Google Hummingbird was a rewrite of Google’s algorithm that consciously anticipated the needs of searching on mobile devices, in particular by enabling conversational search.

Hummingbird set the stage for dramatic advances in search.

Google never published an explainer of what Hummingbird was.

However, there are records of Googlers explaining what it is.

Let’s take a look at what Google’s Hummingbird update did, how it impacted natural language search, and what Googlers and SEO industry experts had to say about it.

Google Hummingbird

The Google Hummingbird update was put into place in August 2013 and announced one month later, in September 2013.

The Hummingbird update has been described by Google as the biggest change to the algorithm since 2001.

It was also described by multiple Googlers as a total rewrite of the core algorithm.

Yet, despite the scale of this update, the immediate effect was so subtle that the update was largely unnoticed.

It seems contradictory for an update to be both wide-scale and unnoticeable.

The contradiction, however, is made more understandable when Hummingbird is viewed as the starting point for subsequent waves of innovations that were made possible by it.

Hummingbird Defined

The update was called Hummingbird because it is said to make Google’s core algorithm more precise and fast.

We all know what fast means.

Arguably the most important part of Hummingbird is the word “precise” because precision is about accuracy and being exact.

As you’ll see in the following linked conversations by Googlers, Hummingbird enabled Google to be more precise about what a query meant.

And, by moving away from matching keywords in a query to keywords on a webpage, Google became more precise about showing pages that matched the topic inherent in the search query.

A Complete Rewrite Of The Core Algorithm

Former Google Software Engineer Matt Cutts described Hummingbird as a rewrite of the entire core algorithm.

That doesn’t mean it was a brand new algorithm but rather the core algorithm was rewritten in a way that makes it able to do its job better.

In a December 4, 2013 video interview, Matt Cutts said that the Hummingbird algorithm was a rewrite of Google’s core search algorithm.

Matt Cutts explained (at the 1:20:00 mark of this video):

“Hummingbird is a rewrite of the core search algorithm.

Just to do a better job of matching the users queries with documents, especially for natural language queries, you know the queries get longer, they have more words in them and sometimes those words matter and sometimes they don’t.”

Some people think of Hummingbird as a component of Google’s core algorithm, much like Panda and Penguin are parts of the core algorithm.

Matt Cutts makes it clear that Hummingbird was not a part of the core algorithm. It was a rewrite of the core algorithm.

One of the goals of the rewrite was to make the core algorithm better able to match queries to webpages and to be able to handle longer conversational search queries.

Hummingbird Affected 90% Of Searches

Matt Cutts followed up by sharing that the precision and quickness of Hummingbird were present in 90% of searches.

Matt said:

“And so Hummingbird affects 90% of all searches.

But usually just to a small degree because we’re saying this particular document isn’t really about what the user searched for because maybe they said, ‘Okay Google, now how do I put a rutabaga up into space, what really matters is rutabaga and space and not how do I’.”

Hummingbird And Natural Language Search

When Hummingbird came out, some in the search community advised that it might be a good idea to change how content is written in order to match how searchers were searching.

Common advice was to convert articles to use more phrases like, how to.

While the advice was well-intentioned, it was also misguided.

What Hummingbird did was to make long conversational search queries understandable to the search engine.

In Matt’s example, Google was ignoring certain words in order to better understand what the search query really meant.

In the old algorithm, Google would try to rank a webpage that contained all the words in a search query, to do a word-for-word match between the search query and the webpage.

What Matt was explaining is that Google was now ignoring certain words in order to understand the queries and then use that understanding to rank a webpage.

Hummingbird enabled Google to stop relying on matching keywords to webpages, and instead, focus more on what the search query means.

That’s what he meant when he started his explanation of Hummingbird by saying:

“Just to do a better job of matching the users queries with documents, especially for natural language queries…”

Is There A Hummingbird Patent?

Some of the things that Hummingbird was doing with search queries was rewriting them by using techniques like query expansion.

For example, there are multiple ways to search for the same thing, using different words.

Five different search queries can be equal to one search query, with the only difference being that they use different words that are synonyms of each other.

With something like query expansion, Google could use synonyms to broaden the group of potential webpages to rank.

After Hummingbird, Google was no longer exact matching keywords in search queries to keywords in webpages.

