Where is Google going with Hummingbird?
Google has a better understanding of search queries than ever.
…it’s now time to show that you have a better understanding of your buyers.
The Hummingbird algorithm returns results that are more likely to represent what a user is looking for when querying Google; it’s now up to SEOs to determine what users are looking for too – not just what Google wants to rank.
What is Hummingbird?
Additions like the knowledge graph demonstrate how Google has improved its understanding of search queries in recent years, and Hummingbird is a significant step towards integrating that understanding into the organic search results.
Back in 2010 – only a month after the Caffeine algorithm was rolled out – Google purchased a San Francisco-based data company called Metaweb, stating that
[People are asking] hard questions, and we’ve acquired Metaweb because we believe working together we’ll be able to provide better answers.
Fast forward to September 2013, and a statement from Google’s Amit Singhal:
We now get that the words in the search box are real world people, places and things, and not just strings to be managed on a web page.
These real world people, places and things are called entities, and with the help of Freebase – Metaweb’s collaborative knowledge bank – Google is getting better at classifying these entities.
The move towards entity-based search means that when a user queries Google, the results returned are the best pages in relation to the answer, rather than a page that answers the question directly.
How does Google understand entities?
The principle shouldn’t be unfamiliar to SEOs.
Keywords help Google to determine the relevance of a page. When multiple sites use a certain anchor text to link to a piece of content, Google is given a good idea what that content is about.
Google understood that a lot of websites stating a certain page was about a certain keyword meant that the page would be a good quality result for people who search for that keyword.
Instead of using links to determine what content is about, Hummingbird uses clauses across the web to determine what a subject is about. This subject is an entity, and the clauses are what you learned about in school: a simple sentence structure of [subject] [verb] [object], also called a triple.
Metaweb compiles these clauses/triples, which means that Google has access to an enormous database of them, as well as mining those it finds across the web.
If a person, place or thing is often mentioned in relation to something else, Google is able to better understand which pages might be relevant to a search term.
- Richard Branson owns Virgin Media
- Virgin Media is part of the Virgin Group
- Virgin Media has stores in the UK
- Virgin Media is boosting its network coverage
- You can invest in Virgin Media
- David Tennant advertises Virgin Media
When a user searches for ‘Virgin’, Google can now determine that users are likely to be looking for the company. Helpfully, Google even tells us what searches it considers similar to “Virgin Media” at the foot of the page.
Users are often looking for ways to get in touch with Virgin Media (see “virgin media contact number” and “virgin media customer service”) so some results are returned that allow users to do that; numbershelpline.co.uk, Facebook and Twitter.
Likewise, users may be after Virgin Media’s stores, so it displays locations using the Knowledge Graph and Google Places.
Users might be looking for more information about the brand…hence the News stories and Wikipedia entry, as well as the Knowledge Graph.
The most relevant results, though, are the company’s home pages, which is why these are displayed first.
Hummingbird and meme search
Entities are essentially memes.
That is, memes in the original sense of the word coined by biologist Richard Dawkins: a person, place, thing or idea whose relevance is determined by public perception of its meaning. If a person, place, thing or idea is widely mentioned as part of a certain statement, Google will begin to associate that entity with the statement.
Once an entity exists within the public domain (and therefore Google), the more often it is mentioned as part of different sentences, the more easily Google will determine what the entity refers to. This means that as news breaks and old entities are mentioned in new contexts, Hummingbird ensures that Google is displaying the results that are relevant to what people are actually talking about in relation to the search query.
When Richard Branson goes into space, Hummingbird means fresh results are returned that users will most likely want to see when they search for “Richard Branson” or even “Virgin”.
Meme search is why Google highlights Symantec.com when you enter “Norton antivirus” into the search box. Symantec is mentioned in connection with Norton antivirus so often that it’s essentially understood to be a brand search. Google understands these terms to be one and the same.
Where is Google going with Hummingbird?
Like Caffeine, Hummingbird is an infrastructure update – the new algorithm allows Google to incorporate new technologies; much like Caffeine paved the way for Panda and Penguin, Hummingbird will enable new updates that will improve the quality of search results and weed out the spam.
Potentially this could mean new and harsher penalties against poor quality sites, but Hummingbird itself is not a penalty. The aim is not to determine quality, but relevance – QA comes later.
Social signals are also becoming a more important ranking factor.
This doesn’t mean that +1s are a ranking factor, or tweets or Likes; but how people actually talk about entities across the web – on social channels; on blogs; in comment strings…
Bad reviews may become harder to escape, and PR will become even more important for brands; if people generally talk about your brand negatively, negative results will be returned for your search queries. Semantic mark-up on your site is as important as ever, but glowing reviews on your site will only get you so far now Hummingbird allows Google to see the bigger picture. Public perception of the service you give is taken into consideration.
It may also be that Google will continue attempting to assign more “weight” to the opinions of Authors, but with the Authorship Program apparently scuppered, it’s unclear whether a form of Author Rank will ever find its way into the way Google might understand entities. It does, however impact how Google interprets the quality of content.
The more things change…
Hummingbird does not change how Google determines the quality of content; only the relevance. In fact, content that precisely satisfies users’ search queries (think Q&A sites) will be returned in the SERPs less often.
Whether Google uses ‘quality authors’ to determine the relevance of content to a search query in future is irrelevant; Authorship is just as important now in demonstrating a page’s quality, and a reduced reliance on keywords means that it’s even more essential to ensure content is written well.
Hummingbird simply means that it’s more important to comprehensively answer users’ questions, and not just match users’ search terms. Google purchased Metaweb in order to provide better answers to hard questions. Hummingbird does not grade the quality of the answer – it allows Google to understand what users are actually asking for.