Semantic search has been a popular topic of conversation for years - and for good reason, as it created a massive shift in how modern SEOs evaluate and plan content.
In this guide, I'll run through how you can integrate semantic search into your SEO strategy, step by step, for even better results.
While it sounds complicated, the term "semantic" simply means the relationship between words and the meaning of words when combined.
Google uses semantics to understand a searcher's query intent and return better results. Semantics has moved Google from a keyword-matching engine to an intent-matching engine.
When you incorporate semantic search into your strategy, you use related topics and words to add meaning and depth to your content.
You're no longer targeting keywords; you're targeting topics.
For example, Google uses semantics to understand when someone searches for "how to make a cake," or “making a cake” they’re looking for "cake recipes."
With the engines understanding semantics better, you may not even mention directly on your page the phrases you rank for.
Almost any language has words with multiple meanings. Search engines rely on the context provided by previous searches and other terms within the current query to understand which meaning the user wants.
Search engines do the same when analyzing your pages; if you mentioned "Apple," do you mean the brand or the fruit? If you said "Tim Cook," "Steve Jobs," or "Cupertino" in the same article, the confidence that you’re talking about the brand increases.
Once upon a time, it was enough to spam a page with the keyword you were looking to rank for.
Google has taken giant steps over the past ten years to improve the way they understand semantics; here are some highlights.
The Knowledge Graph was Google’s first step toward a deeper understanding of the meaning behind words.
A good competitor analysis helps you better implement your knowledge graph.
One way Google uses the Knowledge Graph is to display knowledge panels in search results.
The Knowledge Graph is a map of entities and their connections to other related entities. Entities are "things," so that could be:
Google states that information for the knowledge graph comes from various sources, including public authorities, license data, and factual information directly within content or structured data.
The knowledge graph helps add information to the SERP, but it’s also potentially beneficial for site owners by giving their users at-a-glance information such as:
Hummingbird is often referred to as an algorithm update, but it was much more than that.
Think of other algorithms, such as Panda and Penguin, as small parts of a car engine, like the air filter or the pistons. Hummingbird is akin to the main engine block. When that gets changed, it can’t really be considered the same engine as before.
A core reason for the Hummingbird update was to bring semantic search to the forefront of all search queries. It does this by applying the understanding of entities from the knowledge graph to search results.
With this update, Google moved to understand the intention and meaning of words rather than merely matching the text within content to a search query.
RankBrain is a machine-based learning system that is part of Google’s core algorithm.
RankBrain associates words with topics, likely by using software similar to word2vec as mentioned in this Google blog post.
By matching words with broader topics, RankBrain can associate never seen before queries with historical search data for similar queries.
User-behavior insights from that historical data are then used to predict the best result for the novel query, and the SERP is reorganized accordingly.
For example, RankBrain may associate the search "large cat that is a predator" with the search "big cats."
Nothing changed in how SEO works from this update because you can’t optimize for RankBrain. Optimizing for RankBrain is the same as creating targeted useful content for users, which wasn’t anything new to the industry.
you optimize your content for users and thus for rankbrain. that hasn't changed— Gary 鯨理／경리 Illyes (@methode) September 26, 2019
Bidirectional Encoder Representations from Transformers, better known as "BERT," helps Google understand longer search queries.
A search query is often a string of words, usually containing one or several keywords, instead of a sentence.
Typically, keywords within a query are matched to pages across the web.
With BERT, instead of looking at a word within a query and matching each word individually to content, BERT infers the intent by looking at how the query words could change your meaning when used together.
Google’s BERT announcement post has some great examples of BERT in practice.
Now you understand the importance of semantic search, let’s run through some steps of what you can do to incorporate it into your strategy!
First, the term "semantic" means it’s all about content. To include the right semantically related topics, you need to improve the way you research.
As always, start with improving your keyword research. Here is how I begin my analysis.
Build out key topics you want to cover, highlighting broad, high-volume keywords you’d like to rank for.
In this example, I’m researching the topic of "chanca piedra," a tropical plant with medicinal properties.
Use Advanced Web Ranking’s keyword research tool to get a list of related searches for your broad topic.
Pick keywords that you consider relevant to your main topic and assign them to groups for a better management.
Once they are added to your project you can also evaluate keyword difficulty and decide which are the best keywords to target.
In addition to this, you could also use the Ahrefs Keywords Explorer.
Put the topic you’re writing about into Keywords Explorer, go to "Phrase match" in the left sidebar, and pick out a list of keywords with different intents.