This was something different that began happening after the Hummingbird update.

Bill Slawski wrote about a patent that describes things that the Hummingbird algorithm is said to be able to do, especially with regard to natural language queries.

Bill writes in his article:

“When the Hummingbird patent came out on Google’s 15th Birthday, it was like an overhaul of Google’s infrastructure, such as the Caffeine update, in the way that Googles index worked.

One thing that we were told was that the process behind Hummingbird was to rewrite queries more intelligently.”

The patent that Bill discovered and wrote about describes a breakthrough in how search queries are handled.

This patent described a way to make a search engine perform better for natural language search queries.

Thanks to Matt Cutts, we know that Hummingbird was a total rewrite of Google’s search algorithm.

Thanks to Bill Slawski, we can read a patent that describes some of the new things that the Hummingbird update made possible.

Does The Hummingbird Update Do New Things?

Similar to what Bill Slawski touched on about the patent he discovered, Matt Cutts said that the Hummingbird update allows Google to remove words from a mobile search query.

Matt Cutts said at a Pubcon 2013 keynote session that Hummingbird allows the algorithm to remove words that aren’t relevant to the context of what a user wants to find from a mobile voice search query.

You can watch Matt discuss Google Hummingbird in this video at the 6:35 minute mark:

“…the idea behind Hummingbird is, if you’re doing a query, it might be a natural language query, and you might include some word that you don’t necessarily need, like uh… [what’s the capital of Texas my dear]?

Well, ‘my dear’ doesn’t really add anything to that query.

It would be totally fine if you said just, [what is the capital of Texas?]

Or, [what is the capital of ever lovin’ Texas?]

Or, [what is the capital of crazy rebel beautiful Texas?]

Some of those words don’t matter as much.

And previously, Google used to match just the words in the query.

Now, we’re starting to say which ones are actually more helpful and which ones are more important.

And so Hummingbird is a step in that direction, where if you are saying or typing a longer query then we’re going to figure out which words matter more…”

There are three key takeaways from Matt’s explanation of what Hummingbird does:

• Google no longer relies on just matching keywords in the search query.
• Google identifies which words in a query are important and which are not.
• Hummingbird is a step in the direction of understanding queries more precisely.

Hummingbird Did Not Initially Affect SEO

As previously mentioned, some SEOs advised updating webpages to make them match longer conversational search queries.

But just because Google was learning to understand conversational search queries did not mean that webpages needed to become more conversational.

In the above video recording of the 2013 Pubcon keynote address, Matt goes on to remark that Hummingbird doesn’t affect SEO.

Matt observed:

“Now, there’s a lot of articles written about Hummingbird, when even when just the code name was known, people were like, okay, how will Hummingbird affect SEO?

And even though people don’t know exactly what Hummingbird is they’re still going to write 500 words about how Hummingbird affects SEO.

And the fact is it doesn’t affect it that much.”

The Effect Of Hummingbird On Search Was Subtle

Matt next describes how the changes that Hummingbird introduced were subtle and not disruptive.

He said that the effect of the Hummingbird update was wide but the effect itself was small.

Matt explained:

“It affected 90% of queries but only to a small degree and we rolled it out over a month without people even noticing.

So it’s a subtle change, it’s not something that you need to worry about. It’s not going to rock your world like Panda and Penguin.

It’s just going to make the results a little bit better and especially on those long-tail queries or really specific queries, make them much better.”

Hummingbird & Long-Tail Keywords

Cutts continued his discussion about Hummingbird by describing its effect on sites that targeted extremely specific long-tail keywords.

We have to stop here and talk about long-tail phrases in order to better understand Matt Cutts is talking about because this part of the Hummingbird update had an effect on some SEO practices.

Long-tail keywords are search phrases that aren’t searched very often.

Many people associate long-tail with keyword phrases that have a lot of words in them – but that’s not what long-tail is.

Long tail, within the context of SEO, simply describes keyword phrases that are rarely searched for.

While some long-tail phrases may have a lot of words in them, the amount of words in a search query is not the defining characteristic of a long-tail search phrase.

The rarity of how often a phrase is used as a search query is what defines what a long-tail search query is.

The opposite of a Long-tail Search Query is a Head Phrase Search Query.

Head phrases are keyword phrases that have a high search query volume.

