Voice Assistant Study
Discover the Story
Soundmind is a company, which supports senior citizens. It customises and manages Alexa (Voice Assistant) to deliver high quality care. When we started to work for Soundmind, the tool was already used, however, there were still many questions to be asked.
I worked with a great team: Yiqiao (Karen) Li and Ran Hao. Together, we found out why currently people use Voice Assistants and what are their expectations for Voice Assistants development. In our research we provided recommendations for Voice Assistants designers aiming to make sure that the future Voice Assistants meet people needs and wishes.
Duration of the project: 14 weeks, January - May 2020
Research methods: Interviews, Diary Studies, Co-Design Workshop
My role: We divided our roles equally - I have participated in all research stages. I took a leading role as a liaison with our stakeholders - client, professor Pamela Pavliscak and our research participants.
At first, we wanted to investigate how seniors are using devices provided by Soundmind. However, this appeared not to be possible - we had a very restricted timeline and Soundmind couldn’t provide us in this particular moment with access to the research participants. We had to think how to tweak our research, so it will be useful for our client, but also feasible for us to do!
Eventually - we came up with the main research question:
How Voice Assistants, in any form - affect people’s day to day life?
We wanted to look into the current use and perception and eventually investigate the future use of voice assistants.
In our study design, we explored 3 different ethnographic research methods: Interviews, Diary Studies and Co-Design Workshop. Each of them provided us with different insights and naturally guided us to the next research step.
First, we recruited our study participants. We used the snowball sampling and asked our friends and friends of friends for help. We were looking for people who use Voice Assistants on a daily basis and who are 21 years old and above. Eventually, we found 3 Voice Assistant users who were interested in our research and were willing to be interviewed.
It was already a good start! Before creating an interview script, we dreamt that we can combine the interview with an observation. However, two of our participants were located in a different city than us. We had to perform the interview online and this interfered with our research design idea. However, we already knew that we can add an observation component to the next phase of our research (diary studies).
Before the interview, we asked our participants 3 questions:
Which Voice Assistant do use (e.g. Siri, Alexa, Google Assistant)?
For how long are you using it?
How experienced are you in using the chosen Voice Assistant? (1-5, 5 = very experienced)
This allowed us to prepare for the interview and tweak our questions. We asked questions aiming to find out how our participants started to use their Voice Assistant, what they use it for, what are high and low experiences related to it and finally what type of relationship our interviewees have with their Voice Assistant. We used an affinity diagram to analyze our data.
Based on the interviews we have created an empathy map. This tool helped us to create our major findings. We have categorised the comments from our users into 4 major categories: Thinking, Feeling, Saying and Doing.
What did we learn?
People find Voice Assistants convenient and reliable but are not emotionally attached. There are several research papers and articles that addressed how people perceive voice assistants as real life companions (eg. MIT Technology Review). However, based on our interviews, people drew a clear line between Voice Assistants being just a tool and an actual person. As one of our participants said: “[Alexa] is like ‘personal assistant’ - is when she is needed, disappears when it is not needed”.
People remain skeptical about their privacy and data protection. Despite the fact that people are using it on a daily basis, they are very aware of the potential privacy issues. Participant no 1 mentioned during the interview: “It is a weird feeling to talk with someone who doesn't exist. How it works is somehow intriguing and off putting at the same time”.
Efficiency is the major reason while people use Voice Assistants. All participants said that Voice Assistants allow them to save time and multitask. They noticed that they save time on typing as they can just communicate their thoughts by speaking. “It saves me more time, basically I can just tell Alexa to turn on the lights” said Participant 2.
Presence of others has an effect on the use of Voice Assistants. One participant uses Alexa for information (e.g. news, radio) or quick tasks (e.g. checking weather, time) when alone; while with grandchildren around, the use of Alexa changes. Another person only uses a Voice Assistant at home and almost never speaks to it in public. As Participant 3 said: “I almost never speak to Google in public, because it’s embarrassing for me… it’s weird”.
In the Phase 1 of our research - Interviews, we have discovered that Voice Assistants are efficient and convenient, but are not considered to be companions.
We aimed to investigate the social aspect of using Voice Assistants in the Phase 2 of our research. To do this, we decided to go for the Diary Study research method. We hoped it will allow us to observe how users interact with their Voice Assistants.
Once again, we recruited our participants. The selection criteria were the same: we needed people who use Voice Assistant on a daily basis and who are older than 21 years old. Luckily for us, all participants who took place in the first phase of our research agreed to participate in the Diary Study.
This time, we asked our research participants for 5 days to record their interactions with their current Voice Assistants, describe them as persons and predict how future, perfect Voice Assistants, might look/behave like. We used Indeemo software to do this.
