In 2018, Google admitted that over 15% of queries had never been seen before.

However, Google search is still able to serve relevant, high-quality content that satisfies a searcher’s intent when they input a search query — even when the queries are totally new to Google.

How can it possibly figure out what searchers want even though it’s never seen the queries before (or keywords don’t line up with the content served)?

Through semantic search.

What is semantic search?

Semantic search is when search engine algorithms incorporate search intent and the context of words (and how they relate to each other) to better satisfy user intent (what exactly a user is looking for) when a user inputs a search query.

This provides a better user experience because searchers are more likely to find the content they’re looking for even if Google has never seen their query.

In simpler terms, Google and other search engines are better able to provide relevant results to queries that aren’t entirely keyword-focused by understanding the natural ways that we use language — the contextual meaning of words. As in, not just their exact meaning, but how those words are related to each other and to other words.

Search intent is simply what a user wants to find when they input a search query.

For example, if a user asks a question, they want an answer. If a user types in something like, “What is the best cake recipe?” Google understands, through semantic search, that it should provide recipes to the user.

If a user types in something like, “Best TVs of 2021,” Google understands that the user wants to see products for sale on an ecommerce page, or maybe in a blog post review, even though the user hasn’t said specifically that this is what they’re looking for.

Google is able to understand the intent behind a query. In these cases, Google understands what type of content to serve to the user.

However, there’s a second element to semantic search — understanding how words are related to each other to serve relevant results to the user even when the query doesn’t have exact keywords or related keywords to the content it serves up.

This is best illustrated by an example.

A semantic search example

Say that a user is trying to remember a movie but doesn’t remember the name of the movie. Google will use the context of the words to serve up content that it thinks satisfies the user’s search intent, which is to find the movie.

An example of semantic search using Lord of the Rings.

In this case, the query never includes the keyword phrase, “The Lord of the Rings.” However, Google still understands that the query, “what is the movie with the elves and the ring?” is related to The Lord of the Rings.

Semantic search is the method by which Google has understood the intent behind the query so that it supplies the type of result the user is looking for — in this case, an answer to a question — by understanding the context and the relationship between words. It even takes synonyms into account.

Google knows that elves and rings are related to each other in a specific way. When the term “movie” is added into the mix, Google sees that these three words are related to “The Lord of the Rings.”

Because Google is serving an answer to a question, the results show up in the form of a featured snippet, which is content pulled from other sources and displayed directly on a search engine results page (SERP).

The featured snippet is the first result on the page and often gives users what they’re looking for without the user ever having to click. This is called a zero-click result.

How does Google accomplish this? It uses machine learning. Essentially, Google learns over time how words are related to each other and what types of results give users what they’re looking for.

While it may once have been the case that the query above didn’t make sense to Google and didn’t return relevant results, over time, Google learned how these words were related to each other to better understand what users were looking for.

This resulted in users finding the content they’re looking for even though the search doesn’t include straightforward keyword phrases.

Why does this matter to marketers? Because semantic search engines like Google and Bing see search intent as an important ranking factor. If you’re just focusing on keywords and not on a searcher’s intent, you’re less likely to have your content indexed highly for those keywords.

How did semantic search change SEO?

Before semantic search (as far back as 2010), the main focus of search engine optimization (SEO) was getting as many high-quality backlinks as possible (off-page SEO) and inserting keywords “properly” (on-page SEO) on a web page.

Many marketers focused on long-tail keywords (mostly) and never even considered latent semantic indexing (LSI) keywords because, while they existed, Google’s search algorithms didn’t take them into account.

After the release of the Hummingbird update, which was focused on revamping Google to take search intent into account, everything changed. No longer were keyword research and keyword searches as relevant as they were before.

No longer would practices like keyword stuffing into web pages reward marketers and SEOs.

The semantic web was born.

RankBrain came next in 2015, introducing machine learning into Google’s algorithms, further increasing Google’s capability of understanding search intent, taking into account factors like a searcher’s location, or using personal data to deliver personalized results.

An infographic of Google algorithm updates that shows the lead-up to semantic search over the years.

More updates followed over the years. One of the most significant was BERT, which took natural language processing to a new level in 2019. BERT uses machine learning and artificial intelligence to even better understand the context of words, making LSI keywords more relevant than ever before.

What this means for Google, its searchers, and marketers

Essentially, Google stopped just looking at keywords in order. It now looks at how keywords are related to each other (and to synonyms and to other words that aren’t even in the query) to better understand user intent.

The search experience was improved. Search intent became an important ranking factor. As voice search becomes more and more common, Google and other semantic search engines need to be able to better understand how natural language works. These updates address those needs.

Marketers now need to focus on “closing the loop.” A loop is when a searcher clicks on a result on a SERP, doesn’t find what they’re looking for, returns to the SERP to try another result, and continues this process until they find what they’re looking for.

Content that “closes the loop” is content that satisfies search intent. For example, if a searcher inputs “best cake recipes 2021” and gets a result at the top of the page for 2019, they’re likely going to return to the SERP.

Once they find an article for 2021, they will likely stay on that page and not return to the SERP.

The loop has been closed.

How to do SEO for semantic search

Because semantic search has had such a profound impact on how search engines work, SEOs and marketers now need to focus on semantic search when creating content if they want their content to rank and draw traffic.

Start with keywords

Keywords are still relevant in SEO, just not in the way they used to be.

It was once the case that marketers and SEOs could insert keywords in specific places in their content and be assured that their content (as long as it was a certain length) would likely rank.

Now, keywords are simply a starting point. They help you to better understand what searchers are looking for.

