Google’s last infrastructure update, named Hummingbird, will be the focus of my latest post of the SEO101 series, sharing the basics of SEO as I learn them. I will be directing attention to what Hummingbird is, what we know of how it works and what this means to Google users and site owners. Hummingbird is a huge overhaul to Google search and is an important factor in understanding the evolution of search to what it is today.
What is Hummingbird and what changes does it bring?
Launched on 30th August 2013, Hummingbird was one of the largest shifts in the way Google works since the caffeine update in 2010. Google stated that Hummingbird would affect 90% of searches, however for such a large change it had a relatively small impact on websites rankings, especially compared with topic classifier algorithm changes (such as a Penguin and Panda). Simply put: Hummingbird could be described as the overall infrastructure (or meal) for Google search; Penguin and Panda are topic classifiers (or ingredients) within this, as well as over 200 other ranking factors (other ingredients) which all help to determine the ranks of pages within Google.
So what has changed since Hummingbird was released? To put it straightforwardly, Google can now understand the intent and meaning of searches better, therefore offering more contextual results to users. As Amit Singhal, head of Google’s core ranking team described it, “People communicate with each other by conversation, not by typing keywords”, and therefore the Hummingbird update tries to serve these needs. The way people search is constantly changing, especially with the rise of searches on mobile devices and by voice; therefore Hummingbird is Google’s evolution to deal with these changes and improve the finesse with which they can return results, in particular for long tail queries.
How does Hummingbird work?
So how does Hummingbird provide us with more contextual results? Google has integrated many of its projects into Hummingbird, such as its knowledge graph.
Google Now, a synonym engine, a query revision engine and other projects, which all work together to provide more context and relevance to the results users see. This Google patent gives us a deeper insight into how Hummingbird works by describing the synonym engine and the query revision engine in more detail.
The query revision engine adds to or changes the search query after searching, but before retrieving documents from their index meaning users will not notice a difference when they type in a query. This in theory will reduce the amount of SERP’s that are produced (if search query 1, 2 and 3 essentially mean the same, then only 1 SERP needs to be presented to users).
So to simplify some of Google’s technical jargon in their patent, it’s easier to use an example. A user types “where is the best place to eat a pizza in London”. With Hummingbird, Google can look for synonyms of the word “place”, which could be a “restaurant” or “finish” for example, and will then use a confidence score to work out which word is more relevant to the query. In this case, Google will work out that when the word pizza is near the word “place” in the query, the term restaurant has a higher confidence score and will then replace “place” with “restaurant”.
What does this mean for Google users and websites?
So what does that mean for searchers and site owners? Hummingbird presents the ability to answer questions such as “how tall is the Eiffel tower?” as shown below:
If you are a sports fan, try typing in your favourite team to see what kind of results Google presents to you (like the screenshot below of my beloved Aston Villa). Type in questions, and Google will now answer them, and give you comparisons. Try typing in “do a barrel roll”; Google employees have a good sense of humour!
Previously, Google ignored what it deemed to be “irrelevant” keywords from queries, however it has now introduced full sentence indexing to better understand the context of a query. This means that Google has the ability to understand long-tail keywords in more detail, which is particularly important with the shift to people using voice search and mobile which is generally more conversational in nature. Site owners with real content now benefit more than ever before in terms of rankings as these ‘stop words’ are now acknowledged, in theory pushing well written content higher up the SERP’s.
Points to Remember
In essence, Hummingbird is about matching pages by meaning, rather than keywords. Here are the 3 main points to remember:
- Hummingbird is not an algorithm, it’s a primary infrastructure. This isn’t a topic classifier update like Panda or Penguin, this is a fundamental change to Google’s search framework.
- Google has started indexing full sentences, to further understand the context of a search term, instead of stripping out what it deems to be “irrelevant” words. This means real content will be recognised more by Google, echoing the “content is king” phrase in the SEO industry.
- A query revision engine has been added that revises or adds to your search term after submitting to Google but before retrieving documents from the index. This, integrated with Google’s knowledge graph, synonym engine and other processes, ensures the results users receive are contextualised more than ever.