SEO’s Guide for Topic Modeling – Stop Leaving Traffic on the table

A single blog post can rank for 1000+ keywords. For example, if you have an optimized page for cat food, Google is able to recognize searches such as cat snacks, cat treats, cat meals, etc. as relevant and your site is eligible to appear for all these keywords. As a Digital Marketer, you need to target a topic rather than a keyword for your next SEO campaign.

When you search for a keyword Apple, how does the search engine decide which results to show? Apple the Fruit or Apple the mobile company. It depends on the user intent if the search engine knows that more people searching for Apple, clicks on links from Apple the company, the results from the mobile company will get prominence.

With the 2013 hummingbird update, Google has gained the immense capacity to understand the content, context, and user intent.

The search results are not just driven by keywords, there are multiple factors into play – Keyword, content, context user intent, and local factors.

To understand this better, let’s run a search on Amitabh Bachchan, the popular Bollywood actor from India.

The first search Just ‘Amitabh Bachchan’ returns generic results about his personal profile and latest news. Without a qualifier word, the search engine does not understand what I am looking for about the actor. His bio, Wikipedia page, and recent news get prominence.

Search without a qualifying keyword

The second search ‘Amitabh Bachchan movies’. Google understands the intent behind my query, and show me results related to his movies. The search results include a Knowledge graph with the popular movies, a different Wikipedia page about movies, and his IMBD profile.

Search for Amitabh Bachchan Movies

The Search for Amitabh Bachchan Music returns a totally different result.

Search for Amitabh Bachchan Songs

The search for his controversy related to the Panama papers leak returns a totally different result

Search for Amitabh Bachchan Panama

Of Course, the results are driven by user intent. Adding a qualifier word, in the end, tells the search engine what information I am looking for about the actor.

The term Amitabh Bachchan is associated with so many different topics – Movies, music, reality shows, Controversies, politics, and fans.

How does the search engine decide how ranking should change when you search for movie, music, or reality shows of Amitabh Bachchan.

Let’s find out…

What is topic modeling-

A Topic model is a computer program to extract topics from texts. It determines the topic by looking at the list of keywords appearing statistically in a meaningful way. A Topic model can analyze blog posts, a book, journal articles, or any kind of unstructured text.

While Google Algorithms are complex black boxes, researchers, and content marketers learned through experiments, trial, and error what can really work in content marketing and SEO.

By having a basic understanding of how search engines interpret the text, quality of content, context, and user intent, you can produce quality content that is highly relevant.

How search engines use topic modeling for content ranking –

Google’s algorithm discovers and ranks authority websites with deep content coverage on a given topic. The more in-depth content you write about a niche, the better your chances of ranking higher in search results.

The depth of your content is decided by the coverage of a given subject. A search engine will use a topic model to discover the relationship between a topic and different words, search terms, and phrases. Google might be having a massive database of semantic relationships between various terms.

In order to rank better, you need to create a content cluster covering various search terms and related terms on a topic. An interlinked cluster of deep content would mean an authority site on a niche.

Topic Clustering for Authority

The search engine’s job is to satisfy a user’s intent on every search.

You need to decide on a focus topic and the subtopics on a given subject. The next step is to discover various terms, phrases, and keywords related to that topic and sub-topics. A search engine would analyze a million pages to understand the relationship between various topics, sub-topics, and key terms.

You may not be able to replicate what search engines do, but just by understanding the key concept behind topic modeling, you can easily create in-depth content in your niche. If you want to understand the topic models, refer to the documents below.

Further reading on topic models

Term Frequency-Inverse Document Frequency

Latent Semantic Analysis

Latent Dirichlet Allocation

Topic clustering process –

We have no way to know exactly which topic model or modeling tool Google might be using. But here are some factors (Content marketer’s wild guess or learning from trial and errors.) which Google might be using to decide the relationship between two words, terms, or phrases. Google has a massive database or corpus of synonyms and has a very good understanding of how they relate to each other.  The Search Engine is able to relate different keywords, synonyms, and their relative importance for a topic. Here are some common relationship –

Semantic relationships

The search engine might look at the co-occurrences, how two terms are related to each other, how frequently they appear together, and if do in what order? How frequency impacts the topic and the information related to it. This is how Mashable defines Semantic relationship

semantic search definition by Mashable
Semantic search definition by Mashable

Keyword Proximity

The searcher may use a different combination of keywords, a search engine needs to understand how the distance between two words impacts the search intent. When two terms appear together in a document, what is the distance between them? Do they appear close to each other? How authority websites treat these terms.

