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 a search for a keyword Apple, how does 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 a prominence.
With 2013 hummingbird update, Google has gained the immense capacity to understand 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.
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 Knowledge graph with the popular movies, a different Wikipedia page about movies and his IMBD profile.
The Search for Amitabh Bachchan Music returns a totally different result.
The search for his controversy related to Panama papers leak returns a totally different result
Off 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…
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 on 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 which is highly relevant.
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 relationship 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.
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 of 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.
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 –
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
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.
How other websites or blogs are linking to with those terms and phrases. The search engine would scan the blog, forum, websites for those terms, how they are cited or linked to.
The search engine use 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.)
Search engine’s biggest advantage is 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.
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-
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.
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.
In order to build a topical authority, you need to ensure that there is 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.
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 which 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.
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.
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. 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.
Next step is to use the following tools
When you search a topic in Google, the bottom of the search results page will show a list of sub-topics and search terms which are ordered by the user intent. Create a list of these terms in an excel.
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 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 topic and sub-topics in your niche.
Search for the sitemap of any authority site in your niche to understand the organization of topic and sub-topics. Do this exercise for 2-3 websites.
There is no replacement of common sense, if you are an authority in your niche, you already have a good understanding of various topics and sub-topics.
Keyword tools extract a list of keywords from a database. The Keyword research tool will not be able to give you relationships or 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.