October 24, 2016 | Author: Patricia Kelly | Category: N/A
1 2016, MindMeld 02 Table of Contents About This Research... 2 Methodology... 2 Executive Summary... 3 Key Research Find...
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
©2016, MindMeld
0
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
Table of Contents About This Research .................................................................................... 2 Methodology ................................................................................................ 2 Executive Summary...................................................................................... 3 Key Research Findings................................................................................. 4 About MindMeld ......................................................................................... 17
©2016, MindMeld
1
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
About This Research This research was undertaken by MindMeld, developer of the MindMeld Platform, an AI-based platform powering intelligent voice interfaces for any commercial device or application. The purpose of the research is to understand: • • • •
Usage patterns of voice assistant technology Demographics of voice assistant users Satisfaction of respondents with voice assistants Demand from consumers for innovation in the voice industry
To understand attitudes surrounding voice-driven natural language technology, MindMeld studied smartphone voice assistant users. This group constitutes the largest single user base of advanced voice interfaces, whose attitudes and preferences are likely to influence the deployment of intelligent voice interfaces in other products and services. To reveal leading indicators of attitudes and interest in voice in general, we assessed: Patterns in adoption Frequency of use • Satisfiers and dissatisfiers of voice assistants • Demographics and technology habits of voice assistant users • •
Methodology The research focused on users of smartphone intelligent voice assistants to infer facts about people’s attitudes toward and usage of voice interfaces in general. An online questionnaire was conducted in January 2016 and completed by 1800 respondents who were: Smartphone users Over the age of 18 • Equal numbers of males and females • Geographically distributed across the United States • •
©2016, MindMeld
2
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
Executive Summary Intelligent voice interfaces have long been a favorite of science fiction, and recent AI advances are finally making them a reality. Already common in smartphone assistants, voice interfaces are now appearing in cars, wearables, smart televisions, and connected home devices such as Amazon’s Echo. This research study of smartphone users assesses adoption, attitudes, and satisfiers and dissatisfiers for intelligent voice assistants like Siri, OK Google and Cortana. We consider these factors to be leading indicators of trends related to voice interfaces in general. Key findings about voice users include: • • • • •
45% adopted within the last year More than 28% adopted within the past six months alone 55% use voice assistants regularly (daily or weekly) 55% indicate overall satisfaction with voice assistants Only 13% indicate dissatisfaction
These findings, combined with those of our Q4 2015 survey, lead us to conclude that 2015 was a tipping point in the adoption of intelligent voice assistants. However, reported satisfaction does not obscure the fact that users expect more from these assistants. Users want to see improvements in areas where technology providers are applying innovations in AI and machine learning, namely natural language understanding (NLU) and accuracy in search and discovery. Similarly, basic speech recognition, while much improved in recent quarters, still leaves something to be desired. Improvements in these areas will drive greater use and engagement by most users (72%). App and device vendors should note the uptick in voice assistant adoption, which may signify a broader consumer desire to discover content and navigate bodies of knowledge by voice. Voice users demonstrate much higher app use than non-voice users, and more than 40% of app users express a desire to see intelligent voice capabilities added to the apps they use.
©2016, MindMeld
3
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
Key Research Findings 1. Voice assistants are popular – used by 62% of smartphone users Intelligent voice assistants (Siri on Apple iOS, OK Google on Android, and Cortana on Windows) are no longer a peripheral smartphone feature. Rather, they are used by almost two thirds of smartphone users in the US. The percentage of smartphone users using intelligent voice assistants has remained consistent from the prior quarter. There is, however, a doubling in the percentage of people reporting using Amazon Echo/Alexa. While this is still a small number, its growth indicates voice assistants ascending to devices other than smartphones.
