pages with higher information gain scores may get ranked higher because they provide more unique and valuable information. This helps prevent users from seeing repetitive information across multiple documents, thereby enhancing Google’s search experience.

You need a thorough grasp of entities, their attributes, and attribute values, along with a keen understanding of the current information available in the search landscape in order to effectively apply this concept in your keyword research. 

Focus on presenting new or proprietary information—not just replicating the articles that currently rank. Incorporate user research and first-party data to validate new keyword variations without compromising quality and search experience. 

This is where semantic keyword research departs from traditional methods: Instead of only focusing on metrics from third-party tools (like search volume), understand the information gaps and fill them through your content. For instance, before Zapier created their templated ‘integrations’ pages, the keyword volume for connecting niche tools with one another was likely negligible. Yet, as the product addressed this need, the demand grew, validating the importance of focusing on user intent and behavior, rather than just keyword metrics. 

This approach shows that understanding the entire search journey, including the sequence of queries and the entities mentioned, leads to better content strategies and improved user satisfaction.

In keyword research, the semantics matter

Traditional keyword research still carries value, but there are more advanced concepts that search engines use to understand users that you should incorporate into workflows to ensure your pages are competitive. 

• Instead of relying solely on keyword suggestions from third-party tools, incorporate different data points from user research, SERP analysis, entity analysis, and topic research. 
  

• Understand the topic that you’re planning to cover and the subtopics that make it up.
  

• Research the entities your content is about, their attributes, and the values that are commonly mentioned alongside them to advance the existing information landscape.

In terms of practical steps and tools to incorporate in your process, I’ve prepared a handy checklist that features the steps needed to conduct semantic keyword research, the tools you can use to get started, and additional resources to help you grasp the concepts mentioned.

---

Lazarina Stoy - SEO & Data Science Consultant 

Lazarina is an organic marketing consultant specializing in SEO, CRO, and data science. She's worked with countless teams in B2B, SaaS, and big tech to improve their organic positioning. As an advocate of SEO automation, Lazarina speaks on webinars and at conferences and creates helpful resources for fellow SEOs to kick off their data science journey. Twitter | Linkedin