This week Tim Butler of Innovation Visual led a webinar in association with Liverpool University to talk about voice search and how its continued growth will impact marketing in the future.
You can watch the full video of the webinar below:
Before the webinar began, attendees where asked to fill in a poll, offering us an interesting insight to how many people are actively using voice search and what voice search devices people are using. The results are shown in the graph below.
Interestingly 25% of those tuning into the webinar had never used voice search. Over 70% of viewers had used their smart phone to conduct a voice search, making this the most popularly used device, of those who took part in the poll.
The concept of voice-controlled computers certainly is not new. Using voice to interact with computers has in fact been around for some time. Google's first voice-powered search was actually a phone service. You would dial a number and then speak your search query. The result would then be read back to you.
App-based Google voice search arrived in 2012, in a format you would recognise today.
The Siri App arrived earlier than Google's, in February 2010. To begin with, it was a standalone app and then in May 2010, Apple bought the technology. It took until near the end of 2011 before Siri was integrated into the iOS operating system, as we use it today.
Microsoft's digital assistant, Cortana, was even later to the party, being first demoed in 2013.
The device that changed the game, however, was Amazon's Echo, using the Alexa digital assistant, announced in November 2014.
The technology that we are using today for voice search is quite new, although the concept is not and this is due to the way it works.
When you say, ‘OK Google' or ‘Hey Siri' and then ask something, you might be talking to your phone, but it is simply a conduit for the information, a microphone and a speaker because the computer power required to answer your simple question is immense.
What held back early voice technology adoption was the need to use very specific commands and speak very clearly. You had to know the commands and you certainly would not want to have had a strong accent. It was not easy to use and so it was not better than the alternative of typing a query in.
This has been overcome by computing power, Artificial Intelligence and Machine learning. Your simple request is sent to servers where your voice is translated to encoded text, which can then be used as the search criteria with the results going back through voice encoding before being sent back to your handset. This whole process happens within a few seconds. It is impressive.
However, the AI means that the translation of the audio to text is now much more complex including data such as location, previous search you may have made and even emphasis in your voice. It is this power and complexity that has allowed computers to really ‘understand' what you are asking – something that you take for granted every time you have a conversation with a human being.
The devices used for voice search are now very familiar to us. Amazon's Echo and Google Home are called voice first devices. They do not have screens, so it is purely an audio relation we have with these devices. Although, this is set to change, as Amazon now have an Echo with a screen!
Smartphones, running iOS or Android, have voice control at their core, as well as Apps that are voice controlled.
Apple's CarPlay and Google's Android Auto are evolutions of your phone and car's technology.
Voice search goes beyond just the devices above and extends into desktop devices with Microsoft, Apple and Google Chromebooks all having voice control/search built in.
The Google Chrome browser supports voice search too. It may, in fact, be these devices that see the greatest growth, as people stop being so self-conscious about asking their computer a question out loud in the office.
Voice control goes even further than these devices and extends into what might be referred to as dumb devices, meaning ones that use a voice interface but are very restrictive in their functionality.
The car is interesting because Jaguar Land Rover and SEAT are putting Alexa technology in their cars and the CarPlay / Android auto links to the inter enable smart search.
If you think about the direction that car technology is going in, with the introduction of autonomous vehicles being continuously talked about, then if you are not driving a car you can be searching online. Why do you think then that Google is so interested in creating autonomous cars?
There are other non-internet connected voice controlled devices such as the famous robot dog from Sony, which you could control by voice.
Voice control is something we find very appealing as humans. It is not these dumb devices that we are explicitly talking about, but the more you use voice control the more it becomes the norm.
The graph above shows data from a study in the USA conducted in 2016. The most common reason to use voice search is multi-tasking. You can search while you drive, cook, eat etc.
The faster results are interesting. Voice searches tend to jump straight to the first result, bypassing the Search Results page, so people do not have to choose one of the results to dig deeper into. Busy people can avoid menus and choice overload, and get straight to the point.
Something many of us will have experienced is the difficulty of typing on devices. It all relates to convenience and that is why people are using voice search. It is easier and quicker than it is to type the same search query.
We can speak 150 words per minute. We can only type 40 words per minute. People use voice search now because it is easier for them.
With products such as Alexa & Google Home being in your Home focused, it is clear these are fulfilling a need from the data above.
Searching on-the-go and in the car, makes sense. You are on the move and need answers quick. You might need an address of somewhere you are trying to get to or want to know the nearest place to get a coffee. This is the driver for a lot of these local searches because people are on the move.
Perhaps the surprising statistic is the lack of use at work, but this is probably because we are still a bit self-conscious about saying ‘Hey Siri' at our desk with our colleagues listening.
There is certainly an argument to suggest this will change, and as soon as people start losing that inhibition, the uptake at work could well be rapid.
The growth of voice search has been dramatic. The Amazon Echo, with Alexa, has only been around for 3 years. The data for the whole of 2016 shows:
The AI technology behind Amazon's Alexa and Google Assistant means voice search devices are now very useful and their sales reflect this.
This growth rate is massive and it certainly does not look like slowing in the immediate future.
