On September 26, Google announced that for about a month, a new algorithm had been put in place. The algorithm was called “Hummingbird.” But despite it’s animal-inspired namesake, hummingbird was nothing like Panda or Penguin. According to Amit Singhal, a change this big hasn’t been made since he first joined Google in 2001. It was a complete re-haul of the search engine.
If that’s true, why did this fly under everybody’s radar for over a month, and what does it mean for the future of SEO?
Refining the Long Tail
The primary purpose of the new algorithm is to deal with “conversational search.” Unlike Panda or Penguin, this update doesn’t deal so much with specific sites or links. It doesn’t change the way that Google values sites. Instead, it changes the way that Google evaluates queries, and the meaning behind pages on the web.
The new algorithm switches away from word-matching and toward concept-matching. Previously, conversational search algorithms were only connected to the Knowledge Graph. Now, semantic search techniques have been brought into the organic search results.
By and large, this change doesn’t really effect major search terms. It does, however, effect 90 percent of queries.
The purpose of the update is to determine the meaning of a search query. It’s far more likely to affect the kinds of searches that only one person ever searches for (which is a surprisingly common occurrence). This is all about the extreme long tail, and if you play it right, this could become an opportunity to pick up traffic that would have been impossible to capture in the past.
The update leverages Google’s vast Knowledge Graph, which contains information about 570 million concepts. It understands some of the relationships between those concepts, for example, the fact that Eiffel Tower is a tower, in a specific location, with a specific height. Google’s ultimate aim is to make longer queries return better results than short queries, by virtue of the fact that they give the search engine more information. Traditionally, we all know that the opposite has usually been true.
Scott Huffman told Forbes “we want to get to a natural conversation” between the search engine and the user.
More emphasis has also been placed on the Knowledge Graph itself. We can expect it to start turning up more often for more queries. Needless to say, this keeps users on the search result page longer, making them more likely to click on an advertisement.
Google has shared examples of how this impacts a few specific queries:
- Query: “acid reflux prescription:
◦ Before: This used to be taken somewhat literally, returning a page about medications that treat GERD.
◦ After: Now it returns a page about treatments for GERD, whether or not they involve medication.
- Query: “pay your bills through citizens bank and trust bank”
◦ Before: This used to take users to the homepage for Citizens Bank
◦ After: Now it takes users directly to a page with an app that allows them to pay their bills
- Query: “pizza hut calories per sliece”
◦ Before: This once took users to a somewhat relevant but not necessarily authoritative result
◦ After: Now the search engine understands that Pizza Hut is a business, and returns the nutrition information provided by that business.
These examples are cherry-picked, of course, but they can help give us an idea of the direction Google is taking things with this update. This about moving away from “results,” and toward “answers.”
This will be achieved, in part, by better understanding question words like “how” “why” “where” and “when.”
The Future of SEO
David Amerland, author of Google Semantic Search, told Search Engine Land: “From a practical perspective, the need to identify the USP of each business and become authoritative within it is now a key criteria for continued SEO success.”
To be sure, we’ve already felt for some time that a unique selling proposition is key to SEO. But in the past, this advantage was only true because of the way it affected users and web influencers, who ultimately influence rankings. Now, the influence of your unique selling proposition goes beyond that. We can actually expect it to influence search results directly, and that this will become an increasingly common occurrence.
This all comes down to the concept of “entity search.” It’s about categorizing semantic data, splitting it up into concepts, and identifying how those concepts are being related to one another. Clearly, Hummingbird is still in the rudimentary stages of this, but this is the first time it’s been applied directly to the organic search results, rather than just the Knowledge Graph answers.
While we’re far from the end of keyword-matched search results, it’s clear that SEOs who want to stay relevant will need to address some of these issues:
- We are gradually moving away from matching keywords and toward matching problems.
- With 100 percent “keyword not provided” hitting hard, it’s going to be very difficult to determine user intent based solely on traffic data, and user intent is going to be the key to Hummingbird.
- We will need to start shifting away from keyword research and toward market research, identifying what potential customers want to know, and answering those needs better than anybody else on the web.
- Emphasis on keywords isn’t necessarily going to “hurt” you the way that Panda and Penguin hurt you. Instead, this is about missing out on opportunities. An excessive focus on keywords means you will sacrifice long tail traffic and fail to drive visits from original search queries. This will make it harder and harder to keep up.
- A big part of the reason for this update is the focus on mobile. The need to understand conversational search is the direct result of targeting mobile searchers, who will be asking Google questions conversationally, rather than typing words and hoping for a good match. Increasingly, SEOs will need to work together with designers to create experiences that are suitable for mobile, and this will eventually take things further even than responsive design.
- Ideally, Hummingbird levels the playing field. Precise and complex searches are less predictable than “fat head” searches. However, it’s entirely possible that things could go the other direction. With a semantic understanding of the search query, Google may opt in favor of returning big brand results that don’t match the specific query, but match the concept. It’s not yet clear whether these issues will significantly effect the playing field, or in which direction.
- Keep in mind that long tail searches make up the vast majority of search traffic. The top 10,000 keywords only make up 18.5 percent of the searches. If Hummingbird successfully sends users to the most relevant pages for these long tail searches, this could lead to some dramatic changes. Since none of these keywords individually send very much traffic, however, we wouldn’t expect to see sudden spikes or losses.
- The increasing prevalence of the Knowledge Graph is going to render many traditional search results unnecessary. Simple pages that address simple concepts will be increasingly ignored in favor of the Knowledge Graph results in the search page. SEOs will need to provide value that the knowledge graph can’t, and they will need to get better at converting visitors into members of a repeat audience.
- A major component of the data-mining that has led to the Knowledge Graph and Hummingbird is Google+. It should go without saying that being involved in Google+ gives the search engine more data to crunch in relation to your business, and the concepts surrounding it. This has the potential to go one of two ways. Google can rob you of this information and use it for its own benefit with things like the Knowledge Graph, or it can use this information to better identify you, and when users might be asking questions your site is capable of answering. Only time will tell how this plays out. (I suspect elements of both).
I know I’ve already touched on this, but I feel it’s important to emphasize that Hummingbird hasn’t really created any “losers.” (If your site tanked circa October 4th, that’s Penguin 2.1.) This change is one of pure opportunity. It’s about how Google retrieves and presents data to users. The advantage is going to come from the long tail, so changes aren’t going to happen overnight.