Next, go to the "also rank for" report and select potential related topics. This report is excellent for discovering related entities a page is ranking for.
While we can't rank well for all these keywords in one article, they’re likely topically relevant. It helps to reference these keywords in your content and plan out related articles you can build specifically on these topics.
Next, search the keyword you want to rank for and take the URL of the top-ranking page.
Use the Top Sites and Top Sites in Time reports in Advanced Web Ranking for an in-depth SERP analysis.
To see if you can find more keyword ideas go to Ahrefs Site Explorer, put the URL into the tool, and then go to the "Organic Keywords" report in the left sidebar.
You can see the different related keywords that the page is ranking for; use it to further build your keyword list.
Next, go back to the SERP you were just on.
Use the people also ask (PAA) feature to discover other content ideas.
If you select a question, the PAA expands and shows questions related to the one you clicked. Use this to your advantage to find related questions that can enhance your content.
You can also use the related searches block at the bottom of the page for more ideas.
I often find I discover most of these via keyword research, but it's always handy to check.
Getting the intention right is critical. If the searcher is in the research stage, creating commercial content isn’t going to work. Instead, if relevant, cover the topic well and create pages for both informational and commercial content.
You can get an idea of the type of content to create by searching the keyword/term and comparing the content’s intention on the first SERP.
Detailed outlines for content helps to:
Here are some briefing tips for when you hand off to a writer.
Start by first creating a heading structure influenced by both your PAA and keyword research. Here is an example:
Since the Hummingbird update, Google understands how all questions and searches you discovered earlier are related, so don’t be afraid to create a comprehensive page instead of a page for each target keyword.
When briefing, provide your writer with related terms that could be included or explored. We discovered some of these when we used the "also rank for" report in Ahrefs.
As an example, we discovered earlier the phrase "Phyllanthus niruri," which is the scientific name for Chanca Piedra.
While we can consider the above information core to what we need to include, you should also use an expert on the topic to write your article or fact-check it.
An expert can ensure your content is factually correct and provides value above and beyond what could be provided by a copywriter.
They’ll also naturally include semantically related topics simply because it’s their field of expertise.
When hiring an expert, encourage them to provide their own insights; you’re likely to also get a unique angle that could be different from competing articles.
Experts won’t be required in some cases, as the topic isn't complicated enough to require a specialist. In these cases, continue with thorough research, briefing, and fact-checking yourself.
If budgets don’t allow the hiring of an expert, try to include quotes from experts on the topic you’re writing about. Again, this provides valuable information and allows multiple experts to discuss related topics that maybe you missed in your research.
To make sure you've included semantically relevant phrases in your content, use a text analysis tool to look for missing terms in what you've written.
The Ryte Content Success tool performs a TF-IDF analysis on a small corpus of data (the pages ranking on the SERP).
The tool picks out words that competing articles use and suggests you include them in your content.
It also has a content editor where keyword suggestions pop up while you write.
Frase.io is another excellent tool for researching and analyzing content.
It also scrapes and analyzes the content ranking for a given topic and provides insights other URLs have included on things like:
Be aware, your primary reason for researching related entities is to create better content for users. We’re not adding related entities for search engines.
As I mentioned, related terms should be within the content naturally if it’s been well researched and briefed, but tools like this can certainly help.
We've focused on what the audience reads, but now it's time to look beyond the visible content.
Some semantic HTML markup can help search engines understand your page layout.
HTML5 is the latest version of the HTML spec, and it provides semantic markup you can use rather than the standard <div> element.
An example of Semantic HTML5 tags are:
<article> <header> <footer> <main>
These tags can help crawlers understand exactly what the role of the content is, so it’s worth including.
Structured data provides context to what your content is.
For example, instead of Google trying to work out whether a page is an article or a product, structured data will directly say "this is an article."
John Mueller has previously explained the role of structured data in ranking and made it very clear its importance on Google’s understanding of the page.
“... we do use structured data to better understand the entities on a page and to find out where that page is more relevant.”
Adding structured data doesn’t directly impact rankings; however, Google having a better understanding of the content on a page can help.
By adding this markup, you can also acquire rich snippets, which can help your organic click-through rate.
Semantics play a role in organic search that has never been more important. While Google improves its ability to understand semantics, the fundamental principles of SEO remain the same. Continue to finesse your content creation process, from researching and briefing to delivery.
Hopefully, this guide has helped you better understand semantics and how it impacts the modern SEO process.