Recent technological breakthroughs have paved new and exciting means for people to search information on the web. These latest advancements continuously affect the search industry in so many ways.
Search engines now use artificial intelligence and machine learning to better understand user intent and interpreting search queries, giving us the most relevant answers to our questions – and more.
One key area that is quickly making progress is voice search. According to Bing, a quarter of all searches performed on the Windows 10 taskbar are voice searches. Last year, Google announced that voice searches account for 20 percent of search queries on mobile devices. The popularity and adoption of voice-powered digital assistants, such as Cortana, Siri, Alexa, and Google Assistant, are also soaring at a rapid pace.
In this episode of Search Engine Nerds, I had the opportunity to interview Christi Olson, Head of Evangelism for Search at Microsoft, to discuss the current state of voice search and what we can expect in the future. Olson also offers tips on how companies should approach voice search and shares examples of low hanging opportunities that search marketers can take advantage of today.
Where are we at and how good are we today with voice search?
Christi Olson: I think you can break it down into two areas when you think about how good is voice search. The way I break it down into those two areas [are]:
- How good is the technology at understanding what we are saying?
- How good are the results that we get back from the technology?
Let’s break it down that way and start there.
When we think about how well do the different platforms understand what we say and how we say it, it has improved leaps and bounds over the last 10 years.
I remember trying to use a voice technology about 2003, 2005, and when I’d speak it would get every other word wrong. It didn’t understand what I was saying. It’s this thing called “word error rate”.
Where we’re at today is about 95-96 percent understanding of what people are saying – meaning that between 4-6 percent of the time, it gets it wrong. When you think about that, that is absolutely amazing. It’s about the same as a human translator, in terms of getting conversation correct. That’s in English.
What I would say is for foreign languages, it’s still developing. It’s still learning. Part of that is that it’s been trained in English. A lot of our work has been done in English.
When you start into, let’s say Spanish, you have Mexican Spanish, Spain Spanish, Castilian Spanish. You have all the different varieties and varietals, and how people say things and the enunciation. It’s still learning, so the word error rate is not quite as amazing in foreign languages as it is in English today, but it’s been getting better.
As more people use essentially voice search and digital assistance and talk to them, it learns, it improves. It understands where it’s going right or wrong and it gets better over time.
Brent Csutoras: There [are] situations where I’m proposed some sort of automated system on a telephone where I’m trying to get through something. There are situations where I think people, any level of error can be very frustrating to people. Do you feel like there are specific industries that can get away with this more and are doing better with it? What’s the acceptable level of error rate in all industries right now?
CO: To break that apart… when I’m thinking of voice search, I’m actually thinking of the digital assistants and the technologies that when you ask a question to Cortana, Siri, Alexa, Google Assistant, that it gets back to you with a response. The technology on that end is actually quite good, and it understands us and it tends to get it correct.
The technology, however on the voice phone systems and the dial-in systems where you call customer service and you have to sit there and yell, “Customer service,” and it doesn’t seem to understand you – they’re using a whole different set of natural language processing (NLP) and AI technology on the back end.
That actually is super frustrating, that it’s not to the same level as the search engines have gotten their digital assistants.
I would say for the listeners and people out there, if you are essentially doing a customer service line where you dial in, I would look at how do you improve the technology you’re using on the back end because that is going to be very frustrating to customers. And if you have a frustrating experience and they’re already upset when they call you, you are not going to be able to maintain that loyal customer relationship or it’s going to be difficult for you to overcome that hurdle. When you have a bad experience and you’re already frustrated and you have an even worse experience, it sort of puts your customer service person in that death spiral of anger.
Do you think that it’s better for companies to wait until they have a technology that really suits their audience? How much can a company afford right now to annoy their audience in exchange for utilizing a new technology?
CO: For voice search specifically, as it pertains to digital marketers, you almost can’t afford not to play in the space and to start looking at investing in voice search. And the reason I say this is when we look up, and I’m going to use a couple of different stats.