Which voice assistants have you tried? 45% 40%
38% 36%
35% 28%
30% 25% 20% 15%
9%
10%
4%
5%
0% 0% Siri
OK Google
Cortana
Alexa
Other
None of the above
©2016, MindMeld
4
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
2. Voice assistant adoption cuts across age and gender demographics Adoption of voice assistants cuts across multiple age groups of smartphone users. The highest penetration is among the 18-24 age group and the lowest among those 65 and older. This is consistent with data reported in the previous quarter’s research. There appears to be no gender bias in use of voice assistants.
Which of voice assistants have you tried? (By Age) 60% 50% 40% 30% 20% 10% 0%
52% 41% 32%
26% 9%
41%
33%
32%
37% 25%
13%
age 18-24 Siri
7% 4%
4%
3%
age 25-34
OK Google
Cortana
age 35-44
Amazon Alexa
None of the above
60% 50% 40% 30%
50%
44%
39%
30%
29% 30% 23%
20%
11%
9%
10%
26% 26%
5% 5%
4%
3%
0% age 45-54 Siri
age 55-64
OK Google
Cortana
age 65+
Amazon Alexa
None of the above
Which voice assistants have you tried? (By Gender) 50% 40%
39%
38%
30%
34%
27%
20%
10%
10%
37% 29% 9%
4%
4%
0% male Siri
©2016, MindMeld
OK Google
female Cortana
Amazon Alexa
5
None of the above
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
3. The percentage of long-term voice users is growing Although Siri was added to the iPhone in 2011, the actual adoption of voice is still a very new phenomenon. While our previous survey reported that 61% of users had adopted voice assistants within the year prior to that survey, the current survey reports 45% of voice use starting within the past year, and 28% in the past 6 months alone. The combined data of these two consecutive quarters indicates that adoption of voice assistants may have reached its tipping point in 2015.
When did you start using voice search/commands? 30%
28%
27%
25% 20% 15%
15%
17% 14%
10% 5% 0% More than 3 years Between 2-3 years Between 1-2 years Between 6 months ago ago and 1 year ago
Within the last 6 months
When did you start using voice search/commands? 70% 61%
60% 50% 40%
44% 27%
30% 20%
15%
10%
11%
0% More than 3 years
14% 19% 9% Between 2-3 years ago Q1 2016
©2016, MindMeld
Between 1-2 years ago
Less than 1 year ago
Q4 2015
6
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
4. Frequency of voice assistant use is increasing Respondents’ voice use frequency is growing. A comparison between this quarter’s results to last quarter’s reveals regular (daily and weekly) voice use increasing by 6%, to 55% of users.
How often do you use voice search/commands? 60%
55% 49%
50% 40%
33%
36% 31%
31%
30% 22% 18%
20%
15% 15%
10% 0% Every day
At least once per week
At least once per Less than once per Regularly (every month month day + every week) Q1 2016
Q4 2015
©2016, MindMeld
7
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
5. Satisfaction levels with voice are holding steady About half of voice users report being satisfied with voice as it currently stands, while 39% report feeling neutral. A minority – 13% – report dissatisfaction.
How satisfied are you with voice search/commands today? 45% 39%
40% 35% 30% 25%
27% 21%
20% 15% 10%
7%
6%
5% 0% Very satisfied
©2016, MindMeld
Somewhat satisfied
Neutral
8
Somewhat dissatisfied
Very dissatisfied
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
6. Despite high usage and satisfaction levels, users want to see improvement in voice interface performance Low dissatisfaction does not preclude users from desiring an improved voice assistant experience. Users recognize limitations in their assistants, despite high voice assistant adoption, usage and satisfaction. Even as performance improves steadily and support for new applications grows, some core technology areas remain unaddressed, to the point that they generate user objections. The main voice assistant dissatisfiers are: • Automatic Speech Recognition (ASR): 44% wish their assistants were better at understanding the words they’re saying. This number was exactly the same last quarter. The problem is attributable to automatic speech recognition (ASR). ASR companies are making enormous strides but continue to strive for greater accuracy. • Natural Language Understanding (NLU): 28% wish they could speak more naturally to their assistants. Most assistants have made dramatic improvements in NLU, including gaining some dialog management (DM) capabilities. DM allows users to ask follow-up questions, while the assistant maintains context from the original question. More than a quarter of respondents still desire better NLU performance from their voice-enabled assistants. Compared to last quarter, there is a 1% in this complaint. • Information Retrieval (IR): 27% of respondents wish their assistants were better at giving them the results they ask for. This is about accurately understanding user intent and returning the most appropriate result for the query. For voice interfaces, IR is a different challenge than general web search, because results may need to be presented via voice response or tightly contained in a small screen UI. Compared to last quarter, 1% more people report this complaint.
