In December, the makers of voice question-and-answer platform MindMeld released results from a survey of 1,800 adult smartphone users. They wanted to know two things: how many of the people they surveyed had used voice search, and when they’d started searching via voice.
Thirty-seven percent had not yet tried voice-assisted search. However, 60 percent had started using it within the past year, and 40 percent had started using it in the past six months. Their (admittedly self-interested) results suggest voice search is poised for rapid growth.
On-page SEO has always prioritized the keyword phrases users type into search engine fields. Before Panda, people often stuffed on-page copy with as many potential keyword variations as possible. Panda cleaned up a lot of keyword stuffing, but SEOs still spend a significant amount of time researching the perfect keyword phrases for their clients.
The shift toward voice search could mean big changes in keyword usage and SEO as we know it.
What Is Semantic Search?
With the introduction of Knowledge Graph back in late 2012, Google signaled its belief that semantic search was the future. Semantic search is the effort to put a searcher’s query in context. Search engines learn to understand the intent behind a query so they can deliver more accurate results.
The first thing semantic search attempts to do is understand how a search term fits into different contexts. The servers that execute Google’s search algorithm, using a process called machine-based learning, build context around different terms using a number of techniques.
Computers analyze existing documents to understand how a term is used. They learn which other terms commonly pair with the key term — co-occurrence — and which other terms seem to mean the same thing.
As the search engine builds webs of connections around each term, it begins to understand what the term is about. It also begins to identify searches based on their context. For example, it learns when searchers want to learn how to make a fruity spread for their toast (jam) and when they want to learn to improvise as part of a rock group (also jam).
User Characteristics and Behavior
The types of users who search for the term, as well as their behaviors, reveal a lot about user intent. Computers might note, for example, when large numbers of users search for one particular phrase at a point in time, an indicator that the term is tied to a major news or sports event. Computers also note which search results engender particular user behaviors, and they draw conclusions based on those behaviors. For instance:
- Users make a certain query and click one particular link.
- Users who click that link always end up bouncing back to the search results.
- Computers decide the link must not be satisfying the user’s query.
- The page’s ranking drops for that particular query.
As computers gather data on context and user behavior, they learn to predict with more accuracy which links will satisfy a particular search query. The result isn’t based on how skillfully the copywriter inserted particular keyword phrases but on how well it satisfies a user’s query.
In the future, the keywords in your copy will matter less, and your page’s effectiveness at fulfilling the searcher’s intent will matter more. To successfully interpret voice searches, search engines can no longer take keywords so literally. They have to understand what the user wants, whether or not the query incorporates keywords.
Is Keyword Research Still Necessary?
As Google has disclosed more and more about machine learning, some in the SEO world have decided keywords no longer matter — and that’s incorrect. It’s still essential to know the keywords for which your pages rank for two essential reasons:
- It helps you understand how Google interprets your pages and how well Google thinks your pages perform when it comes to particular keywords.
- Performing keyword research ensures your keyword phrases are still relevant, and it lets you know how queries are changing around particular subjects.
The presence of a keyword phrase on a page, however, isn’t going to be enough to earn a ranking. You’re also going to have to broaden the way you think about keyword phrases. Instead of using them verbatim, think of all the different ways someone might make an inquiry about the keywords for which your page ranks.
Which brings us back to voice search.
Imagine someone driving in a car near Toledo, unlocking his iPhone, and asking where to get a burger in Toledo. Siri won’t just pull up pages that say “burgers in Toledo.” Results will depend on:
- Most likely, the driver wants a burger from a nearby restaurant and doesn’t want to drive across town.
- Positive review data. It won’t be enough to write a keyword phrase like “best burgers in Toledo” into your copy. Your assertion will have to match data on Yelp, Zomato, and other review sites.
- The searcher might want a cheeseburger, a meatless burger, a bison burger or a turkey burger. Restaurants offering a more complete menu are more likely to top the search results.
- Past user behavior. If users have clicked the result and then left positive reviews after their visits, the website will appear more often when others have similar queries.
- The search engine will most likely choose a restaurant at which people check in on social media or a restaurant with an active social media presence. Your inbound link quality will also provide considerable context related to whether the search engine has confidence in your page.
Your content in the era of voice search has to do more than just incorporate a keyword. It has to provide the best possible content related to a query, supported by signals of being liked by others. If you can’t deliver the best content, you have to deliver something unique. Competitor research becomes more important than ever.
Better Content for a Voice-Driven Era
By all means, keep putting keywords into your content, but remember the words themselves are no longer enough. When you evaluate your pages, ask yourself:
- What questions should this page answer for visitors? Is it currently answering those questions?
- Does it offer complete information? In other words, would someone return to the search results after checking this page?
- In what contexts do people search for a page like this? Is it ready for mobile search? Will the first impression cause people to want to stick with the page?
- Is the content approachable and easy to read? Can a visitor scan it and understand immediately what the page is about?
SEO used to be about completing checklists: include keywords, build inbound links, add a title tag, etc. Now, the checklist is only the beginning. The user will get an increasingly bigger vote in how well each page ranks.
Inbound links used to be the ultimate vote of confidence for any page, and they were a factor that SEOs could manipulate. Today, for search engines increasingly accessed by voice, SEOs must optimize pages for user behavior above all else. In 2016, the user reigns supreme.
What adjustments have you made to accommodate for this shift? Let us know in the comments below.