Smart coach VUI
year: 2018
company: Myotest SA
project duration: 4 weeks
my role:
- experience design
- user research & testing
- algorithm design
- copy, audio content creation
PROBLEM & CONTEXT - EMPATHISE & DEFINE
~9 min read
Running has never been very fun for Jessica but she knows that being active is good for her health and also that since she is quite the foodie, running is a great way of keeping her weight in check. Moreover running only requires a good pair of shoe and that’s it.
She has been running for a few months already but lately she has been feeling like her usual run are lacking excitement and her motivation is definitely going down. She saw some improvement quickly after she started running but after a while it felt like she was not making much progress anymore. She has recently purchased a smart watch that can track her running session and provide her with an overview of her performance. She likes it but there are a lot of different parameters that the device is tracking and she is confused about what they mean and how to use the data to become a better runner.
How might we help Jessica enjoy herself while running and also learn more about the way she runs and how she might improve ?
PROCESS - IDEATE, PROTOTYPE & TEST
The aim of this project was to create ​a unique in-run experience​ for casual runner. I wanted to ​make this experience entertaining​ by ​challenging ​the user and make it informative​ by delivering ​engaging content​ through ​visual and audio feedback​.
Jessica, our casual runner, wants to ​enjoy herself by​ experimenting new things in a fun way and wants to​ learn something about how she runs so that she can become better at it.​
In order to make the run more entertaining, I thought about bringing in some challenges that would motivate her to try out new things and break up her boring routine. I came up with the idea that we could create a challenge for each of the 10 parameters we measure in real-time and take the opportunity to educate her about what each one of them mean and how she can use them to run faster and injury-free.
For this project, I was working with our very talented in-house designer and we sat down together to define the strategy we wanted to use to develop a proof of concept. We decided that we needed to create a prototype with those challenges and test them with real runners as soon as possible. We wanted to focus on audio feedback as much as possible since we had learned from a previous project that the visual guidance we had developed to use while running presented some usability issues. We also decided to develop the prototype on the Apple Watch mainly for technical reasons.
While the designer was working on the visual language of the experience, I set out to work on building the challenges and the audio feedback content and flow.
I started by investigating which metrics out of the 10 we measure in real-time were more actionable and easier to understand and play with when you are a casual runner. We asked a pool of 50 runners at a local charity race to fill up a survey about their running habits and what metrics was important for them. For most of the runners, time, pace and distance were the primary metrics that were being tracked and paid attention to. None of the advanced metrics that we were measuring were currently used but some of them were stated as relevant or of interest (cadence, step length, power). Thus I decided to focus on one of the easiest parameter to understand and to modify when running: cadence. Cadence relates to the number of steps you are doing per minute while running.
Once I knew on which parameter to focus, I had to work on the flow and the content of the experience. I decided to create a flow based on a regular training session but with the challenge as the main part of the workout. Basically the runner would warm up, then be challenged by the system and finally cool down. For the challenge, I decided to simply ask the runner to reach a specific cadence value (threshold) that would be different than the normal value at which they would be running usually.
While thinking about that, it occurred to me that some runners could have trouble with the challenge and that I also needed to think about strategies for helping them and adjust the difficulty. I created 3 use cases:
1) The runner does not need any help and can achieve the challenge on its own.
2) The runner needs a bit of advice but afterwards he/she is successful in fulfilling the
challenge.
3) Despite the advice given, the runner cannot reach the required threshold so we
have to lower the threshold for the challenge to be successfully completed.
This helped me when creating the algorithm for setting the cadence threshold value to reach. Ideally, the lower the natural cadence, the more increase we should ask for and the higher the natural cadence, the more difficult it would be to still increase it. Thus I created 2 different algorithms, a static one requiring the same increase for every runner and a dynamic one adjusting the increase to the natural individual cadence of the runner. Although the static algorithm allowed for comparison between runners, the dynamic resulted in a physiologically more sound challenge for everybody.
