Wondering about the future of search? Odds are, the answer won’t be found anywhere near your keyboard.
By 2020, at least 50 percent of searches are going to be through images or voice, and people won’t just be adding items to their Amazon shopping list. Digital marketers across virtually all industries will need to change their approach to content creation and promotion if they want to keep up in this voice-driven, AI-enhanced world.
To find out how to optimize content for voice search, we reached out to Will Scott, CEO of Search Influence. Here’s what he had to say.
How is voice search different from text search?
The two most obvious ways in which a voice search is different from the text search bar are the query length and the format of the query.
Over the last several years the length of the typical Google query has increased significantly. Some of this is, no doubt, searchers learning that the more information they put in, the better response they will get. Some of this behavior is also likely to have evolved from the types of queries that are used in voice search.
On the query format, voice search lends itself more to asking questions. The possible responses to a question query are somewhat different in that they want to provide the best answer.
You’ll notice, for instance, that for many obvious question queries, Google has started presenting an answer box. There’s a leading mega-result at the top of the page which seems to be trying to answer authoritatively the question.
In the case of voice-search appliances like the Amazon Echo and Google Home, you can see how this could lead to much more of a query and single response than to the list of links or options to which we’ve grown accustomed.
What impact will AI have on search, particularly when layered in with this shift to voice?
AI and machine learning should help to infer from ambiguous data the meaning of the question asked, as opposed to just looking for keywords within the question. Voice tends to be more free form, and there will be much more extraneous data due to the length of the query.
Often times, when we talk about AI, we are really talking about machine learning. Machines learn through new inputs and tailor their understanding of the world around them based on bigger and bigger data sets. So, the combination of faster computers and bigger data allows the machines to have a greater corpus of potential questions and answers upon which to base decisions.
How does this change how we should be thinking about online content?
The biggest change in thinking about online content will be a greater focus on answering questions. The more your content answers questions, the more likely it is that your content will rise to the top.
From a practical perspective, this shouldn’t call for much of a change on the part of marketers and content producers. We have always recommended to our customers that when thinking about important content areas, they should look to their frequently asked questions, whether on their website or via direct interactions with customers.
Tools like Schema and JSON-LD become more important to assure that the question and the response are both clear to the search engines as they come crawling. In cases where the question calls for a response that has multiple options, the use of tabular data and grids seems to be having a positive impact on one’s ability to take the top spot or, as Dr. Pete Myers of Moz calls it, “Position 0”.
A final, important, consideration is context. We will learn over time, with more data, which queries are most likely to occur where. For instance, a query in the car is very different than a query at home, the office, or on the sports sideline.