The Reality of Voice Search
Every year since 2017 has been dubbed The Year of Voice Search. Multiple statistics have led the industry to believe this. The best example being the mention popularly attributed to comScore that “50% of all searches will be conducted via voice by 2020”. This quote specifically though, was taken out of context and originates from a completely different source. It comes from then-Chief Scientist at Baidu, Andre Ng, who referenced the combined potential growth of both voice and image search relative to China and Baidu back in 2014. At that time, search via voice and image accounted for roughly 10% of all queries. Neither Baidu (nor Google) have recently commented on the market share of voice search compared to text-based search.
Voice Search Compartmentalized
We also need to keep in mind that voice search exists in two separate silos that are often incorrectly combined. There’s voice search via mobile devices, and there’s voice search via stationary home assistants. Both are growing, with home assistants seeing an anticipated global growth of 35% this year. And despite both serving similar purposes, they often ingest very different queries.
There exists a big difference between the commands normally associated with a smart speaker query and the questions normally associated with a mobile voice assistant. For example, a voice command provides a definitive answer (“Who wrote The Old Man and the Sea?”) or leads to a specific act (“Alexa, turn off the kitchen lights.”), whereas a voice query (“What’s the best pizza place near me?”) requires multiple search results as the output in an SERP format. Voice commands intended for digital assistants also aren’t included in Google Ads data, whereas inputs via voice on mobile phones are. For these reasons, the latter is what most advertisers interested in voice search should focus on when considering volume potential.
Where to Focus Your Efforts
Focusing on mobile queries is a start but diving deeper in relation to the queries themselves is even more important. As it relates to the queries associated with mobile voice, the focal point should be on the following:
- Answering general questions
- Location-based queries
Answering General Questions
Answering general questions is one of the main outputs of voice search. How Google Voice – for example – replies to questions is heavily based on what it deems to be the most relevant content. Often, the reply via a voice solution is a word-for-word reading of a featured snippet:
In order to be included as a featured snippet, a site has to be deemed a knowledgeable source by Google in relation to a particular query. In order to be deemed knowledgeable, that site has to be a provider of premium, original and trustworthy content. For most advertisers, this would mean the inclusion of relevant, thought-provoking (or thought answering) content that deserves to be elevated. Content that informs consumers without hinting at an agenda. If a site is aiming for inclusion into the query-based side of voice search, improving their site content and overall domain authority is the best route.
When it comes to voice and location-based queries, Google’s recently released breakout of voice search categories (below) would lead one to believe it’s not a large piece of the pie. But in reality, we have to assume that a) it’s a big piece seeing as location-based queries are incredibly mobile-focused and b) the below is exclusively focused on smart speakers, not mobile devices. As noted, smart speakers (commands) will receive very different queries than mobile devices (queries).
In order to capitalize on location-based queries, advertisers need to focus on location-based keywords and local listings optimizations. Keywords containing the phrase “near me”, “nearest”, “closest”, etc. should be a part of a location-specific advertiser’s keyword builds. A large portion of the recent keyword expansion Empower is undertaking includes longer-tail versions of these types of terms. Head terms containing phrases such as “location” and “near me” will also capture traffic from longer tail queries (i.e longer format, natural language) normally associated with voice due to match type limitations expanding as of the last few years.
Local listings optimizations can go a long way towards voice relevance as well. It’s paramount for advertisers with physical locations to keep their Google My Business, Yelp, Yellow Pages, Bing and all other listings as up-to-date and fully fleshed-out as possible.
What’s Next for Voice Search?
Comprehension. In order for voice to be as widely adopted as most stats make it seem, it needs to be able to comprehend human speech at a near perfect rate. We mentioned this in a previous blog post, and comprehension rates are close, but not near the levels of accuracy they need to hit in order for language to be discernible regardless of context or environment. On the plus side, advancements in artificial intelligence are helping to bridge the gap. Google’s RankBrain is an example, a natural language processing AI with an end goal of establishing more predictive outcomes as it relates to speech combinations. When the tech encounters a new phrase or question, it produces a “best guess” answer based on the searcher’s meaning and intent. The ability to react to – and learn from – ever-changing speech patterns in a conversational manner is the type of functionality missing from voice’s current iteration. There’s a mile-wide chasm between a voice assistant programmed to answer canned questions and one that can truly participate in ever-evolving conversation.
Another possible next step was previously discussed as well, but this falls more in the wheelhouse of appointment-driven advertisers. The idea that digital assistants and voice search can be used as true personal assistants. Google Duplex is the perfect example. In this potential evolution, a person could ask their digital assistant to schedule an appointment for them, and it would be able to perform the act of setting up an actual appointment (based on the individual’s calendar) in the real world. This would entail interacting with someone on the other end of the line, maneuvering through scheduling options and finalizing the appointment while providing a confirmation notification for the person attending. This could lead to an increase in phone-based appointments and interactions as it relates to such advertisers as well as an uptick in call center interactions (which, in most cases, are trending downward as more and more consumers continue to favor online scheduling) in the near future.