Having figured out the method for setting the threshold value for cadence, the next step was to focus on use cases 2 and 3 and the way to adjust the difficulty of the challenge.
In order to help the runners achieve the challenge, we wanted to provide the runner with advice. The first step was to define a precise timeline for the audio feedback.
I defined various timeouts that would lead to specific recommendations or actions from the system to help the runner. I also devised a strategy to lower the threshold in case the runner would struggle to reach the target value after a given amount of time. Since we wanted a prototype with a way of testing rapidly the experience, we set the length of the experience to be around 5 min.
Having now defined the objective and the flow of the challenge as well as the various ways to help the runners achieve it, I needed to think about the content of the audio feedback. So I investigated what makes a good feedback and how to give it properly.
Feedback can either serve as redirection ​(encourage to change pattern) or reinforcement ​(encourage to repeat pattern). According to best practices, feedback should be:
- Focused on acts, not attitudes
- Goal-oriented
- Multidirectional
- Supportive
- Continual
- Specific
I also decided to do a quick research on the current devices giving audio feedback while running and what the customer pain points were with those. The existing issues with products doing audio feedback and must-have settings that came out were the followings:
- Male/Female voice choice
- Language choice
- Frequency of feedback choice by time or by distance
- VOLUME OF FEEDBACK
- MANAGING MUSIC (PAUSING, VOLUME)
Basing myself on the results from the user research as well as the side investigations I conducted, I created the audio feedback for the prototype according to the timeline of the experience. Then together with the designer, we reviewed the visual design and he worked on the code implementation for the challenge algorithm and the audio feedback in our prototype watch app. He did an amazing job at getting very rapidly a working prototype ready for testing.
OUTCOMES
Before extending it to other parameters, we wanted to try out our prototype with real runners. We asked 5 runners to give it a try and watched them as they did. Despite the last minute fixes we had to do, what striked us was the fact that people were excited and that they came back to us with very positive comments:
« I love the talking. Great when running. I definitely prefer that over watching a screen. »
« I loved the comments about « higher cadence reduces injury risk », gives meaning to what I just did. »
« Love that the voice overlaps nicely with the music playing. »
« The duration between feedback is too long. Would be nice to introduce a verbosity level: talk more, talk less. »
« Can’t wait to learn about other parameters and how these impact my risk of injury and my performance. »
« It said my cadence was 135 steps/min and then 165 a few seconds after. I don’t believe these numbers. »
We learned the hard way that any issues with the threshold algorithm or the real-time measurements would greatly affect the experience and that, way before the settings of the audio feedback would become an issue. We also had to tweak the feedback flow and experiment a few times in order to bring more coherence to the experience. We should have definitely set 2 to 3 KPIs to measure and ask runners about in order to make the decision-making easier for the right strategy to apply as well as to get the stakeholders buy-in. Unfortunately, this project did not turn into a real product. We were not having the bandwidth and the resources at the time to focus on projects that clients did not specifically order.
There are other features we would have wanted to try out such as using sonification to help runners relate to the parameter being measured more closely. For example using a metronome sound for the current cadence and then increasing it at the same time as the runner is increasing his/her cadence until the target value for the challenge has been reached. We also tried making a longer challenge with various parameters to play on, and would have loved introducing gamification elements to share the challenge with friends and push each other. We also wanted to make the runner explore running with higher and lower cadence values and collect feedback on whether he/she would feel better while running. If given the opportunity, we would definitely have investigated those.
I am convinced however that this product would have been a key differentiator and a very enjoyable way of « onboarding » users to our technology and educating the market about its potential. I believe that Jessica would have enjoyed this product and that it would have helped her become a better runner while having fun in the process.
LAST THOUGHTS
I really enjoyed working on this project as I learned a lot about audio interactions and how we can apply them to motivate people and empower them to change their behaviour. User testings was particularly insightful and helped us shape the experience. It was also a huge pleasure to work together with the designer and to be able to learn from him and push the performance of our 2-man team to the fullest.
Thanks for reading.