AI and How It’s Affecting, and Enhancing, Voice Search
July 11th, 2017 by
If 2016 was the year of the Internet of Things, 2017 has become the year of AI, and, in particular, voice search. The way people search on their phone, tablet, or computer is changing thanks to devices like Amazon Echo, Siri, Cortana, and Google Home—as of May 2016, 20% of mobile queries are through voice. You can now search for a business, order pizza, turn on your home’s lights, ask for movie reviews, and more just by asking a question or stating an instruction. According to the 2016 Internet Trends Report, by 2020, at least 50% of searches are going to be through images or voice, whereas in 2015, only 1 in 10 search queries came through voice. Understanding how your content is optimized for voice search can help your business succeed in the near future.
AI, RankBrain, Machine Learning, Deep Learning…What’s the Difference?
With all of the talk about AI these days, it’s best to separate different terms that get tossed around, like RankBrain, Machine Learning, and Deep Learning. The easiest way to understand their relationship is to think of them as smaller parts of a larger whole. AI has been around since Alan Turing first started trying to decode the Nazi’s Enigma in WWII. In the 1980s, we started to see more advancement with machine learning, which provided computers with the ability to learn predictions and statistics without being explicitly programmed. In the current decade, we have begun to see large advancements in deep learning.
Deep Learning
Also known as deep structured learning or hierarchical learning, deep learning uses neural networks with many hidden layers of processing and information. It mirrors the way the human brain works, learning from mistakes to come up with the best, new solution to an input, not one that it was given with code. For instance, Marie Dollé from SocialMediaToday explains how it recognizes your face in a Facebook photo.
“For a face, it will first correlate pixels on a photo, then group them and conceptualize its features (eyes, nose, mouth). Finally, it’ll group all these concepts to recognize the unique collection of features as a single face.”
It’s also how the world’s best Go player, Lee Sedol, got beat by Google’s DeepMind AI dubbed AlphaGo. Go is an ancient game of immense complexity—there are more possible positions on the game board than there are atoms in the universe. This is precisely why we never thought a machine could master a human in this game; it’s more about intuition and instinct. Wired Magazine said that the pivotal play in the game was also the moment that “machines and humanity finally began to evolve together. While the move that set up the machine to win was puzzling to humans, it opened Lee Sedol’s eyes to strategies he hadn’t considered before.” We’ve come a long way since the first game mastered by AI in 1952—Tic Tac Toe.
What About RankBrain?
It’s best to think of RankBrain as a facilitator for deep learning. Unveiled by Google in 2015, RankBrain converts words and phrases into mathematical vectors that can then be used by deep learning. One of the more interesting things about RankBrain is that it searches for phrases that have never been used before on Google and then attempts to find similar words or phrases that may relate, i.e. your rich content that may not match up perfectly with a user’s search but is close. To put it simply, Greg Corrado, a senior research scientist at Google involved with RankBrain states, “If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries.”
So, why are we getting these never-before-seen queries? Well, in some way or another, this will always happen. New products will be released, and people will want to find them. However, the introduction and evolution of voice search as a viable tool has transformed the way we talk to search engines.
The Importance of Semantic Content Marketing in the Age of Voice Search
Gone are the days of planning your content marketing strategy based on one or two keywords. Sherry Bonelli from SearchEngineLand posits, “Now keywords are no longer just keywords. Keywords in the voice search world are long-tail+. The ‘plus’ refers to the conversational phrases that you need to add when optimizing for conversational voice search.” Because people speak things differently than typing, we have to start adjusting marketing strategies accordingly. In comes long-tail+ keywords and semantic SEO.
Long-tail+ keywords should mimic the way people actually talk. Think about what types of questions people ask when they call your business, and then think about any follow-up Q&A you have with them. Also, in addition to thinking about what people might be asking or searching, it’s time to start thinking about all of the synonymous and relative topics to that search. The age of the “best digital camera” keyword could likely fall away. Instead, consider the question, “Google, where can I find a digital camera that is waterproof and can also sync w/ and stream Facebook Live?” Look at everything that’s in that question. It mentions streaming and Facebook Live capabilities, it specifies waterproof, and the whole thing is rooted in a local question: the “where.”
So, if you were to then take a hard look at your content, does it mention syncing devices with live streaming, have you written a blog about waterproof devices that you offer, are all of your physical location pages up to date, and have you claimed your Google My Business listing? Or, have you made the mistake of optimizing for “best digital camera” and called it a day? We’re starting to get to the point wherein Google’s AI technology is becoming as quick thinking, on-the-fly, and complex as us. So, we need to optimize accordingly.
Optimizing for Voice and Conversational Searches
Using long+tail keywords is a good first step to optimizing your content for voice search. Try to understand and predict the user’s next intent or action after finding the answer to their first query. For alternate words or keywords, instead of trying to go for the “be all end all” keyword, think, “tell me how to get to the store,” alongside, “directions to the store.” Experiment with different versions of the same idea.