I believe the last I heard from Google is they were saying over 50 percent of their mobile traffic is going and using voice search now. And they’re already about 50 plus percent in terms of mobile, so that would put roughly about 25 percent of all their search results coming in through voice search today.
Can you avoid that traffic? Can you, as a business, afford to lose out if you are not thinking about that as part of your marketing strategy?
Within Microsoft, when we look at our volumes, we’re seeing we have Cortana integrated onto well over 400 million devices. There are 145 million active users. Would you want to make it so that people can’t find your website and your content when their engaging and working with that technology? And the response really is no.
The question is how much do you invest and where do you invest. The way I typically, when I talk to businesses, I break it down as what are the main types of questions that your customers are asking about your products, your services, and can you create a great experience. Experience, in this instance, could be a web page that answers those questions and it gives them the information they look. Maybe it’s an FAQ, maybe it’s not. But can you answer those questions in such a way that the digital assistants pick it up and that becomes the voice of the digital assistant answering back?
In terms of voice search today, there’s sort of two ways that you can appear through voice search.
- You become that essential trusted answer, where you are the top of the Knowledge Graph. You are the spoken word that comes back from Cortana, Siri, Alexa, the Google Assistant.
- If it doesn’t know the answer, if it’s somewhat of a black and white where it’s not a yes or no answer, but there might be multiple solutions, they essentially give you a web result.
Do you want to have your web result or your brand there to answer the questions your customers have about your product? The answer, to me, is always yes.
I would say it’s not a completely different strategy than what you’ve done in the past. It’s going back and thinking about your customer and answering questions – being customer-focused and customer-centric.
From a base level, how do you approach the voice search question… knowing that the majority of companies don’t even do SEO today?
CO: I would hope that you have a website. Let’s start there. If you don’t have a website, you’re probably in trouble. Sorry, I just have to put that out. Have a website.
When you think about voice search, essentially the search engines, what they’re trying to do is, you ask the search engine a question or you put something in, and it essentially scours the breadth of knowledge within the internet and tries to give you something back. That’s all search results is – you ask essentially a question of some sort and the search is giving you, to the best of its ability, what it thinks the answer is for your question.
Voice search is exactly the same. When you think about how do you develop a strategy for voice search, I always say, “Go back to your customer.”
- What is your customer’s pain point?
- What are they asking?
- What type of questions?
Build that in a conversational natural language into your content.
I will give an example of a Toyota dealership here in Seattle. They end up having an amazing voice search strategy without even attempting it. They came and talked to me because they were like, “We’re getting all these phone calls from across the country and we can’t figure out why. All these people are calling us asking for specific questions about Toyota safety sense.”
What they ended up doing was they went and talked to one of their receptionists and to their customer service department and said, “What are the most frequent questions that people call in to ask you?” Then they started developing a web page because they wanted the receptionist to spend less time on the phone and be able to direct customers to a page on the website to answer those questions.
They developed a really good web page that had text content on the page that was more in a conversational manner and tone – how you and I would speak today. They put in images that walked through the process of what they were answering for the question and then they created a video. It was all about a very specific Toyota feature… They created a really well designed and developed page that essentially became the instant answer coming back when somebody asked about that specific safety feature.
What they found out is as customers were talking both in voice search and asking the digital assistants a question, that was the number one response back. And when it said, they asked the assistant, “Go ahead and make the appointment and call,” it was calling a specific Toyota dealership here in Seattle, instead of calling the local Toyota dealership for that individual.
They contacted us going, “We created this really great content, but how do we keep from getting everyone across the country calling us?” They didn’t set out to create a voice search strategy, but the content they created was rich and well written and well designed so that when a digital assistant is trying to answer a question, that was the go-to response on the back end.
BC: It seems like when you listen to a lot of people talk about voice search, even mobile, even SEO right now, it really seems to all come back to this kind of silo of answering questions, right? Being more specific to answering the question and less about presenting content that might be more topically rich…
Even Gary Illyes from Google recently said in a podcast I did with him that he envisions web copy being better the shorter it is. Not necessarily the shorter it is as like one line, but we were talking about in a mobile-first [index], when they do make that switch, that less content more focused about what people are actually searching for is going to perform better.
