When people run a search, what are they really looking for?
Search has transformed considerably over the past decade, and semantic web technology has become one of the main reasons for it. Users now want search engines to understand their natural language like never before. And what can we say? Google can also understand the user intent behind the queries like never before.
In the old days, when search engines weren’t as complex than they are today, you could barely find the answer to your question in a matter of clicks. When someone typed in “MySpace vs Facebook”, they would probably have ended up with these websites in the first place instead of finding the difference between the two.
What do we have now?
Google not only recognizes the keywords “MySpace”, “vs”, and “Facebook” but also translates the user intent behind these queries to deliver the most precise results. Therefore, what we see is a featured snippet explaining how the two social networks differ from each other.
So, it’s just about bringing up the most relevant results for searchers.
What is semantics
Semantics studies the meaning in language, be it words, phrases, or sentences. For example, the word “child” can mean “kid”, “son”, or “daughter”—all these words technically mean a “child” but each has its own subtle shade of meaning.
And while discussing the terminology may seem a bit tedious, the role of semantics in SEO has become very central in recent days.
What is semantic search
As opposed to lexical search, semantic search is about understanding the meaning of the query instead of simply looking for literal matches.
It means understanding the user intent within a specific context, sorting through hundreds of billions of web pages, and answering the query in the best possible way.
The results of the semantic search bring together the aggregated information from several websites to answer the query in more detail.
In the example with the Eiffel tower, we get both the correct answer and the information about other popular attractions along with their construction dates.
Due to semantic search, people can type in something specific like “free CMS for the small online shop” or something asked in a natural human language like “what CMS should I use for my website?“—and in each case get satisfactory results.
It’s also important to note that often people use quite broad search queries in the hope of getting specific results. By typing something vague like “free CMS“, users don’t explain whether they are looking for a specific freeware or they simply want to find out if free CMS meets their business goals.
To a human, this request could mean a lot of things, but thanks to algorithms, Google can figure out what results will be more relevant given the context.
How search algorithms work
To meet the highest criteria of quality, Google takes into account numerous factors that eventually determine what you see in search results.
For this, Google created a complex system consisting of many algorithms that crawl, rank, and deliver the relevant pages for a particular query. These algorithms have been undergoing minor and major changes for decades and usually cause a lot of fuss.
Let’s now take a look at the recent core updates that have changed the way the Google algorithm works:
With the updated Google’s Hummingbird algorithm, Google transformed the way search results are returned. By implementing the “controversial search”, people can get more precise results that are based on the meaning behind the search queries.
For example, when you are looking for a coffee shop, Google will show you a list of places nearby, if you’ve allowed location sharing. Knowing your location may help search engines go beyond just showing pages with matching keywords.
Search features that go beyond the blue link
Besides showing users better results, Hummingbird also promoted knowledge graphs, rich results, and featured snippets to enable users to see the detailed answer on the surface with no need to click on the links in search results.
To deal with difficult or conversational queries, Google introduced a technology that could analyze the full meaning of a keyword in relation to the other words in a query, not just one by one. This is called Bidirectional Encoder Representations from Transformers, or BERT. With this technology, people can search in a way that feels natural to them, using prepositions (like “to” or “for”) and human language.
However, with increasing search requirements, BERT will probably be replaced by the newer model soon.
This new algorithm is called SMITH, or Siamese Multi-depth Transformer-based Hierarchical. As researchers suggest, SMITH is far more proficient in understanding entire documents and can predict the next passage within the context, whereas BERT can only predict words.
Although SMITH outperforms BERT in a way, the latter one will not be entirely substituted. The newer algorithm will support BERT by doing the more complex tasks that BERT cannot do.
So far, no one knows when the new technology will go live.
Danny Sullivan, Google’s public liaison of search, confirmed in his tweets that the SMITH algorithm doesn’t function yet.
We publish a lot of papers about things not used in Search. I won’t be making a habit of confirming each one someone might speculate about because it’s time consuming & more important, we have tended to proactively talk about this stuff already. That said. No. We did not.— Danny Sullivan (@dannysullivan) January 13, 2021
MUM (Multitask Unified Model)
To help users with complex queries that cannot be answered directly with a snippet, Google has introduced a new technology called Multitask Unified Model, or MUM. At the moment, Google is developing this technology, and it’s expected to bring MUM-powered features in the following years.
Let’s now take a look at how MUM will be able to help us with complex queries.
Let’s say you want to calculate the cost of traveling to Northern Italy. If you ask a person who’s already been there, you’d get a concise answer that would include all the details regarding the flight, hotel, food, etc.
If you asked Google the same question, you’d probably take additional searches: you’d have to look for flights, car rental services, accommodation, and other things one by one. After a string of searches, you’d finally gather all the information that answers your question. So far, algorithms are not advanced enough to give you the answer right away, however, with MUM, Google will soon be able to do that. Thus, you’ll need fewer efforts to get the job done.
How Google finds relevant pages
Now, let’s move on to the key criteria that determine what results users see for a particular query.
