On October 3rd, 2013 Google announced a major search algorithm release called Hummingbird.
Uh-oh.
Does this mean your content-driven business is in jeopardy? Is keyword researchdead? Are you going to have to reengineer your entire content strategy?
There’s no question that the Hummingbird algorithm is only the beginning of change in search optimization, but smart content creators can be prepared to thrive in this — and any — environment that may come in the future.
This release is basically a platform that enables Google to better handle “conversational” search queries.
To illustrate this, consider the difference between these two queries:
- “golden gate pictures”
- “give me some pictures of the golden gate bridge”
The first query is formed the way people have learned to enter entries using a keyboard. This has been our primary input method since web search was born.
Keyboards are not natural human devices, and even for fast typists they are a bit of an awkward device to use, so learning to abbreviate queries to talk to a search engine is a generally accepted practice.
However, the rise of mobile device usage brings some new challenges.
The mobile keyboard cometh
While many continue to type with the keyboards on phones and tablets, they are a bit more awkward to use.
Over time, people are going to increasingly gravitate to voice search in environments where that is acceptable (e.g. environments where speaking to your device is not seen as intrusive).
Voice queries are far more likely to fall into the pattern of the second query above — natural language queries.
As in all things search, Google wants to dominate mobile search too.
Google wants to process “real” speech patterns
Having the best platform for processing conversational queries is an important part of that, and that’s where Hummingbird fits in, though it’s just the beginning of a long process.
Think of Google’s Hummingbird algorithm as a two-year-old child. So far it’s learned a few very basic concepts.
These concepts represent building blocks, and it is now possible to teach it even more concepts going forward. It appears that a lot of this learning is derived from the rich array of information that Google has on all search queries done on the web, including the query sequences.
The Knowledge Graph
These examples involve Google’s Knowledge Graph, where natural language search benefits from the ability to pull real-time answers to queries that understand the specific context of the query.
Note that the Knowledge Graph has accepted some forms of conversational queries for a while, but a big part of Hummingbird was about expanding this capability to the rest of Google search.
I have seen people argue about whether or not Hummingbird was just a front end translator for search queries, or whether it is really about understanding more complex types of user intent.
The practical examples we have now may behave more like the former, but make no mistake that Google wants to be able to do the latter as well.
The mind reading algorithm
Google wants to understand what is on your mind, well, before its on your mind.
Consider Google Now as ultimately being part of this mix. Imagine being able to have Google address search queries like these:
- Where do I find someone that can install my surround sound system?
- What year did the Sox lose that one game playoff?
- What are the predictions for the price of gas next summer?
- What time is my dinner on Tuesday night, where is it, and how do I get there?
No, these queries will not work right now, but it gives you some idea of where this is all headed.
These all require quite a bit of semantic analysis, as well as pulling in additional information including your personal context.
The 4th question I added was to show that Google is not likely to care if the search is happening across web sites, in your address book, or both. Not all of this is Hummingbird, per se, but it is all part of the larger landscape.
To give you an idea on how long this has taken to build, Google’s Amit Singhal first filed a patent called Search queries improved based on query semantic information in March of 2003. In short, development of this technology has taken a very long time, and is a very big deal.
The implications of a Hummingbird search world
It is important to remember that this step forward being described by Google as a new platform.
Like the Caffeine release Google did in June of 2010, the real import of this is yet to come. Google will be able to implement many more capabilities in the future. The implications to search in the long term are potentially huge.
For you as a publisher, the implications are more straightforward. Here are a few things to think about:
1. Will keywords go away?
Not entirely. The language you use is a key part of a semantic analysis of your content.
Hopefully, you abandoned the idea of using the same phrases over and over again in your content a long time ago. It will remain wise to have a straightforward definition of what the page is about in the page title.
I’ll elaborate a bit more on this in point 3 below.
2. Will Google make the long tail of search go away?
Not really. Some of the aspects that trigger long tail type search results may actually be inferred by Google rather than contained in the query. Or they may be in the user’s query itself. Some long tail user queries may also get distilled down to a simpler head term.
There will definitely be shifts here, but the exact path this will take is hard to project. In the long term though, the long tail will be defined by long tail human desires and needs, not keyword strings.
The language you use still matters, because it helps you communicate to users and Google what needs and desires you answer.
3. You need to understand your prospect’s possible intents
That is what Google is trying to do. They are trying to understand the human need, and provide that person with what they need.
Over time, users will be retrained to avoid short simple keyword-ese type queries and just say what they want. Note that this evolution is not likely to be rapid, as Google still has a long way to go still!
As a publisher, you should focus more attention on building pages for each of the different basic needs and intentions of the potential customers for your products and services. Start mapping those needs and use cases and design your site’s architecture, content, and use of language to address those.
In other words, know your audience. Doing this really well takes work, but it starts with knowing your potential customers or clients and why they might buy what you have to sell, and identifying the information they need first.
4. Semantic relevance is the new king
We used to speak about content being king, and that in some sense is still true, but it is becoming more complex than that now.
You now need to think about content that truly addresses specific wants and needs. Does your content communicate relevance to a specific want or need?
In addition, you can’t overlook the need to communicate your overall authority in a specific topic area. Do you answer the need better than anyone else?
While much of being seen as an authority involves other signals such as links, and perhaps some weight related to social shares and interaction, it also involvescreating in-depth content that does more than scratch the surface of a need.
Are you more in-depth than anyone else? If someone has some very specific scenarios for using your product or service, does your content communicate that you address it? Does your content really stand out in some way?
What’s it to you?
As noted above, this is going to be a journey for all of us.
While Google’s eventual destination is easy to imagine (think Star Trek’s on board computer), Hummingbird has only scratched the surface, and the steps along the way are hard to predict. That will be driven by very specific developments in technology.
No comments:
Post a Comment