“OK Google, how can my customers find me with voice search?”
“According to SMITH.co, Being discoverable through voice SEO requires strong domain authority, a focus on customer intent, and developing approximately 29 word answers to questions that are written at the grade 9 level.”
In 2016, Comscore predicted that by 2020, 50% of all searches will be through voice. While that number seems a bit reaching when we look at the current uptake, Google has reported that over 20% of mobile searches now begin with voice, so we’re starting to get there. According to Adweek, 67 million voice-assisted devices will be in use in the US in 2019, not including the over 2 billion devices globally, like smartphones and computers, that have voice interfaces. More than half of U.S. teens and 41% of adults use voice search on a daily basis.i
While you still can’t filter your Google Analytics to determine which 20% of initial searches are coming from voice, voice SEO is becoming a more pressing topic for brands who quite naturally want to make sure that their content is discovered as usage of this new interface emerges. Voice search continues to grow both on mobile and through the millions of smart speakers that have been sold around the world.
First, what are the layers of "searchable" content within intelligent agents? When you communicate with a voice interface, its primary activity is first determining your intent, and then trying to provide you with the best available solution.
Response content comes from three general layers of the AI:
- The “Core” AI (hard-coded content) and integrated APIs it can call for information (ie. Weather, Amazon Commerce)
- 3rd Party content like Alexa Skills and Google Actions developed by brands like yours
- The Internet, where voice SEO comes into play
This blog post and the next focus on how to increase discoverability of your web content via spoken interface.
There are two primary use cases for voice search in this context. The first is a user on a mobile device beginning their search with a voice query. In this use case, the user has the benefit of a screen that displays the initial search result, and can move to touch navigation after the query, digging into a website for further answers.
The second is a “pure” voice query where a user is asking a smart speaker a question and does not have access to a screen. In this latter case, the user journey into brands’ content is limited without a Skill on Alexa or an Action on Google Assistant, and ends after each query.
Think of Skills and Actions as invisible websites designed for voice interaction. Without one, a smart speaker is limited to reading high-level content on the surface of the search engine result.
While traditional search is about keywords, voice search is about semantics, context and – most importantly – intent. But when it comes down to the nuts and bolts of optimization for discoverability, the fundamental principles of SEO still reign.
Brian Dean notes that 75% of voice search results are in the top three traditional results for the query, and that domain ranking matters far more than the ranking of the actual page with optimized content. In other words, your traditional SEO efforts for your domain play most of the role in being selected as the response to a request from an intelligent agent, so long as you have voice-optimized content available for the query. So while the topic might seem intimidating, online voice search should feel like familiar territory for digital marketers, with a few nuanced differences.
Here are some tips for approaching optimization as you work to get your content heard in the world of voice.
Spend time with virtual assistants
This seems obvious, but it’s important to spend time playing with the various interfaces to understand the nuances of how voice interfaces work. Google Assistant can be installed on your Android phone in the Play Store, as can the Alexa App, and smart speakers are aggressively priced to maximize consumer penetration, so it should be an easy expense to pass by your understanding CFO. If you have an iPhone, you’re naturally familiar with Siri, who also runs on Apple’s HomePod. Cortana is running on your Windows 10 laptop, if you’re not using a Mac. Finally, Bixby is your reliable alternative to Google Assistant on Samsung devices.
It’s important to note that while Google Assistant, Apple’s Siri and Samsung’s Bixby use Google search by default, that Microsoft’s Bing is the search engine used by Alexa and Cortana.
Think like a customer
Brand messages have little relevance in voice SEO, as customers have questions about specific topics and content must be available to answer these questions. Just as you would with your traditional content that you work to optimize, brainstorm questions and research key phrases that are at the beginning of your funnel to increase discoverability. What topics can your brand be an expert on when a customer needs an answer that might be answered in an informational search query?
Google’s “micro-moments” are incredibly relevant here, illustrating the four key moments that will lead to a search with voice:
- I want to know
- I want to go
- I want to do
- I want to buy
As you work to determine which content to optimize for voice, consider how these sentences would come to life in the context of your product or service, and ensure that you have available content around the scenarios.
One useful resource in brainstorming potential queries is to use the site Answer the Public which allows you to explore questions based on a topic of your choice.
Think in (grade 9) conversational tones
When you’re thinking about how customers engage with virtual assistants, think in conversational tones. In a typical text search, a customer will input their intent differently, perhaps typing “Best Hawaiian vacations” while in voice search the query will be entered as a question or command, like “What are the best destinations in Hawaii?” or “Tell me about Maui."
This is reflected in research by Jennifer Slegg, who notes that the average query length for voice is 4.2 words, while text is 3.2. Campaign Live illustrates the trend in the visual below.
So if people are asking longer questions, how are these interfaces responding?
Backlinko’s research found that, with Google Home, the average answer selected by the device was 29 words, written at the 9th grade level. The reasoning behind the simplicity of the answers is threefold. First, the device should not be reciting encyclopedic answers to ensure speed and usability. Second, answers should be easily understood by the average person. Third, the device should have no issues in elocution of the words to ensure clear communication.
We’ll discuss the implications of this next week in Part 2 of Search Engine Optimization for Voice Interfaces.
As user adoption of search with voice interfaces continues its upward trajectory, it becomes increasingly critical for brands to ensure that they have a voice strategy in place. As you develop yours, our Voice and AI teams are here to help. You can start by downloading our Guide to Conversational Commerce or reach out for a real-life conversation with our team.