voice search is no longer “the next big thing” in search: it’s here. an estimated 20% of mobile searches in 2017 used voice input and research from comscore indicates voice will account for 50% of all web searches by 2020.
whether or not comscore’s projection proves accurate, voice is undoubtedly becoming a more significant part of the search picture.
so, let’s make that picture clearer. in this guide, we’ll discuss what voice search is, how people are using it, and how you can optimise your content to maximise performance in voice search results. you’ll leave with a mix of information, theory and practical tips that will help you factor voice into your marketing strategy.
what is voice search?
voice search is a web search where users can input their search query vocally, as opposed to via traditional text input. leading providers of voice-enabled search include google voice search, cortana, siri, amazon echo and baidu duer.
the basic steps of a voice search are:
1. the user verbally asks their query, which is captured via the microphone of their smartphone, tablet, computer, wearable or voice assistant device.
2. the search engine processes the query.
3a. if the search is via a device with a screen, multimedia search results are served to the user, similar to the results of a traditional search on smartphone or computer.
3b. if the search is audio-only, an audio response is served to the user. this is the case with most voice assistant devices.
some voice searches involve several rounds of interaction, with the search engine clarifying the user’s intent by asking follow-up questions.
who is using voice search?
global web index’s 2019 voice search report offers some fascinating insights on where voice currently sits in the search landscape. their key findings include:
- 27% of web users worldwide use voice search on mobile.
- 17% of web users have a voice-controlled smart assistant device, while another 34% say they are interested in buying one.
- 37% of web users used voice search on any device in the month leading up to global web index’s survey.
- adoption is higher among younger users. 35% of 16-24s used voice search or voice command tools on their mobile in the past month, while only 11% of 55-64s said the same.
- adoption is not especially high in the united kingdom. only 19% of internet users here used a voice search or voice command tool on their mobile in the past month – a far lower rate than was seen in users from indonesia (38%), turkey (33%) and the united states (25%).
we strongly recommend reading global web index’s report to get a fuller picture of who uses voice search, so you can measure up the current trends against your own audience research. if the demographics in your audience seem similar to those that use voice search the most, this may be an especially important topic for you to focus on.
you can download the report for free here (you’ll have to enter your email address).
how smart assistant voice search differs from voice search on mobiles
voice controlled smart assistant devices currently support a smaller share of global voice search activity than smartphones, and as such, they are arguably of lower priority for marketers targeting voice traffic.
that said, smart assistant devices do represent a significant minority that’s steadily increasing in importance as global smart assistant device sales grow. the global installed base of smart speakers running google home, amazon alexa, apple siri, microsoft cortana and other voice assistant products was projected to reach 100 million units by the end of 2018.
whereas voice search via smartphone or computer works similarly to traditional web search, voice search via smart assistant device is an entirely different process. instead of delivering the traditional mix of links, images, videos and other content, smart assistant voice searches usually deliver audio responses, sometimes asking follow-up questions to help clarify the user’s intention.
this is a new type of search, and people are using it differently. according to google, “assistant queries are 40 times more likely to be action-oriented than search”, with people asking for things like “send a text message”, “turn the lights on” or “turn on airplane mode”.
these queries are highly transactional, with users typically seeking quick, convenient, concise solutions. if you’re targeting voice search traffic from smart assistant devices, all the adjectives from the previous sentence should describe your content.
these stats from think with google offer an interesting window on how people are using voice-activated speakers. the figures highlight that parents are above-trend users of voice-activated smart assistants, perhaps because they have a greater need for the efficiency savings it can provide. we have known it for some time but it looks like finally, the data is here to prove that parents literally do have their hands full.
they also show that 52% of users (whether they were parents or not) asked their smart assistants about information on local places at least weekly – an important point to bear in mind for marketers representing bricks-and-mortar locations.
we are also interested to note that according to google, smart assistant voice search queries are “200 times more conversational” than traditional search. more detailed questions are being asked, resulting in short, specific answers that make sense as an audio response.
this might mean it pays to target the specific parts of a user journey where the intent to do something that completes a conversion is highest. so, for an saas provider selling an email marketing solution, the biggest payoff might come from targeting the search query “what is a good email marketing app?”, rather than “what is email marketing?” at least, that’s just our theory!
another way voice assistant queries differ from traditional search queries is that they tend to be phrased more naturally. “almost 70% of requests to the google assistant are phrased in natural language, not the typical keywords.”
