Danny Sullivan Shares What He Has Learned About Hummingbird

Some of the discussions in this Hangout on Air event were just priceless. It took place at 4 PM ET yesterday, and Danny shares some fantastic stuff on Hummingbird, social signals in search, and more. The following link will take you to my post on Google Plus about the event, and this contains the full video.

One part that I want to highlight in this post is the discussion on Hummingbird. Here is a short 4 minute and 10 second video extraction from the full show.

Hummingbird in a Nutshell

Danny gives us a great explanation of the 3 components of Google’s search service. Basically, this consists of 3 major components:

Hummingbird is a search engine rewrite

As it says, hummingbird was a rewrite of the search engine. Part of this was to enable Google to better understand “entities” within the context of web search. Google has had this as part of Knowledge Graph type queries for a while. This is also the reason that Google has had some capabilities in dealing with natural language queries for the Knowledge Graph, but now with Hummingbird update, that ability to understand natural language queries has been extended to the rest of the search engine as well.

The following chart represents my speculation as to how this may work:

Inside the Query Translator

What we do know for sure is that Google is able to better understand a number of scenarios, such as references to entities, comparisons, query sequences, and the like, much better than they have in the past. As Danny says in the video, Google now understands “Tom Brady” as a person, as opposed to seeing that as 2 words “Tom” and “Brady”. Who knows, this update may even begin to enable some use of AuthorRank.

But this is only part of the story. Hummingbird encompasses a lot more than just bringing natural language search and entity recognition into the search engine. According to Danny, it also lets them do a better job of bringing in new typs of signals such as social signals.

We discuss this at length in the full video, including the importance of leveraging social signals as a ranking factor. While I still see these as being treated differently than a web based link, I agree that use of these signals is something that is very important for Google to start doing.

As always, it is far easier for us to say they should use them than it is for them to figure out how to use them properly.

Lastly, it is important to recognize that there is likely to be much more in here than we now so far. As a result of the search engine it may well incorporate many new types of components that we are completely unaware of. When will we learn about these? Only time will tell.

You can see the full video of Danny Sullivan on The Digital Marketing Excellence Show here:

The Digital Marketing Excellence Show is airs every Thursday. To get notified about upcoming shows, please follow me on Google Plus.

Thanks to Scott Scowcroft for creating the above HOA nugget (aka “The Scott Treatment”), and Ronnie Bincer for his advice on all things HOA!

Comments

  1. says

    Hummingbird seems to be more about matching queries to “entities” that have been mined from a handful of trusted sources (primarily Wikipedia) and modifying queries on the basis of the user’s recent query history (and click history). I have been able to reduce Hummingbird’s influence on my search results by blocking Google’s click tracking and clearing my browser history (both cookies and cached files) so that query results are not badly skewed toward ridiculous biases based on previous query activity.

    The net effect on Hummignbird seems to be that the searcher is sent on a spiral toward results that closely match earlier topics. In an unfiltered Hummingbird query session (which can last for days) you’re less likely to find new content, especially if it’s not associated (internal to Google) with their “entities”.

    Most users are probably not sensitive to the adjustments being made to their queries by the system but I have noticed an uptick in complaints about Google’s search results by both sophisticated and unsophisicated users over the past few weeks.

    I really don’t like Hummingbird and if Google doesn’t give the user more control over these bias settings I think in the long run their strategy will backfire and alienate people because the search results are just inexplicable after a certain point.

  2. says

    I guess Hummingbird is partially an evolution to adapt the search engine to mobile devices queries. Think about vocal searches, they are quite different from “keyboard” searches. Hummingbird can clean this queries to find the intent and give a goog answer in the first try.

  3. says

    Hi ERIC,

    “Hummingbird” is Google’s new (latest version) Harley Davidson, where they can use all possible advanced and modern parts like query rewriting, social signal, author rank and many other signals (to be developed in future, according to needs).

    Personally I think – Social signal shouldn’t be given much weighted as a ranking factor. Great authors have good social reputation, but good social reputation doesn’t always translate that person in to a great author.

    As of now, Google is concentrating more on query part or “query rewriting” using Hummingbird instead of ranking of content on SERPs.

  4. says

    Actually I think social factors are preferable to pagerank. Pagerank has long been abused with until recently, even Google recommended webmasters get as many links out there as they can to increase their ranking.

Leave a Reply

Your email address will not be published. Required fields are marked *

*