Before Improvement After Improvement
the most important. In addition, one-third of non-users pay attention to security issues such as personal data, and this ratio is significantly higher than that of users.
When the aforementioned obstacles are improved, it may affect the non-users’
willingness to adopt a voice assistant. For the fresh users, they are willing to use information-based features such as checking traffic and weather, as well as seeking information through. They are also willing to complete routine work through hands-free voice interaction, such as playing music, making phone calls, and setting alarm clocks and calendars.
In addition, although the proportion of business conduct is low, 35% of current non-users are willing to use the voice assistant to research products before purchasing.
0% 10% 20% 30% 40% 50%
Couldn't provide more life-oriented services Couldn't be able personalize the experience for me
Couldn't understand my diction and accent Couldn't be able to contribute more based on original demand e.g., volunteer with additional
relevant information
Couldn't ensure the privacy and security of personal data
Couldn't respond correctly and quickly Couldn't be able to hold a reasonable conversation
Obstacle of Non-User
Figure 5-10 Obstacle of Non-User
Figure 5-11 Features That Non-users Are Willing to Try
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
Buy products Tell jokes Find a place to eat Set timer Check the recipe Send message Control Internet-of-Things-enabled devices Check/set calendar entry Set alarm Make phone call Research product before purchase
Play music Seeking information Check weather Navigation
Features of Non-User
6 Recommendation
According to the results of this study, 82.5% respondents already know the voice assistant, showing that the voice assistant is widely known in Taiwan. Among the respondents, 62.5% have used voice assistants, and the usage frequency of Siri and Google Assistant has increased significantly compared to the 2018 study. In addition, new competitors have joined the market, showing that voice assistants are entering the growth stage in Taiwan.
At the growth stage, competitors enter the market and create more competition in this market. To strengthen the competitive strategy of the product is quite important.
6.1 SWOT Analysis
In order to maintain market growth as much as possible, the main brands of voice assistants must have new product features that meet market demands or create new users through market segments, which in turn can dominate the market. The following will conduct a SWOT analysis for Google Assistant and make corresponding recommendations for the results.
Table 6-1 SWOT Analysis of Google Assistant
Strengths
● Google has a lot of data: most people in Taiwan use Google Search and rely on the results of Google Search.
● Android mobile phone system has a high penetration rate in Taiwan
Weaknesses
● Lack of Chinese speech recognition ability
● Limited function on the market (Unable to make voice shopping or voice booking a restaurant, etc.)
● A higher potential market than Siri (when improved, a higher percentage of people want to use it)
● Google Assistant users are more purposeful than just having fun
● The new features of the various developments of the English version of Google Assistant are also in line with the needs of the Taiwan market.
Opportunities
● Market awareness and acceptance of voice assistants are getting better and better
● When voice recognition is better, some non-users are also willing to use
Threats
● International brands start to provide localized voice services
● The localized voice assistant system begin to enter the market
6.2 Recommendation - Voice Recognition
Based on quantitative survey results of this study, half of the voice assistant users on the market claimed that the assistant could not understand their diction or accent, and could not respond correctly or quickly.
Voice recognition is the foundation of the voice assistant. Users not only expect voice assistants to listen, but also understand their needs, and then provide customized information or services according to their needs. As Julia Hirschberg and Christopher D.
Manning indicated that speech cognition requires a large amount of linguistic data
collection and analysis. Internet users in this market are accustomed to find the information they need through search engines. Google not only has a lot of data but also understands the needs of market users. Combining these two advantages will help Google Assistant understand user needs.
In terms of a richer understanding of the structure of human language and its deployment in a social context, Google is based in English and tends to be weak in Chinese. Google can work with local research institutions to strengthen local language understanding and strengthen localized speech recognition capabilities.
For instance, the Industrial Technology Research Institute voice technology team has been researching DNN (Deep Natural Network) based speech recognition technology since 2014. Two different DNN models were trained for over two years: FNN (Feedback Neural Network) and LSTM (Long and Short Term Memory). Compared with the traditional GMM model (Gaussian hybrid model), ERR reached 39.4% and 57.4%, indicating that the speech recognition rate of domestic R&D has greatly improved.
Google can work with the organization to more effectively improve the learning of local language structures.
6.3 Recommendation - Practical Features
To expand the user base, Google Assistants can provide more features or localized information, meet the regular needs of consumers, and build customer trust in the brand.