If anyone bought an Echo on Prime Day on the 11th July this year when there was a big discount you are not alone. Amazon sold 3.3 million Echo devices on that one day!
This is a clear example of people's willingness to introduce voice first devices into their homes.
64% of all website visits start at a search engine. It is even more important if you want your business or product to be found by someone who is looking for what you do – they Google it.
We have seen the growth in voice search, so this will impact what we call SEO (search engine optimisation). Below are a few of the challenges now facing marketeers:
Voice search means no screen results. So, if you are number 2 in the results and Alexa only reads out the number 1 result, being second would really suck.
The first result is often a structured snippet shown on the results page, which is then read out.
The structured snippet appears in a box on the search results page either below the paid results (if there are any) or on the right in the knowledge graph.
The example below is from a client of ours, Red Paddle, for when you search about paddle boarding in winter.
This means that when you ask your Google Home device, ‘What should I wear paddle boarding in winter?' you will be told that according to Red Paddle Co ‘if you plan to just go for a recreational paddle' etc.
Just think for a moment what this means for marketing. You can have your brands answer a search query read out by voice devices, but only if you have a structured snippet as an answer to that query.
If a structured snippet appears for a term you want to be found for, you need to ‘own' that structured snippet.
Search engine optimisation used to be about picking some keyword phrases that your target audience were searching for, and then optimising for these, as you knew that if you ranked highly you would get traffic and sales.
Now people are talking to and putting much more detail into their queries, so you cannot just optimise for phrases, you need to optimise for content themes.
People are asking questions and so your content marketing needs to be structured around this. People have needs and the AI device is looking to fulfil those needs.
There are various different ways to find out what your audience is looking for:
The graphic above shows the results from a tool called AnswerThePublic, which is a great way of generating content ideas around a theme.
Most times when you search you are logged in, even if you do not realise it. Your long-term search history and the history of sites you visit is taken into account in the search results you are presented. The AI learns from your individual behaviour.
The AI gives search context by remembering the ‘conversation'. For example, you could try the following search query:
Ask, "What is the weather in Barcelona in February?"
Then
"How much are flights?"
What it will do is return the results for flights to Barcelona from your nearest airport for the month of February.
This is what makes voice search and voice assistants useful now as they can understand conversational voice, not simply rigid commands. It is making the interactions more ‘human'. It is important to consider however, that the AI needs to understand meaning in order to connect information, and this is where schema.org comes in.
The meaning of data is delivered in a large part by structured data and this is defined by the Schema dot org standard.
It is a system for marking up content on web pages that allows the search engines to understand in a structured way what the text in the code ‘means'.
Example:
From the example above you can see that there is a telephone number and it is for Customer Service, and AI will easily be able to identify the information on the page.
Given what we have seen from the data, it is no surprise that lots of the voice searches are related to location - or local searches, as we call them. For example, ‘Find a coffee shop near me', ‘where can I buy flowers', etc
People are not looking around, they are simply turning to search to find things faster.
The person searching could be in the car and on the go. If you are driving and you want to find somewhere to get a coffee, you can do a voice search with Siri and from the answer get the directions programmed into your sat nav immediately.
Localised search is driven on Google and Bing by their local business pages - if you are a business are you sure you have claimed these and that you have optimised them to be found by adding content and linking to them?
One of the greatest challenges is the misspelling/mispronunciation issue. The word accuracy rate of voice devices is now around 95% when only 4 years ago it was less than 80%. However, it is still not 100%, meaning it is not perfect and can still struggle with certain search queries.
For example, Yves Saint Laurent handbags used to be impossible to find via voice search. Changing your current brand or company name to be found more easily through voice search probably is not an option, but if you are thinking of a new business product or brand then it is probably worth considering how easily it can be found through voice search.
Word of warning, be careful about asking Cortana about ‘Jason Bourne' films when you are in the office. The results can be quite different.
People want convenience and one of the results of this is disintermediation.
Google's main mission is to organise the worlds knowledge, but they act as gatekeepers to what you are shown.
If you ask Amazon's Alexa for pepperoni pizza it will order it from Domino's because the company has a deal with Amazon. This can reduce consumer choice and lead to monopolies developing.
For Amazon, the business' strategy is to own the sales channel. They used to be just an online shop, not a technology company, however, the company is now a sales search engine in its own right. Initiatives like Amazon Prime and one-click purchase, have made it even easier for people to purchase goods. You can even order products through Alexa now, demonstrating the power of Amazon through people owning their Echo devices.
Voice search is definitely important because of its growth, but it is just a proportion of overall searches. Typed search is not decreasing, so is still hugely important.
Voice search is a new stream of searches, providing a means of driving additional traffic to your website.
If you are following the methods to improve your website for voice search you should be in line with overall SEO best practices of having great content, using structured mark-up, having the well-constructed code and getting people linking to your site because it is a great source of information.
So, where is the future of voice search heading? Here are three things to think about:
Voice search is growing, so make sure optimising for voice search is part of your digital marketing strategy.
If you would like to find out more about Voice Search and how you can make it work for your business then contact our team, who will be happy to help answer any of your questions.