For example, say you’re a food publisher interested in creating content about the best cakes of 2021. So, you research “best cake recipes 2021,” and this is what you find…

Screenshot of a keyword tool showing keyword research tool showing related keywords for the search term "best cake recipes 2021."

You see that “best cake recipes 2021” only gets 10 searches per month. Related to this search, “trendy cakes 2021,” gets over 500 searches per month. If you’re interested in creating an article with the most potential for traffic, “trendy cakes 2021” is a better keyword.

Study the SERPs

When the phrase “trendy cakes 2021” is input into Google, this is the result.

A search results page for "trendy cakes 2021."

When the keyword phrase “best cake recipes 2021” is input into Google, this is the result.

A search engine results page for "best cake recipes."

Though the two keyword phrases are similar, the results on the SERPs are profoundly different. One search gives a list of trends (and not how to make the cakes), while the other result gives a list of recipes.

Even though these keywords may seem similar based on keyword phrases alone, it’s now clear that had you created a list of recipes for “trendy cakes 2021,” your content likely wouldn’t have ranked for that keyword phrase.

Determine intent

Intent has now been determined for each keyword phrase by looking at both the keywords and the SERPs. Determining search intent is critical because it provides a number of insights.

  1. First, it indicates the type of content that is ranking. For the keyword phrase “trendy cakes 2021,” the content that’s ranking is image-heavy articles showing pictures of cakes.
  2. Next, many of the results aren’t just about cakes in general, but about a specific type of cake — wedding cakes. The featured snippet in the image (which can be considered the top result that Google believes the searcher is looking for) is not of cakes in general, but of wedding cakes.

So even though the term “wedding” doesn’t appear anywhere in the keyword phrase, it’s now clear that this is the type of content that can rank for this keyword phrase — a list of wedding cakes. When people search this phrase, what they want to find is image-heavy articles that include wedding cakes.

This is major because, if you had just created content about types of cakes and disregarded the wedding aspect, your content might not have ranked even though you included content related to the exact keyword phrase.

Determine content format, type, and angle

Continuing with this example helps determine what the content itself should look like — how to create high-quality content.

The top result for “trendy cakes 2021” is an image-heavy article, not a video or a text-heavy article. This indicates that searchers want to see images (and not read a bunch of text or see a video of all the different cakes).

The same result is also a list of these cakes. In fact, the featured snippet, though it doesn’t show the images, does show a bulleted list of cakes (above).

If you want to create content that has the potential to show up in the featured snippet, you’ll need to format your content as a list and include lots of images — this is your content’s format and the type of content you want to create.

Finally, what’s your angle? What will set your content apart from the top-ranking content? It’s possible that creating an article with videos of how to make the cakes alongside the images might be more valuable to searchers and provide a fresh angle that Google will reward.

It’s also possible that making a much longer list would be rewarded.

Create high-quality content that closes the loop

Your goal is to satisfy search intent by creating content that closes the loop (the searcher stays on your page and doesn’t return to the SERP).

To do this, your content needs to be high quality. High-quality content is content that the user finds to be useful — more useful than other content that’s available.

Continuing with the example, a high-quality list of trendy cakes isn’t going to be 2 or 3 cakes — it’s going to include many cakes, and it’s going to include images of those cakes.

Those images need to be high-quality images — low-resolution images aren’t going to work. You also need many different images of each design, as can be found in the top-ranking content.

If you check all these boxes, your content is high-quality for this search query.

What makes high-quality content?

What makes content high quality depends on the type of content you’re creating. In the example above, the quality of the images was important, but what about other types of content?

For example, the search query, “what is semantic search?” returns results that are text-heavy.

Though there are some images, what Google has determined is the best result that satisfies user intent is a great deal of text that answers this question.

However, the article can’t just be a simple answer to the question that’s only a paragraph or two. Google has determined that the articles need to be long-form. The top-ranking result is over 2,000 words long.

What makes content high quality depends on what Google believes is a high-quality result. That’s why studying SERPs is so important. It’s how you understand what searchers are looking for and what type of content is getting rewarded with the top spot.

This applies to all types of different searchers and content types. For example, if all the top-ranking videos for a keyword phrase are 5 minutes long, it’s likely the case that a long video would struggle to rank.

However, if all the videos are 30 minutes long or more, then a short video is unlikely to rank.

Study the top-ranking content and look for commonalities to determine what constitutes high-quality content for each search query.

Add schema markup

Schema markup is code that tells Google what type of content your web page is made up of. It’s a critical component of achieving rich results.

Search engines struggle to understand what content is when it’s not text-based. Unlike humans, they cannot look at a web page and instantly tell that something is a product or a recipe, or just an article.

It helps search engines determine that you have a specific type of content that is relevant to the searcher’s intent.

Rich snippets make your results on a SERP larger because extra data can be displayed.

An example of a rich snippet on a SERP.

You can see in the image that not only does this result take up more space on the SERP, but it also has a great deal more data that you might not see in other results.

This has higher value for a searcher because they don’t have to click right away to see if the information they’re looking for is included in a search result.

It also makes it more likely that the loop will be closed if they do click on the result because they get a better idea of what kind of information is on the page before clicking.

To add schema to your website, you can use tools like a Schema generator or a plugin for a content management system (CMS) like WordPress.

Get a complimentary SEO audit

In the modern world of Google search, semantic search is more important than ever. If you want your content to rank, you need to study the SERPs, determine search intent, figure out what type of content is ranking, and create high-quality content that closes the loop.

Want to see how you’re doing with SEO? Get an instant SEO audit below. Or, schedule a free consultation to see how intent SEO can boost search traffic revenue by 700%.