Screenshot by author, March 2022

Because there are so many people using the internet, spammers figured out that it was easy to rank for rare search queries so they began targeting millions of long-tail search phrases in order to attract thousands of site visitors every day and make money from ads.

Prior to Hummingbird, many legitimate sites also routinely targeted rare keyword phrase combinations for the same reason as the spammers, because they were easy to rank for.

After Hummingbird, Google began using some of the techniques that Bill Slawski reviewed in his article about the Google patent.

This change to how Google handled long-tail keyword phrases that Hummingbird introduced had a profound effect on how content was written, as many publishers learned it was not profitable to focus on thousands of granular long-tail search queries.

Cutts explained this long-tail aspect of the Hummingbird update:

“So unless you are a spammer and you’re targeting, ‘how many SEOs does it take to change a light bulb,’ and you’ve got all the keywords, you’ve got 15 variants of it, you’ve got a page for each one, you know.

If you’re doing those really long-tail things, then it might affect you.

But in general people don’t need to worry that much about Hummingbird.”

Despite his confidence that this change wouldn’t affect normal sites, Hummingbird did affect some legitimate non-spam sites that optimized webpages for highly specific search queries.

Hummingbird Was A Step Toward Conversational Search

Because Hummingbird was a rewrite of the old algorithm, which made it more precise and fast, it can be seen as a step toward today’s more modern search engine.

All of that one-to-one matching of keywords in the search query to keywords on a webpage was gone.

Combined with other improvements, such as the introduction of the Knowledge Graph, Google was now on its way to developing a deeper understanding of what users meant with their search queries and what webpages were really about.

That’s a vast improvement over the old search engine that matched keywords in the search queries to webpage content.

The improvements introduced by Google Hummingbird may have made this direction possible.

And though Cutts described the initial effect as subtle, these changes eventually lead to a more robust spoken language search experience that had a profound effect on what webpages were ranked and which pages were not ranked.

Search Innovations Sped Up After Hummingbird

What we know about Hummingbird is that it helped Google to better understand conversational search queries; it was a rewrite of the old Google core algorithm; that it helped Google understand the context of search queries; and that Google improved its ability to answer long-tail search queries.

Many significant changes to Google’s algorithm happened within months of the release of the Hummingbird update.

User Intent

Of course, when the conversation is about understanding user search queries, we’re now getting into the realm of understanding user intent.

Being able to remove superfluous words and get to the meaning of what a search query means is a step closer to understanding the user intent.

Fast Conversational Search – June 11, 2014

Conversational search began taking off in a big way in the spring of 2014, about six months after Hummingbird was introduced.

That was when Google was able to integrate the moment current events into the search results.

Read: Let Google Be Your Guide to the Beautiful Game with Real-time Highlights and Trends

Google Hummingbird was so-named because it was fast and accurate.

This new feature gave Google Search the ability to display sports scores in real-time.

There’s nothing faster than real-time, and sports scores are an example of precise information.

Ok Google Comes Online – June 26, 2014

A few weeks later Google unveiled the “Ok Google” conversational search product.

The introduction of the “Ok Google” voice command could be said to be the moment Google finally achieved its goal of providing a true conversational search experience.

Read: “Ok Google” From Any Screen 

Conversational search depends heavily on understanding what people mean when they ask a question. That’s a huge leap forward.

Many other breakthroughs in conversational search followed

Conversational Search And Planning – October 14, 2014

Pravir Gupta, Senior Director of Engineering, Google Assistant posted an article on Google’s blog instructing how to utilize conversational search for doing things like verbally asking Google to find a restaurant or to give the user a reminder.

Read: Fall into Easier Planning with Google

Maybe it’s a coincidence or maybe it’s not that many of these conversational search innovations were released within months of Google’s Hummingbird update.

Regardless, these kinds of conversational search improvements are the sorts of things that Google Hummingbird was meant to support.

Though our understanding of Google Hummingbird could be better, what we do know makes it very clear that the Hummingbird update set Google on course to meet the challenges of mobile search and caused the SEO community to re-evaluate what it meant to build search optimized content.

More Resources:

Featured Image: Henk Bogaard/Shutterstock

In-post Image #2: D-Krab/Shutterstock, modified by author, March 2022