©2016, MindMeld
9
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
What do you wish voice search/commands did better? Check all that apply. 50% 45%
44%
40% 35% 30%
28%
27%
25% 20%
16%
15%
16%
15%
16%
10% 5% 0%
27% of respondents aged 18-24, and 22% of respondents aged 25-34, also reported wanting more personalized results, compared to 17% for the general population.
©2016, MindMeld
10
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
7. Key performance improvements would increase most users’ voice interface engagement Similarly to last quarter, 72% of respondents report that improvements in core functionality would increase their use of voice assistants.
How often would you use voice search/commands if the problems you identified didn't exist? 80%
72%
70% 60% 50% 40% 26%
30% 20%
28%
17%
14%
14%
10% 0% I would use voice I would use voice I would use voice I don't think my as the default all most of the time somewhat more usage pattern the time than I do now would change
I would only use Would increase voice when I had use (sum of first to 3)
©2016, MindMeld
11
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
8. 48% of those who haven’t tried voice “just haven’t gotten around to it” Looking into why some smartphone users have not tried voice assistants yet, the leading reason by a large margin is “I just haven’t gotten around to it yet. This suggests that negative perception of voice assistants is not the main barrier to use. In fact, from last quarter to this quarter, “I’ve heard [voice assistants] don’t work well” decreased by 4%, suggesting that consumers are gaining awareness of the rapid advances in speech technology.
Why haven't you tried voice assistants? Check all that apply. 60% 48%
50% 40% 30%
25%
20% 10%
14% 7%
6%
4%
0% I've heard they I'd feel weird don't work well speaking to a machine
I don't think I just haven't They don't work they would be gotten around with the apps I useful to it like to use
Other
©2016, MindMeld
12
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
9. Hands- and vision-free interaction is the leading reason for voice use When asked why they use voice search and voice commands, 61% say that voice is useful when their hands or vision are occupied. This reason has a 31% lead on its closest runner-up, “They get me results faster.” As wearables and other screen-less and small-screened devices become more popular, the percentage choosing voice because “It’s hard to type on certain devices” is likely to increase. Interesting to note: respondents 18-24 were much more likely to select “They’re fun/cool” as a reason for using voice (38%). Also, people who use voice search and commands every day often do so because “They get me results faster” (43%).
Why do you use voice/search commands? Check all that apply. 70% 61% 60% 50% 40% 30% 30%
24%
22%
20%
12%
10% 1% 0% They're fun/cool
©2016, MindMeld
They get me results faster
It's hard to type I can avoid They're useful on certain confusing menus when my hands devices or vision are occupied
13
Other
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
10. Voice is primarily used at home, and secondarily in the car While applications with voice interface capabilities are mainly found in mobile and automotive, the home dominates in use of voice assistants. This suggests interesting potential developments for smart home assistants and may tie in with the growth we’re seeing in the adoption of Amazon Echo, one of the first voice assistants to target the home.
Where do you use voice features the most? 45%
43%
40%
36%
35% 30% 25% 19%
20% 15% 10% 5%
3%
0% At home
©2016, MindMeld
In the car
On the go
14
At work
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
11. Voice users are also heavy mobile app users Most voice assistant users are also prolific mobile app users and vice versa. Correspondingly, the majority of people who do not use apps have not tried voice. This suggests that mobile app vendors could improve engagement metrics by incorporating voice in their apps.