Do you see kind of a strategy with all of these things in mind around creating content that really answers individual questions? Are we start getting into a lot of pages that have individual-focused question and answer versus a section that answers a bunch of questions?
CO: I love how you stated that. The way that you stated that question is do you see a strategy for answering questions in individual pages. I don’t necessarily think they’re mutually exclusive.
I think what I’m assuming Gary is saying is SEO as an industry, we went down the path – probably starting about seven or eight years ago and it’s maintained – where we developed rich pages with all of this content on the page. And then we just kept on developing and adding more and more content onto the page because we were trying to rank for all these different words and phrases that were all related to each other from a topical standpoint.
So we have these really content-rich pages, but the question that we’re trying to get at is the value in the content. Is it giving the consumer what the consumer’s looking for? Yes or no.
Or did we create a bunch of pages with a ton of content because it helps us rank better for that page, but it doesn’t add value to the consumer?
It goes back and takes a consumer-first mindset with when you are creating your content, you’re not creating it for the bot and for the search engines. You’re creating the content for your customer and for your potential customer so that you can actually have them find you when they’re looking for something related to your product, your service, your line of business.
Does it mean we are going to go [away] from having a bunch of pages? I don’t necessarily think it’s a page by page strategy. I think it’s about giving rich contextual answers and rich contextual content that provides value to your end user. Less about creating that content for a bot or for a digital assistant. It’s going back to being customer-focused.
Are there other low hanging opportunities that you’ve noticed in discussions when it comes to voice search?
CO: I’m going to take this from [my] context. There are people who don’t know me – I have an 18-month-old son. Because of that, I’ve been using voice search a lot more, trying to answer questions while I have my phone in one hand and him on the other hand, trying to figure stuff out.
What I’d say is I’ve noticed as I’ve used voice search more and more and more, is it goes back to when I use voice search, I ask it a question. I don’t use voice search like I’ve used text search. I don’t go and just say, “Weather.” I ask, “Will I need an umbrella?” I don’t go and say, “Baby rash on behind.” I go like, “Hey Cortana, is it normal to have a rash that looks like this in this location?”
When I start to look at the type of questions I’m asking, where I’ve personally seen the gap is that a lot of those questions don’t have search results for it, as it pertains to brand or products or services. A lot of times, especially on the baby side, I’m pretty sure the company, either Boudreaux’s Butt Paste company or Desitin, could have owned me from a search result standpoint of answering every single question I’ve had in the last six to eight months. But what I’m finding is I’m not seeing anything related to either of those brands. I’m getting forum sites because I’m finding other moms who are asking the same questions and there’s still no answer for that question.
Going back to thinking about your consumer – what are the questions they’re asking as it pertains to your products, to your brands, your services? Because that’s where there is some really low hanging fruit today.
People are asking questions over and over again, and when they’re using voice search, they’re just now doing it in a much more natural language tone and conversation.
The other thing that is a low hanging fruit for brands and for people is, right now when we write content, we don’t really write it in how you speak. I had a lot of fun with this conversation with Google back in February at the Moz Local, going, “Okay what does this mean in terms of content writing? Does this mean we write content more in a natural language?” I went back and forth with this one. I’m not saying essentially create a page on your site that is a question and answer, a question and answer, a question and answer. That’s not the answer I’m giving.
[I mean] answer the questions your customers are asking, and answer it in a more natural language instead of writing this superfluous long content that is convoluted and doesn’t really mean anything.
BC: Another way you could do it is sometimes you can answer as if you’re repeating the question. A lot of people ask this question, and when you’re approaching it from this way, this is how you would do it, right. Sometimes you don’t even have to do the Q and A. You can actually find ways to incorporate that question into a creative way with the content.
CO: Exactly. Some of this from the conversational tone. It goes back to you writing for your customers, so make it sound like you’re writing to them and not just a corporate drone writing content based on brand guidelines and standards.
