A decade ago, search was all about keywords.
Search engine algorithms relied on pure text based analysis of the pages of your site to figure out what your site was about. This approach ended up failing because spammers found ways to manipulate those algorithms. Then, Google came along with an algorithm that analyzed links to your site from other sites to see how authoritative you were on a topic.
The link based algorithms are better, but they still have a high dependency on analyzing the keyword strings on web pages to understand what they are about. You can refer to this as the algorithm remaining dependent on "strings" (as in strings of text on your web pages).
There are two major problems with this approach:
1. It requires users to be smart enough to enter in search phrases that are mechanically constructed to bring up the right results, and this was often a tricky process for many people. Short search phrases (for instance, "window screens") actually return better results than a longer one ("I want new window screens for my 2-story house"), although the longer one might be more descriptive of what the user really wants.
2. It requires users to clearly understand and communicate their intent. For example, a query on the word "jaguar" can mean a car, an animal, an operating system, or a guitar, or even be a reference to a football team (which latter intent would be better expressed as "jaguars" but mis-spellings are common in search queries).
The concept with semantic search is to develop algorithms that deal with both of these problems. One of the first steps in this process is to teach these algorithms to understand the relationships between things. Here is an example, using a sports team, the New England Patriots:
- They are a sports team that plays football in the National Football League
- They play in the Eastern division of the American Football Conference
- Other teams in the same conference are the Buffalo Bills, the Miami Dolphins, and the New York Jets
- They have players on the team, including Tom Brady, Vince Wilfork, and many others
- They have the 29th pick in the upcoming NFL draft, which starts on May 8th
- The team has a schedule of games that it will play in the 2014 season
- and much more
These are all relationships that humans understand quite intuitively, and that search engines are working to understand as well. There is more to semantic search than this, though. Consider the ability to understand context. If I run a search on "French restaurants," the search engines know that I am located in Southborough, Massachusetts, and will offer a "carousel" of French restaurants near me:
There is more to context than just this simple example. Let's look at a sequence of queries that combines a knowledge of relationships and "conversational context," starting with "Show me pictures of the Empire State Building":
Now, let's show off a little bit of semantic search in action with the following query: "How tall is it?":
Already we see some semantic magic. To process this query properly, Google had to understand that the Empire State Building was a thing with an attribute called height. In addition, Google had to remember the prior query and recognize that "it" refers to the Empire State Building.
In fact, you can go further with this and try another query, such as "show me pictures of its construction":
Traditional keyword based analysis (an analysis of the "strings") would fail to return good results for these queries. You can actually test this for yourself. Try the above query sequence using Google's regular search engine, and you will see that it does not succeed. Here is what the "how tall is it" query returns:
Then repeat the sequence of queries using Google's voice search function. You can do that by going to Google and clicking on the microphone icon on the right of the search box:
For now, a lot of semantic search concepts are wired by Google into their voice search functionality, but not the traditional web search using your keyboard. In essence, the voice search showcases much more knowledge of "things" and keyboard based search remains much more dependent on "strings."
Semantic search also understands much more, including where you are, the time of day, your past search history, and many other factors that help it process your queries quite differently. The following graphic helps capture some of the many components that go into semantic search:
Thanks to David Amerland for his contributions to this graphic.
Semantic search is all a work in progress that will unfold over many years, perhaps a decade or more, but the reason to learn about iy now is that it is starting to be used now. This impacts the way you think about the content you put on your site, and it affects the way you market your business. Semantic search pushes you to view the entire system in a very holistic way. Adapt to this broader mindset, and you will gain a clear edge over your competition who doesn't get it.
One great resource for learning more about semantic search is David Amerland's new book, aptly titled Google Semantic Search. David is great at simplifying the core concepts, making them easier to understand and take action on. The book shows how businesses are successfully leveraging these core concepts today, and thriving as a result.
Chapter 3 of the book has a wonderful example of a bakery business that combines many elements of semantic search to their advantage. Amerland tells the story of how this bakery shares its vision of how they make their bread - and persuades the potential buyer that they are really buying a work of art.
On top of that, they support a major cause - that of fair farming practices, and this becomes a new element in how they increase their exposure and reputation at the same time. This bakery crafts a campaign across multiple platforms to make sure they are where their customers are. In the process, they weave a powerful web of semantic context across the web, driving their search results through the roof.
The 11th chapter on the four V's of semantic search (volume, velocity, variety, and veracity) is simply brilliant. It takes all the material in the first 10 chapters and crystallizes it into clear concepts that provide the reader with a reference platform for guiding their future digital marketing strategies.
The book is not a one-size-fits-all solution. If you are looking for that, it does not exist. Instead, Google Semantic Search teaches us what we need to know about the search engine world we live in, then equips us with the information we need to make intelligent decisions about how to thrive in this end environment.
The reality is that there is no one single recipe for success for every business. What will work for one person or one business is quite different from what will work for another, and these differences are to be celebrated. In fact, understanding what makes us different and leveraging that in how you promote yourself online is one of the top keys to success in today's environment, and semantic search makes this more important than ever!