13%
Uses shopping apps (eBay, Etsy, etc.)
Uses video apps (YouTube, Netflix, etc.) 12%
13% 10%
Uses music apps (Pandora, Spotify, etc.) 13%
10% 77%
77%
75%
Have tried voice
Have tried voice
Have tried voice
Are open to trying voice
Are open to trying voice
Are open to trying voice
Don't think they'll ever try voice
Don't think they'll ever try voice
Don't think they'll ever try voice
Uses travel apps 10% (Kayak, TripAdvisor, etc.)
14%
Uses connected home apps (Nest, SmartThings, etc.)
Uses local services apps (Yelp, Uber, etc.) 14%
19%
15%
15% 71%
71%
71%
Have tried voice
Have tried voice
Have tried voice
Are open to trying voice
Are open to trying voice
Are open to trying voice
Don't think they'll ever try voice
Don't think they'll ever try voice
Don't think they'll ever try voice
Uses none of the previous types of apps 14% 32% 55%
Have tried voice Are open to trying voice Don't think they'll ever try voice
©2016, MindMeld
15
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
12. Music app users are particularly eager for the incorporation of voice When asked if they would like the apps they use to incorporate voice capabilities, more than 40% of users across of the board in most common app categories indicated an interest in voice. This was true for users of local services (e.g. Yelp, Uber), shopping, travel, and video apps. Meanwhile, 50% of people who use music apps said that they would like to have voice functionality in those apps.
In which type of apps would you most like to use voice commands or voice search? Check all that apply. 80% 68%
70% 60% 50%
50% 40%
41%
41%
41%
40% 30%
33% 20%
20%
23%
28% 22%
19%
26%
10%
10% 0% Local services Shopping apps Travel apps apps
Music apps
Uses said type of app
Video apps
Connected home apps
None of the above
Total population
©2016, MindMeld
16
Intelligent Voice Assistants – Q1 2016 User Adoption Survey Results
About MindMeld MindMeld powers the next generation of intelligent voice interfaces for any device or application. MindMeld delivers human-like understanding of naturally spoken queries for large, complex content domains. Available in both complete cloud or hybrid-cloud solutions, MindMeld is built on an AI foundation for scale and reliability in natural language processing (NLP) and information retrieval (IR). MindMeld offers flexibility to support customer-specified features, content domains, speech recognition engines, and devices. Media and entertainment, consumer electronics, IoT, retail and ecommerce, mobile apps, and automotive are among the industry segments benefiting from MindMeld’s capabilities. MindMeld has been widely recognized as a leader in the field of voice-driven computing and was named by Entrepreneur magazine as one of the "100 Brilliant Companies" of 2015 and by MIT Technology Review as one of the world's "50 Smartest Companies" of 2014. Founded in 2011 and based in San Francisco, MindMeld is backed by leading technology companies including Google, Samsung and Intel. More than 1,600 developers and companies use MindMeld, including Fortune 500 companies.
MindMeld 251 Kearny St, Suite 400 San Francisco, CA 94108 www.mindmeld.com
[email protected]
©MindMeld. All rights reserved. All company, product and brand names, logos and trademarks are the property of their respective owners. No parts of this report may be published without prior approval of MindMeld or attribution to this report. This document has been prepared in good faith on the basis of information available from the conducted research without any independent verification. The information in this document is provided "AS IS" and MindMeld does not guarantee or warrant the accuracy, reliability, completeness or currency of the information in this publication nor its usefulness in achieving any purpose. Readers are responsible for assessing the relevance and accuracy of the content of this publication. MindMeld will not be liable for any loss, damage, cost or expense incurred or arising by reason of any person using or relying on information in this publication. Products may be identified by proprietary or trade names to help readers identify particular types of products but this is not, and is not intended to be, an endorsement or recommendation of any product or manufacturer referred to. Other products may perform as well or better than those specifically referred to.
©2016, MindMeld
17