BC: It’s interesting that you do see quite a bit of the common language come from forums. Maybe a good angle is to really say, “You should be checking forums for the industries that you’re in, looking for questions that are going unanswered. Looking for patterns in how people are asking those questions,” because forum people will put a common language question into the actual subject line of that page.
Right now, voice search is stringing together search parameters, trying to find a connection between the previous, the current and what the next might be. Have you seen examples of companies that are making an effort to really transition that into true conversation instead of just search parameters? When do you think we’ll see true voice search?
CO: It’s interesting. You asked at the very beginning of the conversation where are we at with voice search. I always like using the crawl-walk-run analogy. I would say most companies are probably in the crawl phase, where they’ve heard about it, they understand that it’s coming, and they know they need to do something about it, but they’re not sure what to do or how to do it yet.
In terms of when will we get there, it’s two-fold on when we get there.
- Companies have to create the content so that it answers questions and provides value back to the consumers.
- We’re already well on the way of getting there in terms of adoption of the technology.
To answer one of your other previous questions, are you saying just do voice search because it’s there? …[W]hen I’m talking to brands and companies right now about, from an AI-perspective, which areas of AI should we focus on, I mention voice search as one of the areas. It’s a low hanging fruit because you can do it today without investing in any additional technology. You already have a website. It’s updating the content on your site to be a more conversational tone in nature.
The next area I always talk about is chatbots. Chatbots originally were answering questions provided answers, so it’s FAQ type of area. But you can also create a chatbot to help consumers take action. So I’m asking questions about your products and services, great. Now let me buy it. Take it from answering the question to taking an action.
The third area is like when we think about tools and technology adoption, I can use voice search on my Windows 10 PC. I can use it on my Apple smartphone. I can also use it on Alexa, on a smart speaker device, and you have those voice skills. Skills is the area I’m telling people, think about it but don’t jump in quite yet. You want to really think about that skill strategy.
When you think about it, PCs, I think there’s what? Over 450 million PCs in the U.S. There are over 150 million smartphones in the US.
Then on voice skills, which really are being used on the smart home speakers… think in January the number was about 9 million Alexa devices had been sold to date. The adoption is still growing in that area. And it’s an area that people, like the smart home speakers, consumers are saying they want to purchase and they’re going to use them more, but comparative to PCs and smartphones, which have voice search technology and access the power of the internet and hit websites, the adoption’s already there.
It’s going to be a lot of fun. I think we’re still probably five years off from being screen-less everything. I guess I’m just having a hard time imagining not having something with a screen near me. I can’t imagine working in a holographic ad, more of the hologram-type of world. I imagine the screen is still going to be very important in the next foreseeable future. What it means [is] that you still want to create the content and create experiences developed for a screen.
To go 100 percent voice search, it’s going to start to happen as people get more confident and more familiar with asking the digital assistant a question and getting a response that comes back. As they get more familiar with using that digital assistant, it makes recommendations, and it gives them more value, then that’s going to be that tipping point because then we’re going to be used to engaging with our voice alone and we don’t necessarily need that screen.
More Voice Search Resources Here:
- Are You Optimized for Voice Search? Here’s How to Do It
- How Voice Search Will Forever Change SEO
- Using Voice Search & AI for Content Marketing: An Interview With Purna Virji
To listen to this Search Engine Nerds Podcast with Christi Olson:
- Listen to the full episode at the top of this post
- Subscribe via iTunes
- Sign up on IFTTT to receive an email whenever the Search Engine Nerds podcast RSS feed has a new episode
- Listen on Stitcher, Overcast, or Pocket Casts
Think you have what it takes to be a Search Engine Nerd? If so, message Loren Baker on Twitter, or email him at loren [at] searchenginejournal.com. You can also email Brent Csutoras at brent [at] alphabrandmedia.com.
Visit our Search Engine Nerds archive to listen to other Search Engine Nerds podcasts!
Featured Image: Paulo Bobita
Subscribe to SEJ
Get our daily newsletter from SEJ's Founder Loren Baker about the latest news in the industry!