I was moderating a roundtable session at SES Chicago last week when the conversation turned to Hummingbird and how – according to the young lady I spoke to – it effectively means “Google is simplifying the query from long tail to shorter terms.”
I’ve heard this point of view a few times over the past few weeks, and fundamentally disagree.
To (hopefully) put a different lens on the simplification argument, I’m laying out a few concepts for discussion. “Simple Queryists” please feel free to slice, dice, dissect, and discuss below.
We All Know Larry Bird
Do a search for [Where can I buy a Larry Bird shirt]. Simple Queryists think that Google will shorten the query to something like “Buy Larry Bird Shirt” and match a result accordingly.
I don’t believe this is what’s happening. Rather, Hummingbird is adding a layer of understanding to the query that acts more like an expansion of the query, so that its meaning is clearer.
In this example, I’m thinking Google would interpret the query as the following:
Where (Place: User is located) can I buy (Intent: Purchase) a Larry Bird (Person: basketball player) shirt (Product: [via Association of Product to Player to Team] Boston Celtics shirt #33)
Not exactly shorter.
Let’s break that down a little further, most importantly to see the implied connection between person and product.
With better meaning and context of understanding, I believe Google is connecting, via their Knowledge Graph, Larry Bird the basketball player with the context of the team he’s most famous for, and the product that matches the shirt number he wore. The query is then more exact, and should return a more exact and relevant result. By using prior search data along with “big data,” – other user behavior factors – the new more intelligent algorithm can predict the best match of content to a user’s intent.
And that’s a fundamental difference. Intelligence through experience and predictability is applied to improve results. And, if it’s not exactly a match, real-time user behavior – clicks, query modification, dwell time and page interactions – can help finesse future results.
Big Bird Isn’t a Large Bird
Do a search for [How old is big bird] and you’re treated to a complete history of the Muppet character’s creator (including his birthday).
The term “Big Bird” exists in Google’s Knowledge Graph as a known personality entity, so rather than looking for facts on large winged animals; Google intelligently associates a query mentioning “Big Bird” as a request for information on the large yellow Muppet.
The potential reasoning is that the query is expanded to include the recognition of certain query elements as a distinct entity, so that the query is seen as follows:
How old (Attribute: Age) is Big Bird (Character: Muppet [Attribute: Creator] Person: Caroll Spinney)
Google even serves up a box with the actual age of the puppeteer, delivering additional, useful information based on specific attributes (Age) in the data associating Big Bird with Carroll Spinney and with the query string “how old.”
Both these examples illustrate Hummingbird’s better understanding of query meaning by recognizing and then connecting entities and attributes contained within search queries to existing knowledge data, then applying context to each query – the who, where, when, what and how of intent:
- Who is searching – based on personalization and prior behavior.
- Where they’re searching from – location and locality of query.
- When the search is happening – time of day, seasonality, and dayparting.
- What is the search being conducted on – desktop, mobile, tablet, or emerging platforms.
- How the search query is formed – conversational aspects of the user’s question.
SEO Isn’t for the Birds
There are many discussions I’ve read or have been a part of since the Hummingbird announcement. Ultimately, the SEO community didn’t really notice the original launch (one month before the announcement), because it didn’t really affect site rankings or search engine traffic of existing brands.
But that will surely change.
As search engine query intelligence improves – through entity recognition, entity authority, attribute recognition, and intra-entity connections – expect more relevant and diverse results, as search engines seek to promote not just the better single result but also the better associated results.
What are SEO folks to do? No panic necessary.
- Create human-focused content that answers known queries and needs. (Consider meaning, medium, and message.)
- Identify entities, attributes, and connections in content (through schema.org and other markup) for search engines to digest.
- Ensure content is crawlable, indexable, and sharable.
- Seek, enhance, and inspire authority signals via social, PR, and good old marketing!