Keyword Proximity definition by Searchmetrics
Keyword Proximity definition by Searchmetrics

Citations and Anchor text connections

How other websites or blogs are linking to those terms and phrases. The search engine would scan the blog, forum, websites for those terms, how they are cited or linked to.

Co-citation and Anchor Text
Co-citation and Anchor Text

Trusted domains

The search engines use the trusted domain as their baseline. How trusted domains and authority sites are treating these terms and phrases would determine the relationship for Google. What frequency they co-occur, where do they co-occur (Headlines, main content, sub-topics, etc.)

Queries from searchers

The Search engine’s biggest advantage is the searcher’s data. How searchers use a combination of terms and phrases in their query and the link they click helps the search engine determine the relationship between various terms and phrases. The Rankbrain update helps Google understand the searcher’s reaction in real-time and modify the search results on an ongoing basis.

How topic modeling helps with SEO

Understanding of topic modeling and SEO can help in organizing your content in topic clusters and content themes thereby driving a razor focus in your niche and creating more authority with in-depth content. You can use topic models To-

Improve Existing Content

The topic modeling can help you improve content for conversion, quality, and relevancy. By creating topically relevant content, you can rank better for relevant terms. This will also mean razor targeting users and creating better value for them.

Discover Relevant Terms

Topic modeling helps you understand the words you should include in your blog or article to increase topical relevance. Topically relevant content can rank 100s of different keywords related to that topic and drive a diverse set of visitors. If the length and depth of your content coverage on the topic are great, the search engine is likely to treat you as an authority site in your domain.

Discover content Gaps

In order to build topical authority, you need to ensure that there are coverage gaps. The topic modeling exercises help you discover such gaps. You can improve your existing content to cover such gaps or add new content as required.

Exclude off-topic posts/keywords

If your website contains a good number of off-topic posts, it will confuse the search engine. You need to interlink relevant content, top of the funnel content should be higher up in sitemap. Topic modeling exercise will help you decide, main and subtopic of your site and also exclude content that is not relevant. You can either remove off-topic posts or add relevant content to improve topical relevance. This will ensure your site is focused on a topic and thereby driving a better topical authority.

Earn better ranking

The better ranking is the result of topical authority and relevance. You can expect a 5-6% increase in organic traffic by increasing content relevance.

How to create a topic model for your next Article –

Search engines have a vested interest in maintaining secrecy about how their search algorithms work. There is no official confirmation on how topic modeling and clustering works. It’s a work of hit and trial by marketers and SEO folks to best guess how a topic is related.

To start with topic clustering, understand the focus topic and sub-topic in your niche. Create a list of topics and various terms associated with it. The next step is to create pillar topics and sub-topics. Download this quick guide on how to create a topic model using pillar page models.

The next step is to use the following tools

Search Suggestions from Google –

When you search a topic in Google, the bottom of the search results page will show a list of sub-topics and search terms that are ordered by the user intent. Create a list of these terms in an excel.

  1. Run a search with your main topic and sub-topics. For every search, Google will show related search suggestions. Make a note of all keywords appearing in related searches.Search Related to Amitabh Bachchan
  2. Next, open top 5 search results from each query. Make a note of H1, H2 and H3 titles
  3. Run a google search with a Question. i.e. Which is the first movie of Amitabh Bachchan. The Google answer box will appear. Expand the answer box by clicking on the questions. Make a list of all Question titles

Search for Amitabh Bachchan Answer box

Udemy course content structure –

Udemy has done great work on creating the topic structure for their courses. This is particularly useful if you are working on a popular niche. Search for bestseller courses in your niche. Make a note of the Course title and chapter headlines of the popular course.

Additionally, you can also look at the table of content of any famous book on your topic to understand various topics and sub-topics in your niche.

Authority websites in your niche

Search for the sitemap of any authority site in your niche to understand the organization of the topic and sub-topics. Do this exercise for 2-3 websites.

Common sense –

There is no replacement for common sense, if you are an authority in your niche, you already have a good understanding of various topics and sub-topics.

The relation between topic modeling and keyword research tools

Keyword tools extract a list of keywords from a database. The Keyword research tool will not be able to give you relationships or the relative importance of a search term. They do not give data on co-occurrence, relevance, or any other signal. You need to use a mix of keywords and topic modeling tools to understand the keyword hierarchy and topical relationship.

Tools for topic modeling


Topic Modeling FAQs –

What is semantic connectivity in SEO?
What are the LSI keywords and why they matter?
How do I determine the depth of coverage on a topic?

Further reading


Leave a comment