On September 17, 2013, Google held an event for their 15th birthday party, and part of that event they announced some updates, including the existence of a new platform that they call Hummingbird. There was a lot of confusion about this initially, as their own post about the 15th birthday updates did not even mention Hummingbird.
Initially people (including me) believed that Hummingbird was primarily about natural language search capability, but it is much more than that. Here are the major points about this new platform:
- Hummingbird is a complete rewrite of the Google Search algorithm.
- It was live for about 3 months prior to Google revealing its existence, and almost no one noticed.
- It does bring natural language queries into web search, so that is a part of it, but only a part!
- It enables Google to begin using new types of ranking signals, such as Social signals and/or AuthorRank.
- Apparently, it also allows them to potentially do different things with link signals than they have in the past.
Let's take a closer look at what Humingbird is, and how it will impact digital marketing!
Hummingbird is a complete rewrite of the Google Search Engine
Wow! Those are big words. But that is in fact what Hummingbird represents. However, it is important to clarify that Google's search service actually involves 3 major components as shown here:
It was Third Door Media's Danny Sullivan that clarified this for me. Sullivan was at the 15th birthday party, and I had the opportunity to have him on The Digital Marketing Excellence Show (this is a live broadcast event for which I am the host) to discuss Hummingbird on October 31st. The show uses Google Plus' Hangout on Air technology, and in the one I did with Sullivan he discussed the makeup of Hummingbird. Here is a 4:09 video extraction of some of Sullivan's comments on Hummingbird:
This notion that Hummingbird is a rewrite of the search engine is incredibly important. However, given that few people noticed that it had happened could make you question why it is a big deal. By design, Hummingbird was intended to provide Google with new capabilities within their search platform without disturbing the current ranking landscape that much. The reality is that the significance of this rewrite will unfold over the next few years.
A platform rewrite of this scope is only done when the prior platform has become a limit on the capabilities of the product. Google's original search engine has been around for a long time. They have been able to bolt on algorithms such as Panda, Penguin, and many others, but there were fundamental limitations on its capabilities.
So now Google had a new platform, and no one noticed until they told us it happened. The upshot of all this? Google had a new search engine that has many new capabilities. Hold on tight - a whole new ride is about to begin.
Natural language queries
This is the thing that everyone talked about initially with Hummingbird. This capability was first made live by Google long before Hummingbird (in May of 2013), but it's application was limited to the Knowledge Graph. An example sequence of searches demonstrating this capability is as follows:
and then follow that query with this one:
Note that in the 2nd query that Google recognizes that the word "he" refers to Tom Brady, because this query follows the query "show me pictures of Tom Brady". By itself, this is a pretty big development. In the case of this example, the answer is drawn from the Knowledge Graph. The Knowledge Graph is actually an entirely new component to consider in the overall Google product that I did not show in my prior diagram. A revised review of the Google service looks like this:
As with Tom Brady's height, some of the queries Google gets from users can be answered directly by the Knowledge Graph. It is this part of Google that started being able to handle natural language queries in May of 2013.
Another component of this is the notion of entities. A person, such as Tom Brady, is an entity, as is his team the New England Patriots. In addition, there is an increasing understanding of concepts related to entities (Tom Brady is a person), attributes, and relationships (he plays for the New England Patriots, which is an entity that has many players on it, he has a wide, he has a certain height, etc.).
All of this was part of the Knowledge Graph as early as May of 2013. What Hummingbird did was allow non-Knowledge Graph queries to benefit from the concept of entities and natural language queries. An example of this is to follow our query sequence above with yet one more query:
In this example, the result is not pulled from the Knowledge Graph, but Google does return a first result that answers the question asked. You can see a more extensive set of conversational queries that are supported by Hummingbird in this presentation.
New Types of ranking signals
One of the other big areas that Hummingbird enabled was the ability to process "social proof" signals differently. This includes links, by the way, as they are just one form of social proof. The bigger news though is that it also includes the ability to do more with social media related signals. Here is what Danny Sullivan had to say about it in this 5 minute video extracted from the Hangout on Air that I did with him:
Based on what we hear in the video we learn 2 things:
- Google has historically not had the infrastructure to process social signals as a ranking factor.
- They do now.
That is a big deal. While it is not clear when they will begin doing so, the long awaited comformation that social signals, and in particular Google+ signals, can impact search may be coming in the near future. As Sullivan says in the short video nugget provided above, this is long overdue. Social behavior is a way that people that don't own web sites can vote on the best content on the web. You can read more about this in the article I published on Search Engine Land: What Everybody Missed About Hummingbird: Social Signals.
