Those of us who work in SEO tend to have the same overreaction to any new Google announcement; anxiety, constant worrying, mistrust, mood swings…perhaps I’m embellishing somewhat but, typically, website owners greet Google announcements with a certain degree of suspicion and many already believe they’ve been hit by Hummingbird.
Of course, the fact that Google isn’t being particularly transparent is only adding fuel to the fire; they have referred to Hummingbird in none of their documents (such as the recent article of their 15th anniversary on their official blog) and avoided the specifics of the update in unofficial statements made by Amit Singhal. Indeed, Singhal’s announcements have added to the uncertainty surrounding Hummingbird.
When revisions are put in place by Google, particularly in the case of significant updates such as Hummingbird, the best course of action is to steer clear of any attempts to fully comprehend its effects right away; attempting to understand Hummingbird now is based solely on suspicions. The most beneficial thing you can do is to avoid taking action until things have calmed down and we have more information about what Hummingbird is. Examine the initial records, associated documents and variations and don’t get impatient; go slowly and observe how the update is working, look into it and only once you have done all this, attempt to discover the most likely answer.
It’s an unscientific technique and as a result, the answers cannot be guaranteed as 100% accurate. However, when dealing with Google updates, I think it’s a brilliant technique to apply.
The initial records in this case are the accounts of the media event in which Google revealed Hummingbird and the article produced by Danny Sullivan following the aforementioned event, which directly mentions Singhal’s statements.
Associated documents would be the patents that are likely at the root of Hummingbird, as well as interpretations and comments made by industry experts.
This article will cover the outcome of what I have learned from these documents.
Why compare Hummingbird to Caffeine?
In Amit Singhal’s Hummingbird statement he mentioned that Google’s algorithm hadn’t been restructured this significantly since 2010 and the implementation of Caffeine.
The issue here is that Caffeine didn’t modify the algorithm, but the infrastructure.
The intention of Caffeine was actually to enhance the way Google indexes the huge number of online documents that it crawls, producing a new and greater selection of results for users.
On the other hand, the intention of Hummingbird is not to improve the indexing procedure; rather, it aims to provide improved comprehension of the goal behind a search term, thus providing users with results that are more appropriate to them.
Still, we can establish that Hummingbird revises Google infrastructure, seeing that it manages over 200 elements which comprise Google’s algorithm.
Although he himself was perhaps unaware of it, the connection that Singhal made between Hummingbird and Caffeine shows us:
- That this new update wouldn’t exist if Caffeine had not been set up three years ago, and it should therefore be seen as a progression for Google and not a complete upheaval.
- Furthermore, Hummingbird ought to be seen as Google’s most earnest effort at resolving the algorithm problems that originated with Caffeine.
Allow me to clarify that last point. With Caffeine, result pages were overloaded with substandard results. In response, Google produced smaller updates such as Penguin, Panda and the EMD update, to name a few.
However, while these revisions were efficient in regards to head- and middle-tail search terms, they were less useful for long-tail and conversational terms (or “verbose queries” as labelled by Singhal), which have grown ever more common, largely due to users’ quick embracement of mobile search.
The advancement of Google’s organic language recognition, the enhanced capability in determining entities and ideas using Metaweb technology and enhanced through Knowledge Graph, as well as the colossal progress made in result page customisation, have provided Google with the abstract and practical equipment not just for resolving long-tail search term issues, but furthermore providing a new beginning to Google’s advancement.
When speaking of Hummingbird, Amit Singhal stated that it had provided an “opportunity” to look at existing tools, already implemented by Google in an attempt to comprehend search term definitions, and re-evaluate how they could be brought together to deduce how search terms and documents could be matched, regarding the goal of the term and the associations the documents offered, rather than simply “random coincidence” which can occur with “early search engines”.
How exactly does Hummingbird function?
One of the tools already implemented by Google that Amit Singhal specifically mentioned in his statements was synonyms.
Google has used synonyms for quite some time. Observing the timeline posted by Google for the 15th anniversary, it tells us that they’ve been utilized for 11 years now. In addition, we can see that disambiguation (intended to be a linguistic breakdown of search terms) has been functioning for 12 years.
In 2012, we saw an article from Vanessa Fox in which she examined Google’s development of synonym matching.
