Augmented Reality (AR) provides a compelling experience to train new Model 3 customers about their car.
How can Tesla train 3x number of customers about their new vehicles with limited time and sales people? With the constraint that the solution has to be some kind of mobile app, I deviced an interactive AR onboarding experience that lets customers learn about their car before they receive it.
Click the image above for an interactive prototype.
An example of the second design using a universal card menu.
Tesla has a problem. Too many people are buying their cars and they are not getting the premium experience that previous Tesla customers enjoyed. With the release and subsequent production increases of the Model 3, Tesla is now making nearly 3x as many cars they did just a year ago.
"The delivery of the cars is where the investment is needed. We need to deliver three or four times as many cars." - Elon Musk
As they ramp up production, one of the curious bottlenecks which has come to light is the new customer onboarding and learning experience. Tesla can no longer afford to give each customer a 30 minute session with a salesperson explaining the features of the car. Tesla is actively experimenting with alternative solutions to this problem but so far has only left users unprepared and confused with their new car.
"No offense, 5 minutes is totally inadequate, especially for BEV newbies transitioning from a 20th to 21st century vehicle." - Tesla forum
It may well be the car of the future, but most of its customers are used to gasoline cars of the present, which function quite differently. A little direction at the beginning can make a big impact in improving the user's experience with an innovative new product.
To begin, I wanted to establish what Model 3 reservation holders were expecting from the delivery process. I also talked with those who had received their cars to see what their experience was. I received 30 survey responses and conducted 4 interviews.
One of these members is the president of the Ontario Tesla Owners Club and he provided insights into trends he saw from an even greater number of users. A few highlights became immediately clear:
- New model 3 owners needed more time with the car before delivery. There was just not enough time to rush people through in a timely manner that met both business and user needs.
- Most people do not read the provided materials Tesla sends via email.
- Videos produced by Tesla are helpful but buried on the Tesla support site so few interviewed were familiar with them.
- 60% percent of those surveyed would prefer an interactive app over a live salesman.
- 100% perferred either an interactive AR app or a salesman over other methods such as instructional videos.
- 90% percent have a compatible device for AR.
- Common themes of concern:
- Charging and battery maintenance.
- Center touchscreen.
- Construction quality of the car itself.
"People need time with the car at their own pace." - Ontario Tesla Owners Club
Initially, I had thought to use vision recognition and machine learning so that a new customer could point their phone at the actual car during delivery to learn about their car in an intimate and interactive way. After our initial research, it became clear they needed a compelling solution which the user could interact with before delivery, so that the user was familiar with the car before receiving it. The solution is to project a highly detailed interactive 3D model of a Model 3 in AR using ARKit on iPhone and ARCore on Android.
The Case for AR
You may be wondering, "Why not use traditional media like videos and text to teach people about their new car?" It's all about engagement. If people don't read the instructions, those instructions aren't helpful.
Why do car dealerships put their vehicles inside fancy showrooms? The experience, the intimacy, the engagement. It makes an impact. It makes the car approachable, attractive, and intimate to prospective buyers. Similiarly, AR allows for a far more attractive and personal learning experience for reservation holders.
User Flow and Wireframes
Based on our research, I deciced to focus on key features necessary to driving for the first time, learning about vehicle construction, and battery maintenance.
Detailed instructions on the center touch screen is intentionally left out, as I belive some kind of hand-off procedure from the user's mobile phone to the car computer would be more ideal in explaining those features. That will be explored in the future.
Prototyping and Usability Testing
The design went through several iterations, from sketches on paper to interactive prototypes built with Sketch. Testing these desings proved tricky, as a static clickable mockup does not accurately imitate the AR experience. Often users would ask for a "back button" to change subjects and at the end of a session I would explain you would just physically move and point at something else. I found other AR apps and videos useful for giving context.
Once they understood the basics for how an AR app might work, I focused the testing to see which overlays and info cards would work best. In the future I want to try different testing techniques such as what Apple suggests in their recent WWDC video.
Above I have a few variations on displaying info cards in different settings.
This screen is a good example for what I initially ended up with. The menu on the left with the check marks shows a user how far they are in a particular learning path. As they locate and find the little dots and place them in side the rectange details emerge in the form of cards and the check boxes are added to show progress. On the right hand side there are several buttons, one lets you place a new digital car, the other lets you take and share a photo of your digital car with friends, and the toggle on the bottom right lets you switch the "vision modes."
One of the problems I noticed testing the above designs was that it was a bit overwhelming to users. After the initial walk through, I present lots of buttons at once and it seemed easy for them to get confused. I also noticed things which made sense for a clickable prototype didnt make sense in an actual AR app.
I continue to work on the project and have built a completely different design style. There are two main changes, less secondary menus, and moving details from an object away from floating cards in World Space and instead moving all information to a universal card in Screen Space on the bottom of the device.
Above is an example of a 3D projected overlay explaining how to engage Autopilot.
This shows how the universal menu works and is contextual depening what the user is looking at.
This has been a challenging but also very rewarding project forcing me to re-think common design patterns. AR apps are very different than traditional apps and easing users into the experience is no trivial task.
I believe from my research and testing of this prototype that an app like this could be very useful to Tesla. I set out to build something which would help them on-board new customers and feel this is a success. I also acknowlege implementing some of these features would be difficult and the UI could still use some tweaking to make the experience feel really natural and delightful.
I am actively pursing development of this concept, in particular bringing AR instructions to other objects and not just Tesla cars. While this concept focused on projecting an interactive AR Tesla, the next idea is to use machine learning to recognize objects in the real world. That project can be found at Relayhow.com where I focus on actually building this concept and getting things to work from an engineering perspective.