Eventually, we had to extend the deadline for the study to 7 days. Some of our research participants were travelling. This type of study requires also more from people who agreed to take part in it and we didn’t offer any incentives, which might also influence the motivation of our participants. However, every day, everyone was receiving a reminder to keep on a track with uploading recordings. I also provided personalised comments to all uploads - just to make sure that our test participants feel appreciated and keep being motivated. Eventually, all but one, submitted all recordings and uploads. One person chose to use email and send us voice messages. This created a slight challenge when analysing the data, but eventually we overcame it!
We prepared the controlled tag list and we analyzed the data, tagged and described in the Google sheet. Based on this analysis, the tags were merged and added to Indeemo - this allowed to create a tag cloud and observe the general trends.The tag clouds and our descriptions allowed to identify major research findings.
Based on the outcomes of the Diary Study, we created a Persona of the Future Voice Assistant. Please meet Alex!
Alex is the personification of how Voice Assistant should look in 10 years' time. I have to say that it was great fun to create him. In a way, it surprised me, how people want the Voice Assistant to be more like a companion than the assistant itself. But we will have a detailed look into our findings in a section below.
What did we learn?
Interactions with voice assistants are mostly transactional oriented. Our research participants used mostly their Voice Assistants to perform routine tasks. However, they anticipated a shift in the future and looked forward to having conversations with their Voice Assistants. These are examples of the typical commands given to in this case Alexa: “Turn on the light”, “How’s the weather”, “Play morning music”.
People are also adapting to Voice Assistants. People are always mindful of speaking to Voice Assistants in a different tone of voice. One of our research participants described her relationship with Siri in the following way:”Siri is more like a grandma who doesn’t speak fluently English. You can’t fully converse with her the way you can with Alexa”.
People expect Voice Assistants to be more interactive and intellectual in the future. They want them to be affective, conversational and understanding. They would love to customize and personalize them. Finally, they hope that the future Voice Assistants will be interconnected within IoT.
Thanks to the interview and diary study phases of our research, we already had a pretty good set of data. But now - we were wondering how we can present it to our colleagues in a way which will be interesting for them? Additionally, we already knew what people expect from Voice Assistant in the future. But we were not sure what such an ideal Voice Assistant can do and how it can participate in its owner's life. That’s why we invited our colleagues to participate in the co-design workshop. We provided the following prompt:
“We would like you to think about a typical day at home/work/school - do you remember what it looks like? Now, imagine this in 10 years time - in 2030, what would this day be like with your future voice assistant? Can you walk us through it?
In this activity, each of you will describe a day in the future with VA using post-its notes in any way you like. Then we will regroup and paint a picture of a collective perfect day. What has changed (or unchanged) compared to today?”.
Not everything worked out the way we wanted. We expected that our participants would draw the day, but the majority of them preferred to write it down. Eventually, during the workshop, we didn’t create our common day. We did it within my group and presented to the class during the final presentation.
Based on the outcomes of the Co-Design workshop, we created a User Journey.
It summarizes all comments coming from our colleagues. According to them in 10 years time, Voice Assistants will be a part of our daily life. We will use them almost in every our activity - Voice Assistants will wake us up, will order food for us, help to dress and support us in our working and social life. Eventually, they will help us to fall asleep.
What did we learn?
Our research participants wanted their Voice Assistants to be with them throughout the day. They wanted to wake them up, help them to get dressed, support them at work and maintain their well-being.
Voice Assistants should know the owner, her/his needs and cater to them.
The privacy issues which were appearing in the Interview phase of our research were not mentioned. This was a surprising outcome for us and we definitely want to investigate it further.
Voice Assistants are expected to be part of IoT. They shouldn’t be limited only to the phone or a standalone device.
Voice Assistant as an assistant
Easy to use, user-friendly and accessible
Interconnective, part of Internet of Things
Cater to people’s needs, especially surrounding their daily routines and time managements
Voice Assistant as a companion
Empathetic, patient, understanding
Understand the context of conversations and natural language
Support owners’ wellbeing
Entertaining and fun to talk to
Our research provided us with some answers and the general guidance for the Voice Assistant designers. However, after each phase of the research, we still had some questions left, which we, provided having enough time, we would love to investigate.
How much privacy is important for users? How can creators of Voice Assistants build trust?
Courtesy while using Voice Assistant and its effect on children
Children mimic their family members. Does the way their guardians talk to the Voice Assistant has an effect on a child's behavior?
Use of Voice Assistant with screen and interface
What is the perception of a Voice Assistant with a screen? Do users need/expect it?
Use of gestures while interacting with Voice Assistant
Is there a need for a Voice Assistant to respond to the gestures? Do users need/expect it?
I don’t appear in any of our photos, but my official Photographer role can be summarised in this way :) :