Seeing that article and the cases it offered, it’s obvious that Google had already implemented synonyms, in relation to the objectives behind the search term, to expand the term and provide more fitting results.
However, the article also demonstrates that utilizing a synonyms list or depending on data based on popular terms wasn’t sufficient to guarantee applicable results; for example, the article notes that the term “pet adoption” will supplement “pet” for “cats”, but not “dogs”.
As seen in a previous patent, Amit Singhal was aware that depending solely on synonyms wasn’t an ideal answer, since some words can be synonymous in one situation and yet mean completely different things in another.
For this reason, to provide the most fitting results with semantic search, Google had to recognise context in a way that was quicker, simpler and more appropriate. Hummingbird is Google’s way of dealing with that necessity.
Synonyms are still vital; this was verified by Singhal in his discussion with Danny Sullivan. The way they are being implemented currently was explained in an article by Bill Slawski, in which he analysed another of Google’s synonym patents which looks at synonym recognition derived from terms that appear together.
To put it plainly, words are the spoken depiction of things, and not “things” in their own right, and Google uses search entities to turn words into ideas. An item might have a connection to another item that may vary in other circumstances in which they are used together. By this theory, words are considered in the same way that people and other named things are.
The system utilized by Google when recognising search entities is particularly significant when disambiguating the diverse possible implications of a certain word, and thus improving the data retrieval in line with a “probability score.” This method isn’t dissimilar to Knowledge Graph’s disambiguation method.
Lastly, another model is making a clear contribution to what may possibly be the patent for Hummingbird; co-occurring terms.
Incorporating these components, Google is now hypothetically capable of:
- Improved term objective comprehension.
- Expanding the group of online texts that could possibly satisfy a query.
- Streamlining the way Google dispenses data; if several queries have identical intent, it can produce one set of results and several varied result pages are not required.
- Providing an improved experience when searching, by broadening the term and improving comprehension of the connections between different search entities (additionally including personalised factors), Google is capable of providing results with a better chance of fulfilling the user’s requirements.
- Consequently, Google could also improve result page advertisements as, prior to Hummingbird, the majority of verbose queries didn’t show advertisements in result pages.
Does Hummingbird really impact 90% of search queries?
Numerous SEO professionals have queried the statement that Hummingbird has had an impact on 90% of search terms, purely because they haven’t observed any variations in rank or traffic.
Despite the reality that result pages were chaotic between late August and early September, when Hummingbird was first implemented (although this may be coincidental, it would be a rather timely fluke), the average search term that Hummingbird focuses on is conversational (for example, “Where can I find a good Japanese restaurant near Argyle Street in Glasgow?”), queries which those of us in SEO do not normally follow.
Additionally, Hummingbird focuses on search terms, rather than keywords (and certainly not long-tail keywords), which was thoroughly described in a recent article from Ammon Johns. Consequently, following rankings for long-tail keywords to measure Hummingbird’s effects is completely useless.
Lastly, Hummingbird hasn’t put an end to the traditional ranking components; rather, it is an outline that has been built upon these traditional parts. If a page is authoritative and appropriate for the search term, it will continue to rank exactly as it did prior to the release of Hummingbird.
So, which websites were penalized? In all probability, it was websites which depended solely on optimizing for long-tail key terms, but were lacking in authority. As a result, it is currently substantially more fitting to produce content that is likely to be shared that additionally semantically connects to long-tail key terms, rather than produce countless long-tail keyword optimized pages that are unhelpful and low-value.
Will Schema.org improve my ranking, considering Hummingbird is a move towards semantic SEO?
A persistent, quickly-spreading rumour when we first heard of the update was that Hummingbird was considering structured data to be a major component.
While it’s accurate that Google has been underlining the value of structured data (going so for as to set aside a segment to structured data in their Webmaster Tools), Schema.org should not be thought of as some sort of magical fix. This is a case of muddling the method with the intention.
What we should be doing is providing Google with straightforward and clear context for our content; for this reason, structured data can be of use. Still, structured data alone is not sufficient. As previously noted, if a site has low authority (due to a lack of outside links and referrals), it probably won’t be good enough to rank satisfyingly high, particularly since Hummingbird has now streamlined long-tail key terms.