the limitations of smart assistant voice search
smart assistant voice search is great for straightforward queries that can be answered with an audio message no more than a few sentences long. however, many searches require a more comprehensive response. that could mean a selection of links; it could mean images and videos, or it could mean a more extensive body of text. not every question has a simple answer – and those that do not are generally best answered with a traditional search.
there are whole categories of content that currently don’t work in an audio-only medium. one example is recipes, which on the face of it sounds like something a smart assistant could be a big help with. we thought it would be worth taking a step back from our overview of voice assistant technology so we could actually look at what the technology offers in practice with the currently available hardware and software. to take a walk in the wild we asked our twitter followers who had made use of voice assistants to explain to us a great use for such tech. one of the common threads that came back was using the device in the kitchen as its hands-free. to test this out, we took a look at how voice assistants perform for the troubled chef in need of a guiding voice.
voice recipe assistant’s kitchen test
simple voice assistants like apple’s siri on ios isn’t able to interact with the user multiple times on a single recipe – for example, reading out step by step ingredients as you pull them together – but more advanced systems like amazon’s alexa can do this if asked to add the skill. out of the box, recipe assistance isn’t something alexa can do. the user must add a third-party alexa skill to achieve this functionality; otherwise, it’s just as incapable as siri is. several companies have attempted to develop alexa skills which provide the kind of functionality a novice chef might appreciate. check out this demonstration of allrecipes alexa skill.
we’ve used recipes from allrecipes for quite a while via our smartphone. it’s a great way to quickly get what you need for almost any recipe. however, here is where the technology fails somewhat. allrecipes has built the alexa skill, but it is not yet available to the uk amazon alexa store. if the reviews are anything to go, that could be a blessing-in-disguise. at the time of writing this article, 61% of user reviews were just 1 star with 81% giving 3 stars or less.
in the uk store, recipedia amazon skill by unilever faired a bit better with 39% 5 star reviews but still had 47% 1 star reviews, with many reviewers complaining that the recipe selection is limited and has a bias towards unilever pre-prepared food products.
google home is also able to provide assistance to the chef in the kitchen, and the device takes a broader approach with google having partnered with over a dozen supported cooking partners whose recipes will play nicely with the device. they also provide a helpful google home voice command instructions document to support users in using voice commands to control the device while cooking.
so, although voice assistants could be a great addition to your kitchen in-theory, the skills and functionalities that really wow users and win them over are yet to be developed for many of the dedicated voice assistant platforms. google’s device has a broader approach to recipes which looks like a better strategy, but teething problems on popular platforms like amazon’s alexa could hold back widespread user adoption of this technology.
it certainly seems there are big differences in the ways different platforms allow users to control their voice activated device. as yet there does not seem to be a set of unifying user interface standards across different platforms, which may slow down adoption as users struggle to master using different apps, skills and platforms in slightly different ways.
from our perspective, this seems like a missed opportunity, as it will only serve to fragment what is currently a relatively small global user-base. leaving these problems aside, let’s focus on what your brand can do to try to capitalise on voice search as part of your overall search strategy.
factoring voice into your search strategy
content optimisation for voice search can be done in three key steps:
- create content that answers voice queries exceptionally well;
- make sure the content is suitable to be turned into an audio output;
- add schema mark-up to help search engines understand what the content is.
let’s talk through each part of the process:
1. create content that answers voice queries exceptionally well
voice searches with an audio output use so-called 0-ranking content to answer queries. this is a term commonly used by search marketing specialists to refer to things like google featured snippets and their equivalents on other search engines – the sort of content that would appear as the “card” on the first page of search results, like so:
in google’s case, the 0-ranking spot can be occupied by content from featured snippets, google knowledge graph and proprietary products like google maps.
for most businesses, featured snippets provide the best opportunities to score a 0 ranking. they account for 40.7% of google voice search answers. the most common sources of content providing featured snippets for use in 0-ranking search results include faqs, guides and other informative resources.
semrush did a study into the common traits of content that earned a spot as a google featured snippet, and their findings were illuminating. here are some of the factors they found the best performers had in common:
- the content is easy to scan through on mobile devices, with an average of 22 headers and sub-headers.
- 83% of url’s are secure (https).
- simple language is used, with an average flesch-kincaid reading level of 7th grade.
- visuals are used to break up the text.