Meaning of the query
It stands for establishing the intent behind the user request. For this, Google created language models to determine what keywords to look up on the web. Such language models detect misspelled search queries, understand natural language, match synonyms, and more.
With the synonym system, users are provided with more relevant websites even if they do not contain the exact keywords used in the search.
This helps Google match the query “how to travel on a budget” with sources that contain information about budget-friendly flights, cheap hotels, travel insurance, and other helpful tips related to low-cost traveling.
The second step is to evaluate the content on the web pages and determine whether it is topically relevant or not.
First of all, algorithms analyze keywords: if they appear on a page, headings or the body, this source is more likely to be shown to a user.
Apart from this basic signal, Google evaluates the topical authority of content that includes: in-depth content accompanied by subtopics covering all possible questions, internal linking, and backlinks—that shows how the information meets the search intent.
According to Dixon Jones, an international SEO Speaker and Internet Marketing advisor, just publishing long-form content doesn’t play a key role in earning first-page rankings.
One of the things that strengthen the relevance of a page is Latent Semantic Indexing (LSI) Keywords, or in other words, supplementary keywords that contribute to the topic.
Let’s take a look at our example about “how to travel on a budget”.
There are hundreds if not thousands of web pages containing information about traveling on a budget. However, not all of them are considered relevant. For instance, if Google sees that the keyphrase “travel on a budget” is repeated over and over again, such a page will unlikely be displayed in search results. On the other hand, when you use LSI keywords (like “low-cost”, “discount”, “cheap”, etc”), Google will be confident that this post is about traveling on a budget.
However, be careful not to overdo it. It may seem right to use as many of them as possible. But that is not a good idea. If you scatter all of them throughout the text, your content may seem spammy to search engines and even lead to penalties.
Craig Campbell, an SEO expert from Glasgow, recommends using semantically related keywords moderately and not overdoing it.
Apart from matching the keywords and LSI with the user queries, algorithms also evaluate the reliability of sources. Google PageRank is one of the many elements that make up the Hummingbird algorithm and help it prioritize links that have more value than the others. Long story short, PageRank is designed to calculate link votes and tell Hummingbird which sources demonstrate expertise and authority.
Along with lots of other factors, this one also affects page rankings.
According to Google, 27% of the online global population is using voice search on mobile devices.
As more people today use tools like Siri, Alexa, or Google Home, algorithms are developing to recognize voice search. And because such search queries are more conversational and contain longer, more natural phrases, Google uses semantic search to provide better voice search results.
With this rapid adoption of voice search, you need to determine what queries they will use and customize your content.
What is semantic SEO
Now that we know how the algorithms work, let’s see how to write content given the role of semantic search.
Semantic SEO is the process of optimizing your content for a topic instead of a keyword or a phrase. As algorithms are now getting better at understanding the user intent, you have to provide answers not only to the direct query but also to whatever stands behind it. This means answering additional questions to complement the query.
How to use semantic SEO for higher rankings
In the old-school SEO, you could choose a keyphrase, use it in your title/description, headers, and through the text, and hopefully, rank for this keyphrase. However, with the development of semantic search, the game has become more challenging, particularly for webmasters.
Fili Wiese, a technical SEO consultant at SearchBrothers and ex-Google engineer, suggests taking a more comprehensive approach.
To use semantic SEO properly, you can roughly divide your work into two stages: the content creation stage and the optimization stage.
Content creation stage
The first stage will include: choosing a topic given the user intent, collecting a list of relevant keywords, and creating the content structure.
Your action plan for the content creation stage will look as follows:
- Create a keyword list. Make a list of related keyphrases and LSI keywords on the topic semantically related to your target keyphrase.
- Identify long-tail keywords that address your topic and have a clear intent.
- Clean up the keyword list by combining similar queries into groups, also known as keyword clustering, or grouping.
- Use semantic markup.
Use Google to create a keyword list
While you should not rob yourself by drawing keyword ideas solely from Google, it is definitely the right place to start.
To kick things off, think of the keywords you want to rank for in Google. Let’s say you run a yoga studio in LA. You definitely want to rank for such commercial keywords as “yoga school” or “yoga studio LA”, but you also want your blog to rank for all sorts of informational keywords like “downward-facing dog” or “meditation techniques”. And here’s how you can find all of the most popular queries.
The moment you start typing “yoga” in the Google search bar, you’ll see a number of suggestions. They are all a mix of keywords that are trending on Google and the search terms Google believes you’ll be interested in based on your search history.
Don’t forget to add long-tail keywords to your list. These are usually well-considered queries that are used in the form of questions and sentences.
Consult the section “People also ask” and use those queries in Google search to check what autocomplete results it will bring.
Or, scroll down to the bottom of Google page one and find a list of related search queries that can also bring you the ideas of long-tail keywords.
Google Search also allows looking for synonyms and LSI keywords. They are often confused with one another, however not all related search terms can be synonyms.
For example, “yoga asana” and “yoga pose” are synonyms, whereas “yoga class” is related to the previous phrases but has a separate meaning.
