to voice search usage is conducted to the following 5 companies:
Siri by Apple Inc.
Alexa by Amazon.com, Inc.
Cortana by Microsoft Corporation
Google Now/Google Assistant by Google
S Voice by Samsung Electronics
Each of these
companies provides their own Intelligent Personal Assistant (IPA) with their
own specifications and features. For this research, only data about voice
search is used. Other features or which IPA is better are irrelevant.
2016 (in millions)
2017 (in millions)
Table 4.1.1 Based on U.S. adults, used
on smartphones. Source: (Hwong, 2017)
statistics may not seem impressive; nonetheless Alexa and Cortana have seen a
growth of over 300% in one year. Moreover, the table above only shows
smartphone usage and does not include smart speakers.
Echo, the smart speaker product line with the Alexa software integrated, has
seen a large growth in the last years. In March of 2017, the Consumer
Intelligence Research Partners (CIRP) estimated that Amazon has sold 10.7
million Echo devices since the introduction at the end of 2014 (Levy, 2017). More importantly,
the awareness has grown significantly among Amazon customers. CIRP states that
while in March 2015 only 20% had heard about the Amazon Echo, by the end of
2016 82% knows the device (Levin & Lowitz, 2017).
generally not open about numbers and amount of users, however in 2014 Google
commissioned a study under 1400 Americans from thirteen years and older. The
results are that 55% of teens and 41% of adults use voice search more than once
every day (Huffman, 2014). Moreover, in 2016 at Google’s I/O
Keynote Developers conference, Google CEO Sundar Pichai announced that 20% of
the searches on android phones or on Google’s mobile app, are voice searches (Sterling, 2016). Moreover, Microsoft
recently stated that Cortana has 145 million users, about a third of the total
users that upgraded to Microsoft Windows 10 (Bourque, 2017).
a research by Creative Strategies (2016), 98% of all iPhone
users have tried Siri but 70% rarely uses Siri. The numbers for Android users
are equal, 4% has never tried Google Now and 62% does not use it often. A
returning issue is that users think it is embarrassing to use an IPA in public (Milanesi, 2016).
IPA’s have thousands of features, in order to understand if the growing
popularity will influence SEO, it is important to find out whether the IPA’s
are used for voice search.
Figure 4.1.2 Source: (Enge, 2017)
According to figure 4.1.2, voice search is used by at least 60% of IPA
users. Moreover, Microsoft’s search
engine, Bing, stated in 2016 that 25% of their search queries in the Windows 10
task bar are done through voice search (Sullivan L. , 2016).
4.2 What are the
main differences with ordinary text search?
The main difference
between text search and voice search is the structure and length of the search
queries. Voice search has a conversational tone while text search usually
contain of short keywords.
4.2.1 gives an example on the differences between users search through voice
and through text.
Flight to Amsterdam
How much is a flight to Amsterdam?
What time does flight MU 0727 arrive?
Weather Kuala Lumpur
What is the weather in Kuala Lumpur like
Weather forecast KL
Do I need an umbrella tomorrow?
Where can I buy Nike Air Max sneakers?
Where is my closest IKEA?
Show me the restaurants near me
How long do I cook pasta?
Cortana did a research in 2015 to search query length, 3 months after they
introduced the personal assistant. The results are as followed:
Average amount of words per search
Figure 4.2.1 (Patel, 2017)
4.2.1 shows that the average length of text search is 2 words, which is normal
since most people know how to search a search engine. More surprising is the
fact that most voice searches are only 3 words, while more was expected. There
could be several reasons for this, people might not be familiar with voice
search yet or the error rate is still too high so users try to keep it short
and simple. It does show that the amount of long tail voice searches is higher
than through text search.
4.3 What is the expected
growth of voice search usage?
Industry experts estimations
equals an increase in usage, when assuming this theory is right; the growth of
voice search relies on the large technology companies. Andrew Ng, former Chief
Scientist at the largest Chinese search engine, Baidu, stated: “I think that as speech recognition accuracy goes from
say 95 per cent to 98, 99 to 99.9, all of us in the room will go from barely
using it today or infrequently to using it all the time. Most people underestimate the
difference between 95 and 99 percent accuracy – 99 percent is a game changer.” (Rebecca Merrett, 2015).
the advertising industry, Rebecca Weeks and Romain Bonnet from the large global
media agency Manning Gottlieb, claimed that by 2020, 200 billion search queries
will be done by voice. Also Stephen Kenwright from Branded3, a UK based digital
marketing agency, went even further by stating that in 2020, 50% of searches
are voice searches (Clark, 2017).
The RBC Royal
Bank even claimed that Amazon’s voice device, Alexa, could be a $10 billion
mega hit in 2020. The investment bank says that purchasing through a device
like Amazon Alexa and Google Home will grow significantly and drive sales up to
$10 billion in revenue.
RBC states that there will be 60 million Alexa devices sold in 2020, making a
total of a little less than 130 million devices worldwide. The bank thinks this
will result in 500 million active users worldwide. They assume a 25% adoption
rate worldwide and 40% in the U.S, with an average selling price of $85 and a
replacement cycle of 2 years, this will result in $5 billion by 2020. The other
$5 billion are on generated sales through the voice search device (Kharpal, 2017).
