The Rise Of RankBrain: What Does It Mean?
Search engine optimization has changed a lot since it first became a thing. The days of just making sure you had the words you wanted to rank well for scattered across your pages are so far gone that they’re practically forgotten.
SEO in the modern world is ever changing. The methods used to achieve good ranks have become more technical and varied. What works for some doesn’t seem to work for others, and what worked last month might not work tomorrow. The algorithms employed by search engines to determine result ranks have become both complex and virtually top secret.
Google is, by far, the most popular search engine, and the one most website owners are worried about ranking well on. Although the inner workings of their ranking system is a very well kept secret, the company has been open about general factors that go into the process of deciding who comes out on top. With their latest system updates, they made an announcement that the three biggest factors that make up search result rankings are:
- Content - Real and genuine content on the pages
- Links - Incoming links from other sites that validate the content’s value
- RankBrain - ??????
What Is RankBrain?
Rankbrain is the name Google has given to a machine learning system that is being used to help process the search results that come up when someone runs a web search on the popular search engine. It’s important to note that RankBrain is part of what goes into generating results, it is not the sole determiner, or the name for the algorithm itself.
According to information that has been released so far, the RankBrain system is used in about 15% of all searches run through Google. The system is able to learn on its own and continually improve the search results it returns, supposedly becoming more accurate with more use.
What Does RankBrain Do?
Nobody outside of Google knows exactly what RankBrain does, but they have given some fairly general clues about its effects, as they’ve been known to do in the past.
The RankBrain system seems to have been developed mainly to help the overall search algorithm react in smarter ways by bringing a sort of understanding of the context of search queries. To clarify that a bit, there are many words and phrases that show up in search queries that can have different meanings depending on how they are being used. It’s usually easy for a human to understand the difference, but making a computer program able to tell the difference can be tricky work.
Apples or Oranges?
To illustrate, take the word “apple” as an example. If someone asks you about the “Apple iPad”, you know they are talking about a tablet computer. Other words come to your mind such as tablet, computer, apps, games, etc. If someone asks you about “apple pie” you know they are talking about food, and you might think of things like cake, baking, fruits, etc.
In your mind, we could even say it’s like apples and oranges. Those are two completely different things. But to a computer, it’s not so easy to understand that difference. To a computer, apple is apple, period. Making the computer understand that one apple is a computer and the other apple is a fruit can be amazingly difficult. Let’s not even get into orange as a fruit versus orange as a color.
People are also easily able to form relations and associations between words and phrases, something that is difficult to replicate with computer code. For example, if I say the word “sneakers” to you, you will easily be able to come up with all kinds of other words and phrases. You might think, running shoes, tennis shoes, athletic, sports, or even brand names like Nike or Adidas. If you were looking through a catalog to find sneakers, you might stop on any of those words, because you know they are relevant to what you’re looking for. If you tell a computer to look for “sneakers” it’s going to look for sneakers and just whiz right past all those other words.
Up until now Google has had to rely on human intervention to build these relations and interpret subtle differences in meanings says Google Marketing Expert Qamar Zaman. Real people need to sit and review searches and tell the computer how results could be improved for the next time. That approach might seem to work well, until you consider the fact that Google handles more than 3 billion searches every day. On top of that, they’ve said that about 15% of daily searches are actually search queries they’ve never seen before. That might seem small, but it comes out to about 450 million never-before-seen search queries in a single day. Obviously, trying to keep up with that kind of volume and tune things “manually” is going to be a losing battle.
Rise Of The Machine
So this is where RankBrain comes in. The idea is that this system will be able to learn from previous search queries and make adjustments to results faster and more accurately. Overall, it shouldn’t really affect any page’s rank for any relevant terms. It seems to be working more on making sure search terms are understood correctly, rather than directly deciding what position a particular result should be given in relation to others.
Going back to our “apple” example, it’s not going to matter much to a computer reseller if they aren’t showing up on the results when someone is searching for pie recipes. The person running that search wasn’t going to by a computer anyway, even if he did use the word apple. This is the type of effect that RankBrain should have on results. Hopefully this new addition to the results algorithm will prove to be a benefit to both end users and site owners.
At the very least, it will be interesting to see how results change over time as RankBrain learns more about search — and we learn more about RankBrain.