You can find some synonyms and LSI phrases with the help of the autocomplete feature or in the related searches at the bottom of the SERP.
You can go on and on this way to gather all the queries Google has to offer. And maybe looking at these keywords will help you come up with some more ideas of related queries of your own.
While using Google is a surefire and old-school way of collecting semantic data, you can also speed things up and get better results with specialized semantic SEO software.
Search for keywords with SE Ranking
SE Ranking’s Keyword Suggestion Tool boasts a huge database that goes beyond search engine suggestions. It hosts over 2 billion unique search queries, which means you can have thousands of keywords ideas generated in one click.
For every keyword, you can see plenty of metrics including the keyword’s Google search volume—the number of searches the keyword gets each month. The yoga topic, for example, has 301K monthly searches.
You can use filters to leave out keywords that you don’t feel like taking on. For example, you can filter out keywords with a really high search volume, which as a rule are too vague and overly competitive.
For even more ideas, check the Related and Long-tail keywords tabs. While the Similar keywords tab only features search terms that contain your seed keyword, under the Related keywords tab, you can find all kinds of semantically related queries.
Finally, SE Ranking allows you to easily draw keyword ideas from your competitors. You don’t even need to choose which competitor to analyze—the Competitor Research tool will identify your major SEO rivals and will come up with a list of keywords your competitors rank for while you don’t. That way, you can discover some not-so-obvious topics related to your business niche.
Once you’ve collected keywords from all possible sources and compiled everything together in a spreadsheet, your next step will be to remove any duplicates and move on to keyword clustering.
Look at your list and think about how you can categorize keywords into different groups.
For example, we have a long list of keywords about “yoga asanas”.
Thus, you can group them:
- By their semantics: “yoga poses”, “yoga postures”, “asanas list”, etc.
- By user intent: “how to do yoga asanas”, “what are the beginner yoga poses”, etc.
- By search volume: high-volume keywords and low-volume keywords can help you develop your site’s structure.
Well, that’s about it for traditional keyword research. Pull ideas from search engines or use SE Ranking to speed things up, add new topics to your content plan and create new blog posts and landing pages for your website.
Write semantic markup
Using semantic markup is just as critical for rankings as careful keyword research. Semantic markup, or in other words semantic HTML, is a set of elements that clearly defines your content and gives meaning to it.
For example, an <h1> tag indicates that the embedded phrase is the title of your article. This is both presentational and semantic markup, as you show users and search engines where the title is.
Examples of semantic HTML tags include:
- <h1> through <h6>
With the help of semantic elements, some parts of your content, like lists, can show up as a featured snippet.
Make your content richer, not longer
Simply stuffing your content with the right keywords won’t cut it anymore, your content has to meet users’ intent to rank high. The key is to think about the topic, not just keywords. And your task is to properly interpret these keywords to best address your customers’ needs.
Here’s what Lee Wallis, Head of Digital at Excite Media, Winner for SEO—Australian Web Awards 2021, suggests:
Optimize your page
The next step is to finetune your SEO semantic writing by optimizing the content using structured data, internal linking, anchor text, and more.
Use structured data
To help Google better understand your content, use structured data—a format for classifying the content on a page. Structured data is a language that explains to search engines what content to display in an attractive way.
Let’s look at the AccuWeather website. The general information about their company is covered by Google’s Knowledge Graph: when it was founded, who the founder is, the customer service hotline, subsidiaries, etc.
AccuWeather just helped Google understand the content of a page by including structured data.
Because the structured data identifies particular elements of the weather forecasts, users can see the latest news right in the search results.
Internal linking in semantic search
A well-considered internal link structure will help your content show up at the right place in search.
The structure contains three elements:
- menu link structure
- breadcrumb link structure
- internal link structures in the body of the article
By linking pages together, you not only help readers navigate your site but also help Google crawl it easily and discover new pages. For example, a company that sells sneakers can have a general “sneakers” page that appears higher in search results than a “kids sneakers” page for the query “sneakers for kids”. This seems illogical. However, if the main category links to the “kids sneakers” page, it can help Google prioritize the subcategory page for the given query.
In other words, the smarter your content is linked to each other, the better effect it will have on Google and users.
When setting up internal linking, make sure that your page contains topical anchor texts.
In the HTML code of a page, an anchor text looks like this:
<a href=”https://www.example.com”>Anchor Text</a>
Say your page has a lot of links with anchor text about bikes. Then, Google will use it to evaluate the relevance of your content. Whether you’re using external or internal anchor text, there are some tips to remember:
- Use natural anchor text that provides relevant benefits to users.
- Incorporate long-tail keywords to keep things natural.
- Combine different types of anchor text, for example, branded anchor text or link lists.
Creating content and optimizing it for search is all about relevance and value. Even if your page has high authority, it won’t rank if the content isn’t relevant for your topic. According to the basics of semantic SEO, you have to adjust your pages to natural language search, answer as many questions as you can, and optimize your content to help Google read it properly.