As can be
seen in graph 4.3.1, VoiceLabs, a voice analytics company, estimated the sales
of voice-based devices in the U.S. to be 24.5 million in 2017. This leads to a
total of 33 million voice-first devices in the U.S (Adam Marchick, 2017). With almost 126
million households in the U.S. and the assumption that there is only 1 device
per household, it would mean that 26% of the U.S. households own a voice-first
device (Statista, 2017). Of course, this number in reality is
lower since some consumers purchase multiple devices per households but it can
still be considered a significant increase.
Figure 4.3.1 (Adam Marchick, 2017)
Gartner Hype Cycle for Emerging Technologies
In order to
see the current stage of the product, the Gartner Hype Cycle model can be used.
The Hype Cycle is set up on 5 different stages:
trigger: The stage in which the technology kicks through and is often
presented to the public, but since it is still in the R&D phase, it is not
available for the public yet.
inflated expectations: The stage where the world gets excited about the
technology, social media and magazines will be full of articles about this
‘next best thing’.
disillusionment: The hype about the product is gone, newspapers will not
write about it anymore, mainly due to the technology which could not meet up to
the hype that it created.
enlightenment: Consumers start to understand the technology and are able to
see the benefits, often due to hard work by the technology providers.
of productivity: The benefits and application of the technology has become
general knowledge, main stream adoption has taken place (Anastasia, 2015).
Figure 4.3.2 (Panetta, 2017)
the Gartner Hype Cycle, virtual assistants are currently at the peak of
inflated expectations phase and likely to fall into the trough of
disillusionment soon. Although other sources in this research show that the
virtual assistants are at the slope of enlightenment stage, since the trough of
disillusionment stage could also be right after introduction of Siri and other
IPA’s and the fact that it was not used by the majority. At what stage IPA’s
really are is a matter of time to find out, however Gartner’s model shows an
estimation of 5 to 10 years until the virtual assistants arrive at the plateau
of productivity stage. This stage is reached when 20 to 30% of society adopted
Technology Adoption Curve
While it is
very difficult to estimate the growth of voice search usage or IPA usage in general,
it is very clear that it will grow substantially in the upcoming years. To
estimate the growth, it is important to see how technological innovations
compared to this have grown in the past. Usually this is not a constant growth
with a certain percentage annually, but a rapid growth once the majority starts
to use it.
this further, the Rogers Technology Adoption Curve (figure 4.3.2) can be used
to find out what the situation is right now and what the estimated growth could
for decades, for example Clippit, the old Microsoft Word paperclip. However, to
explain the table below only the IPA software where voice could be used will be
considered. Moreover, the fact that the IPA and its popularity are based on the
upgrades and the general development of its software has to be taken into
Technology Adoption Curve
Figure 4.3.2 (Amy Hoidas, 2017)
groups are explained as followed:
despite the high error rate, enjoys it and believes in the technology. Does
not necessarily believe in the norms and values of the local social system.
gets lower, they are the opinion leaders that use it because they know it
works. Usually younger people with a higher social status.
usually get the information from the opinion leaders, they need confirmation
on the quality before actually using it. They weigh the pros against the cons
before making the decision.
is more skeptical, starts using it because everyone is using it and they do
not want to miss out. Usually from an older generation.
on the past, unlikely to start using an IPA any time soon. Mainly the older
generation that either does not trust the technology or do not see the need
to use it.
innovators started in the first years of Siri, when Apple introduced the iPhone
4S with Siri in 2011. Many customers tried Siri but because of the high error
rate, they did not use it to full extend. The innovators kept using it. While
the technology grew, features and languages got added but still many customers
were reluctant to use it, the early adopters were there. They bought the Amazon
Alexa devices first and use Siri in their car to navigate or to call
the U.S. society is at the beginning of the early adopters stage, since these
large companies are upgrading and optimizing the software; creating a lower
error rate, adding new features and adding new languages. Besides, there is a
significant increase in the usage of IPA’s, when looking at Amazon’s Alexa with
the enormous sales of their Echo device and the primary and secondary research
that has been done. However, the majority have not implemented an IPA in their
the secondary research, it has become very clear that the usage of an IPA
differs per operating system and device. However, with the introduction of
Amazon’s Alexa and their incredible sales right after the release and the
estimated sales this year, voice search is getting more popular. Although a returning
issue for other users is that they feel embarrassed while using an IPA in
public. With the main purpose of this research in mind, it is interesting to
see that at least 60% of IPA users use it for voice search. This shows that
there is a market for this that has potential to become bigger which will
influence the ordinary ways of SEO in the future.
differences of ordinary text search and voice search are mainly in the human
conceptual way of speaking. Moreover, the majority of voice search queries are
only 3 words long, which is surprising since longer sentences were expected.
The reason for this could be the error rate, which makes people speak very
short and clear sentences, to make it as clear as possible for the IPA software
The expected growth of voice search is hard to
define specifically, yet many industry experts expect the majority of search
queries to be by voice in 2020, mainly because of voice search devices like the
Amazon Alexa, which expects to sell almost 25 million devices in the U.S. in
2017. This in combination with the Technology Adoption Curve (table 4.3.2),
shows that the U.S. market is currently at the early adopters stage and
expected is that it will become more ‘booming’ at a certain time. However,
Gartner’s Hype Cycle predicts that the virtual assistant hype will stagger and
they estimate the mainstream adoption to be in 5 to 10 years.