Important caution though, don't go buy a whole bunch of +1s to your content. That will not accomplish anything. Google will weight the signals based on the authority of the person implementing the +1, and in particular their authority on the relevant topic area. Those purchased +1s will do nothing at all for you, except lighten your wallet a bit.
What does make sense is to implement a robust strategy for building up your real world strategy across social media platforms, and if it is relevant to you, include Google Plus as a part of that.
It may also be that Hummingbird enables AuthorRank. If you are not familiar with AuthorRank, this is the concept that Google might track the quality of your content and how well it is received in the market.
There may also be other signals that we have not even conceived of as yet that Hummingbird may enable.
My Hummingbird Predictions - Summing Them Up
1. Social signals will start to matter but no one will notice at first: Google will start with some very basic tests and ease them into the system slowly. They are going to go for very simple and obvious wins that impact a very small percentage of search results. They are not in a hurry with this. Timeframe for this to start? 2014
2. Google Plus will be the first social network to provide ranking inputs: Yes, I know half the SEOs in the world believe Google+ is a ranking factor already, and it is in certain limited ways, but a study I led earlier this year showed that Google Plus does not impact non-personalized results.
Google will try some experiments with Google Plus impacting non-personalized ranking first simply because they get the data in real time, without having to crawl anything. Crawling Facebook and Twitter in real time is literally impossible for Google to do, a fact that many don't fully appreciate, but it is the truth of the matter. Simple access to the data will make Google Plus the first social network to influence non-personalized search results, and it may begin to do so in some small ways in 2014.
3. Google will not punish people not on Google Plus: They can't. Google Plus is growing, but it is not the world, and they can't make their search engine product worse by making use of those signals. They need to use them to make it better, and they will have to find unambiguous signals that they can extract from the platform, and this is another reason for prediciton #1 above.
4. AuthorRank will finally begin to happen: They launched the rel=author initiative to help with that, but it is important to understand that they are tracking every Authorship signal they can, whether or not the content creator uses rel=author. They intend to figure out who writes the best, most in-demand, content and rank that higher, and Hummingbird enables that. Timeframe for this to start? I bet it starts in 2014.
5. Natural Language Search will continue to expand, but it will take a long time: This is not going to happen overnight! Teaching a machine natural language is hard. But, Google is fully committed to this process. They believe that mobile devices will drive increased usage of voice search, and that voice search will increase the usage of natural language queries. They intend to rule that space! Timeframe for this? We will see an update in 2014, but it will continue to evolve for at least a decade.
6. We will see a major link algo shift in 2014: This may be a major improvement in how they handle bad links, or it may be some other type of shift. We know from the interview I did with Danny Sullivan that changes in their capabilities for processing link signals was a part of Hummingbird. Links are still the big dog in SEO, so there is a lot for Google to gain with updates in this area, so I am predicting that we are going to get one!
7. Hummingbird will bring us at least 2 shockers in the next 2 years: Lucky number 7 is my wildcard prediction. My confidence that we are going to be handed some surprises is high though. This rewrite is meant to be their platform for the next 5 or more years, and they did it to alter their basic capabilities.
Final thoughts - Impact on publishers
Tactical Thoughts: There are many who will tell you that the long tail of search is now dead, that the combination of natural language search and 100% secure search has killed it. This is not true. The long tail is alive and well, but its character will be changed.
At a tactical level, what this means is that you need to get more focused on the benefits your products and services bring to customers, and the needs that they have that you can serve. Then you need to build pages around those benefits and needs. Google wants to better understand the user's intent, and so should you.
In addition, I believe it should be a higher priority than ever to clean up any bad links you may have. Don't wait for the penalty to hit. There is much more coming from Google in this arena. Clean house now, and then enjoy the benefits you get when competitors fail to do the same.
Strategic thoughts: If you have not already gone "holistic" with your digital marketing strategy, now is the time. Google has long maintained that its objective is to recognize real world authority signals and rank those that address user needs the best higher than other sites. Their push in this direction has been relentless, and it will continue.
It is going to get harder and harder to figure out some wrinkle in their algo to game. This will roll out in many stages, and over several years, but building real world authority does not happen over night. Get on your game now, get focused, and get a head start on those who hesitate. And, if you are already focusing your energy on real world authority, good for you, the best is yet to come!