Local is going to continue to be a big part of success for a business. Research what interests customers in a targeted area and optimize all of your physical location pages by making sure they’re marked up with schema. Structured data markup from schema.org is crucial because it makes it easier for search engines to parse your site and understand your content within context. Here’s a helpful list of additional best practices that can help with showing up in voice search:
- Claim Your Google My Business Listing: A large part of voice search is based around local. By claiming your Google My Business Listing, you provide Google with more rich information about your business, giving its AI more tools to serve your potential customers with rich, accurate answers to their questions.
- Use Conversational Keywords: This goes without saying, but if people are searching more naturally, then you should be writing more naturally in your content.
- Make Sure Your Content Is Trustworthy: Just as AI can evolve to find us the best possible answers to our queries, it also has the ability to learn from and reject black hat SEO practices. Steer clear of link spamming, keyword stuffing, and untrue statements in your content.
- Set Up a FAQ Page: This may seem like a minor addition to a website, but it’s a simple way to get a lot of good answers to questions, providing content for a vast amount of subjects about your business that could then make it easier for Google to find you.
Because RankBrain plays a big part of voice search, finding never-before-used phrases and pairing them with trustworthy, on-topic content, it’s worth remembering some advice from a former Influencer: “If your content is written in accord with Panda and Hummingbird, meaning it is rich text, educational, and really on topic, RankBrain should not be a problem for you.”
How Google’s Cloud Video Intelligence Is Changing Video Search
Ever wish you could search for a video online or in your cloud library just by asking certain words? Like, “Hey, Google, show me videos with fun cartoon animals,” and getting a result of Zootopia from your library. With Google’s Video Analysis, we’re getting to do just that. This can have a large impact on more than just personal queries. Pretend you’re a Realtor in your local area. By using this tool, you can make your videos more discoverable for potential clients. It’s just another way AI is changing the way we search.
The Future of Google AdWords
Google recently held their Google Marketing Next event. One of the features that they mentioned talks about Life Event Targeting. In the past, Google’s AI had been able to run basic “if, then” scenarios for target buyers. Now, they’re able to parse through multiple languages to predict a similar response for the same event. For instance, they gave the example of those recently engaged. They were able to recognize the different but similar terms and phrases that cultures use when people are in a wedding life stage and then provide them with solutions to queries.
Also, Google is learning to adjust ads and tailor them to each user’s preference. Instead of three people doing the same search getting the same ad display, there will be advertising that suits each person’s preferences for style, graphics, tone, or display. They may have alternate text that is highlighted, one focusing on price and another on the value of a product or service.
Where Have We Seen This Before?
Whenever I used to think about artificial intelligence, I would typically conjure images of a Terminator landscape, or a young Haley Joel Osment sentient robot in A.I., or the philosophical complexities that Harrison Ford faces as a replicant (or not?!) in Blade Runner. There have always been predictions about future technology that we can find in pop culture, from space invaders to complete AI domination. However, I think one that best represents where we’re currently exploring and where we may end up being is seen in Spike Jonze’s film, Her.
There’s no dominating technology that overtakes our lives, no artificial intelligence that turns evil. Instead, it’s a future in which technology is doing what we’re trying to teach it to do now. An operating system’s AI is learning from conversations and adapting and adjusting accordingly. The OS can also see people and judge their facial reactions, something we can already do with facial recognition. Mark Zuckerberg even implemented it in his home.
To quote the film’s production designer, K.K. Barrett, “This is not a future of harshness, but of bespoke details.”
What Does the Future Hold for Search and SEO in Its Relation to AI?
Voice should continue to rise in use; there’s nothing to suggest otherwise. And while this is a bit of a“Wild Wild West” time of development, there’s one nice thing that should placate marketers’ worries—the relationship between the marketer and the client isn’t going to change in some cosmic or complicated way. In fact, it’s the opposite. It gets more personal. Ideally, we’re headed for a future in which our old anxieties about technology are gone—a future in which we no longer worry about technology ruling or destroying us.
Instead, it may be like what we’ve seen in the movie Her. It’s getting more intuitive. More seamless. More subtle. More adaptive. More inseparable from who we are. It may not be too much longer before voice commands aren’t even necessary—Deep Learning may evolve to where simply walking into a room turns on the lights, and those lights then adjust to a brightness or dimness reflecting your facial features and mood…maybe.
Instead of a future of wild advancements of technology, of jetpacks and holograms (well, kind of not those), it’s more a future of subtlety. One where we’ll hopefully be able to teach a program to learn from us so well and so naturally that it doesn’t even feel like teaching. To quote a reaction from those watching Google’s AlphaGo win its matches of Go over Lee Sedol, “The machine did a very human thing even better than a human. But in the process, it made those humans better at what they do.” So, instead of tech moving to the forefront of our lives, in front of our faces on the ride home from work, disrupting family dinners with scrolls and feeds, it could dissolve to be part of the background, one that not only knows when to talk to us but also when we can be left alone to be better at what we do.
Images:
This is a great post, Matthew!
I especially like the tactical takeaways – in a world where people are asking their devices all kinds of questions, the FAQ finally has real search value.