Usability Analysis of Google Assistant
Usability Engineering, User testing
2019
An analysis of Google Assistant’s efficiency in handling daily life tasks and understanding Indian accents in NLP. The study evaluates its performance, identifies limitations, and proposes enhancements to improve its intelligence, contextual awareness, and adaptability for diverse Indian users.
Role
UX Designer
Duration
4 weeks
Team
2 Designers
Category
VUI Usability Engineering
" Is Google Assistant really smart? "
Design Challenge
The voice user interface of Google assistant had usability concerns especially in understanding the Indian accents and the Indian languages it supported. There was need to analyse the nature of the discrepancies and suggest possible solutions.
Research
A questionnaire was sent around to understand the usage of virtual assistants among people. The representation shows the insights collected.
Task Based Scenario
Participants were given the following tasks “You wake up in the morning and decide to go to a nice restaurant. Find out about the restaurant you want to go to. Then you decide to ask your friend to join you on WhatsApp and finally set a reminder for the event."
Cognitive Task Flow
Based on the analysis of the given task, this was the typical cognitive pathway the users went through. As seen in the task flow there in some amount of friction in the interaction of users with Google Assistant.
Transcript Analysis
Users were given tasks to perform using Google Assistant and they were asked to think aloud while they performed the tasks. A transcript was created and analysed.
Evaluation of Voice Assistant UI Characteristics
Applying Jordan's hierarchy of needs to Google Assistant reveals that the product fulfills a range of human needs, including physiological, safety, and social needs, by providing easy access to information, assistance with tasks, companionship, and facilitating communication.
Interface Update Suggestions
Voice input
Google assistant doesn’t understand names and proper nouns. Pauses in speech of users is considered the end of the sentence. Users’ natural way of talking is not understood easily by Google Assistant. The VUI ( Voice User Interface) needs to be updated in a way that it can accomodate pauses and allow room for thought between sentences.
Intelligence
There is only a very limited use of external sources of information (such as calendar or email) to infer potential actions of interest of the user. When the user sets two reminders for the same time it didn't question if it was going to overlap just to avoid error. Also When the user mentioned setting a reminder for “lunch at 1” it set the reminder for 1 am instead.
Output Optimisation
Except for few tasks like navigation,weather etc. the response given by Google Assistant is not consistent. It gives different results based on the way the person has framed their sentences. Another pain point is not being able to retrieve the history of interactions is definitely a weakness of Google Assistant. An option to retrieve history should be included.The chrome related searches get saved in the history of google chrome, there should be a way of controlling that. a person should be able to have an option to not save their search history on the application.
More Projects →