- on average, 33 external link citations are used to support the content’s claims.
- the mobile experience is good, with average google mobile-friendly and google mobile-usability scores of 95/100.
the semrush survey also found that 94% of featured snippets came from top-5-ranking urls.
satisfy these criteria to increase your chance of scoring a 0-ranking and featuring in audio-output search results.
2. make sure the content is suitable to be turned into an audio output
with voice search in mind, we need to ensure content is optimised for use as an audio result. google’s criteria for this are helpfully outlined on the google ai blog:
- information satisfaction: the content must give the user the information they need.
- length: the answer should be concise while encompassing all the essential details needed to answer the query.
- formulation: grammatical correctness is of heightened importance in an audio response.
- elocution: spoken answers must be pronounceable for speech software, with concise headlines and comprehensible summaries.
a good way to organically focus your content around these criteria is to base it around natural language questions. focus on informational queries based on the six ‘w’s of basic information gathering:
- what
- when
- who
- where
- why
- how
these modes of questioning are deeply ingrained in our culture. rudyard kipling wrote about them in his poem i keep six honest serving men (1902), and they continue to define how english-speakers request the most essential information.
3. add schema mark-up to help search engines understand what the content is.
in many cases, 0-ranking content relies on what is taken from webpages tagged with schema mark-up to identify their purpose. this means html tags have been added to the page’s source code to tell search engines what the content is. for more on this topic, take a look at our in-depth guide to using schema markup or visit schema.org
how much priority should search marketers give to voice search?
contrary to some experts’ predictions, voice has yet to overtake typing as the most popular method of inputting search queries, and it looks unlikely to do so for at least another few years.
in light of this, there’s no imperative for search marketers to totally refocus on voice, with the possible exception of those whose websites attract an unusually high volume of traffic from voice search. what we would advise instead is that you consider voice search optimisation as a factor in your strategy.
as we’ve seen, the usage of voice search is evolving. more people are using it in combination with touch input to search for results they would struggle to find via voice search alone. this change is being encouraged by companies like google, whose google home hub chalk can serve both visual and audio results in response to voice search.
we’re inclined to agree with moz founder rand fishkin on how to factor voice into your search strategy.
in the above video, fishkin argues voice search should be seen as additive to, rather than replacing, traditional search.
further, he outlines some smart strategies for using voice search optimisation to acquire traffic. for instance, he suggests answering a simple question at the very start of a piece of content, then going into far greater detail that people may wish to read, e.g. “what’s the difference between slander and libel?” the aim is to get the 0-ranking for answering the question; then get the click-through due to the user wanting to know more.
fishkin also urges marketers to consider how growing adoption of voice search could affect their existing content’s search performance.
some search positions are safer from disintermediation by voice search than others. if your content is of a complex, analytical nature, the risk of it being supplanted by a simple voice search result is relatively low. meanwhile, simpler content covering things like sports scores and quick functions could be at risk of losing visibility.
how to measure voice search traffic
we hope that by this point you’ll have some ideas on how to optimise your content for voice search – from answering questions efficiently to making your prose easier for smart assistants to read.
before you put those ideas into practice, it’s worth considering how you will measure the effect of optimisation on your website traffic. this is relatively difficult to do, as key web analytics platforms including google analytics do not yet offer the capability to see which organic search traffic comes from voice search. we even speculate if this would even be possible since it is not clear if voice assistant devices would be capable of triggering the javascript powering analytics feedback on consumption of website content.
the best solution we’ve come across to this problem is purna virji’s step-by-step guide to testing voice search via ppc.
virji’s method involves using your bing ads or google ads account to identify natural language queries that visitors are using to find your site, where there’s a high likelihood the query has been input via voice rather than typed. we suggest you read through her guide and consider whether you can factor her process into your search performance analysis.
at this stage, most search marketers don’t need to make voice search their top priority. it doesn’t account for as much traffic as traditional mobile search or social, and there’s a degree of uncertainty over how easy the traffic it does attract will be to both measure and/ or monetise.
that said, growing user uptake of voice via smartphones and smart assistants shows this is a subject future-focused marketers should be paying attention to.
our advice is to use your learnings from this article to optimise a single piece of content for voice search. try to get one page to achieve a 0-ranking, and measure the effects on traffic and conversions. this will give you an insight into the most important question of all surrounding voice search: how much does it matter to you and your customers?