Semantic AI-powered SEO Keyword Research Course

Traditional keyword research primarily focused on keyword targeting, density, and difficulty metrics. In contrast, semantic keyword research centres on understanding the user’s holistic search journey, including their content and information needs, motivations, and the context of their searches.

It moves beyond just finding keywords to target on a page, instead aiming to understand the underlying intent, the expected content format and platform, how queries connect, what information the user already possesses, and the brand’s ability to produce relevant content across various platforms. This user-centric approach ensures content is strategically created to address specific user needs and fit seamlessly into their search behaviour.

Keywords are phrases with SEO value, while entities represent real-world concepts like people, places, organisations, or events. In semantic keyword research, entities are crucial for deciphering search queries and their underlying intent.

The Entity-Attribute-Variable (EAV) model helps categorise entities by their characteristics (attributes) and specific values (variables). Google leverages entities extensively through its Knowledge Graph, a structured database connecting facts, entities, and their relationships. This allows Google to understand the context and meaning of queries, disambiguate terms (e.g., “London” in the UK vs. “London” in the US), generate knowledge panels with factual information, refine query suggestions, and power question-answering capabilities. For SEO, understanding entities helps create content that aligns with real-world concepts, improving relevance and visibility.

Information gain is a score that quantifies how much new and useful information a document or webpage provides to users who have already seen other content on the same topic. Google uses this score to prioritise documents that offer novel insights, aiming to prevent repetitive information in search results and enhance user engagement.

The score is calculated using machine learning models that analyse page content, salient extracted information, and semantic representations (e.g., embeddings, bag-of-words). Beyond ranking, information gain helps Google identify and de-duplicate web pages, expand entity collections for the Knowledge Graph, and generate related query/entity suggestions.

For content creators: focus on proprietary or unique insights, regularly update content, incorporate expert opinions, and curate niche material that fills existing information gaps rather than replicating what’s already available.

Google’s Autocomplete APIs (for Search, YouTube, and Maps) offer real-time keyword suggestions based on user input. These predictions are driven by language modelling, frequency models, and reinforcement learning, and become increasingly personalised based on user history and group behaviour.

For keyword research, these APIs reveal Google-suggested query paths, refinements, and augmentations—showing what users are likely to search for next and which related terms trend. While you can’t directly access Google’s extensive user behaviour data (e.g., CTR, dwell time, pogo-sticking), you can infer satisfaction and intent by analysing your own site data (e.g., Google Search Console queries combined with GA4 engagement metrics) and benchmarking competitors. This helps identify content gaps, assess effectiveness for specific queries or entities, and tailor strategies to improve UX.

SERP (Search Engine Results Page) features are elements beyond the traditional “10 blue links,” such as Featured Snippets, People Also Ask boxes, Knowledge Panels, image carousels, and local map packs. Google constantly experiments with and refines these features to enhance UX by providing more direct and diverse answers.

They’re crucial for SEO because they increase visibility, can improve click-through rates, diversify traffic by capturing wider intents, and provide insight into user preferences (e.g., preferred content formats like video or images). Analysing SERP features helps identify content gaps, strengthen competitor analysis, and align your content with how Google presents information for specific queries.

Explicit search intent refers to the user’s clearly stated goal in their query. Traditionally, it includes three main types:

• Informational: Users seek to learn about a topic (e.g., “what is a SERP?”).
• Navigational: Users want to reach a specific website or location (e.g., “YouTube”).
• Transactional: Users intend to complete an action, often a purchase (e.g., “buy running shoes”).

A fourth common category is Localised Search (e.g., “restaurants near me”). These classifications can be done using rule-based approaches (matching keywords like “how to,” “buy,” “near me”) or machine learning models that analyse query text, SERP features, and entity data. More nuanced micro-intents (e.g., “discount” within transactional queries) can be identified by reverse-engineering GSC data and competitor analysis.

Query context refers to the immediate context provided by the user’s search, encompassing both a macro context (broad domains like “medicine” or “sports”) and a micro context (specific terms or subtopics such as “oncology” or “treatment options”). Search engines combine these to deliver highly relevant results.

Session context considers the entire sequence of searches within a session (or across sessions). Google uses this to understand evolving intent and prior knowledge. For example, if a user consistently engages with certain content types or domains, similar results may be prioritised later; conversely, results that didn’t earn engagement may be down-ranked to surface new information. For content strategists, mapping these contexts supports designing content paths that mirror user journeys, interlinking assets, and avoiding cannibalisation.

Unlike traditional keyword research (often reliant on third-party tools and basic metrics like search volume), semantic keyword research emphasises a comprehensive, multi-source approach. Key data sources include:

• Google Search Console: Your ranked queries and performance.
• Competitor Keywords: Both business (revenue) and organic (traffic) competitors.
• Content Gap Analysis: Topics/queries competitors rank for but you don’t.
• People Also Ask & Autocomplete: Common questions and user-suggested query paths.
• Trending Keywords: From sources like Google Trends and Exploding Topics.
• First-Party Research: Interviews, surveys, chat transcripts, and the EAV model to uncover patterns and needs.
• SERP Analysis: Formats, platforms, and features Google prioritises for different queries.

Analysis then categorises keywords by brand, search intent, length (short/long tail), clusters, topics, entities, SERP features, content type/platform, content depth, and even user personas. This multifaceted approach leverages ML and APIs for entity extraction, semantic clustering, and intent classification—enabling a deeper, more nuanced understanding of user needs and the search landscape.