Is there a connection between Hummingbird and the rise in Answers Cards and Knowledge Graph?
Numerous individuals put forth the theory that Hummingbird converts Knowledge Graph to traditional search, and is related to the huge rise in Answer Cards. The idea influenced a number of heated tirades, annoyed by Google’s “scraper” attributes.
This is probably a result of the Hummingbird statement being made at the same time as updated Knowledge Graph elements; however, no clear connection is present between the two.
What numerous people have believed to be a result (Hummingbird resulting in increased Answer Cards, etc.) is likely a basic connection.
Hummingbird significantly streamlined verbose queries into easily understandable search terms, which are at times matched with the ever-growing Knowledge Graph. Consequently, an increased amount of result pages display Knowledge Graph components, as well as Answer Cards.
Still, the ideas behind Knowledge Graph and Hummingbird are very similar.
Is Hummingbird firmly derived from Knowledge Base?
Knowledge Base is powerful and omnipresent in Google’s functionality; however, to condense Hummingbird down to simply Knowledge Base is to over-simplify it.
As we have observed, Hummingbird depends on many factors, including Knowledge Base, particularly in personalised search terms (which ought to be seen as a pervasive factor impacting the algorithm).
If Hummingbird were to firmly depend on Knowledge Base, with no other elements to provide balance, we might come across problems such as those Amit Singhal described dealing with in previous synonym patents.
Is it true that Hummingbird’s release put an end to the link graph?
Not true. PageRank, as well as similar algorithm factors, remain active. It could even be said that links have gained significance.
Indeed, lacking the authority provided by a sufficient link profile, a site will find it harder to rank, now more than ever (take a look at the section above for more on the outcome of sites lacking in authority).
The key factor currently is the context behind the link. We observed this previously with Penguin, however, Hummingbird once again highlights that external links from off-topic, unrelated circumstances are not good.
However, it’s true that Google still has a way to go to develop link elements, as many have criticised this factor. Still, due to entity and context understanding, brand name co-citations and co-occurrences have a bigger impact with Hummingbird.
Is there a connection between Hummingbird and 100% (not provided)?
The two were introduced over the same period of time and that wouldn’t appear to be an accident.
Considering that Hummingbird is focused on search entities, improved data retrieval and the broadening of search terms (a modification where key terms alone are now less important), it makes sense that depending solely on key term information is no longer sufficient.
We ought to avoid solely targeting key term optimization and begin considering topical optimization.
This means we need to consider valuable content, rather than simply “content”. Such things as SEO copywriting will inevitably become synonymous with high-quality copywriting. As a result, we need to begin considering the functionality behind search entities, rather than just compiling synonyms.
If the Hummingbird update means a move towards semantic SEO, then our role in SEO is to improve our optimization not only for things, but for associations that connect things.
How should we optimize for Hummingbird?
Try to respond honestly to these questions:
- When you optimize content, do you do so with a clear target audience?
- When carrying out on-page optimization, do you adhere to these proven effective traditions?
- Implementing understandable data structure.
- Steering clear of problems with canonicalization.
- Circumventing problems with thin-content.
- Producing a semantic content model.
- Topical optimization of content, implementing organic, semantically rich wording and targeting a landing page-focused tactic.
- Publishing valuable content utilizing numerous formats, which you can honestly say is worth sharing.
- Making use of semantic mark-ups, Schema.org and Open Graph.
- Is your goal with link building:
- Improved brand awareness?
- Receiving referral traffic?
- Improving your brand’s perceived authority?
- Contextually associated websites or connected segments of broader websites?
- Is social networking providing these rewards?
- Broader brand recognition.
- Social echo.
- A growing number of links from external websites.
- Brand development and organic traffic.
If you responded positively to each question, you needn’t worry; simply keep doing what you’re doing, remembering to be engaging and original. You probably observed good rankings previously, as well as improved traffic due to your high quality SEO practices.
If you said no to some questions, you simply need to fix your mistakes and abide by these effective SEO practices.
If you truthfully said no to numerous questions, your issues pre-date Hummingbird and you won’t see any improvement unless you significantly change your outlook.
Hummingbird isn’t attempting to revolutionise SEO. It merely requests that we avoid poor quality SEO; shouldn’t we already know not to do this?