Auto Navigator: Disrupting Car Buying for the Sake of the Customer
At SXSW Interactive 2018, we're giving festival-goers a look at how we're applying human-centered design, forward thinking, cutting-edge technology, deep automotive industry expertise, and innovative partnerships to reimagine the car buying process.
the problem
Do you remember how much time you spent at the dealership the last time you bought your car? Did you walk out, keys in hand, and feel completely confident in your purchase?
When it comes to big life decisions, 50% of people say it takes longer to find and buy a car than deciding where to go to college and choosing a child's name, according to a recent Capital One survey. Additionally, 62% of car buyers aren't completely sure they got a lot of money the last time they bought a car. In addition, 78% of Americans admit that the last time they bought a car, they lost confidence that they would get the car they wanted during the buying process.
We wondered why something as exciting as buying a new car should be so worrisome. What we've come to realize as we delve deeper into the car buying process is that some people get frustrated, realizing that the payments are (much) higher than they expected because the total cost of ownership is greater than the cost of the car. But often they don't know this at the very beginning of the buying process, when they set their heart on a particular car. so what? Is it back to the drawing board? Do they start over or accept something outside their budget?
The solution
These customer weaknesses proved to be a problem we wanted to solve, so in 2015, Capital One introduced the first version of Auto Navigator, a cloud-based web application built on a microservices architecture and powered by a suite of technologies including and machine learning.
This development has simplified the car buying process for customers - enabling them to find, finance and fulfill their next car purchase with ease, convenience and confidence.
With the second iteration, introduced just one year later in 2016, customers can browse over three million vehicles from more than 12,000 participating dealers nationwide! We used real-time data, which allowed us to determine how much customers would pay for any of the cars in our database for any combination of their choice (eg cash discounts, terms, etc.). More importantly, customers can now re-qualify for financing without any impact on their credit score before entering any agency.
From the ground up, we've iterated on Auto Navigator, implementing human-centered design where we use empathic listening, tracking technologies, and rapid prototyping with customers to continually build, fail, agile, and improve our customer experience.
What's next on this unsettling journey?
We are in this business to challenge the market for customers. At SXSW, we're taking a look at Auto Navigator's latest development in the form of new functionality:
Augmented Reality Advantage: We've realized that car research is an ongoing process that customers engage in, even when they're not actively looking to buy a car. In the coming months, we plan to offer our customers an augmented reality experience through the Capital One Mobile app, allowing them to scan vehicles through their mobile phones to view individual information related to each vehicle. This may include information such as pre-qualified financing, estimated monthly payment, nearby dealerships where they can purchase the vehicle, and more. Imagine if customers could access this kind of individual information at the moment of need, in real time!
The technology behind it
At launch, Augmented Reality will leverage Apple's recently released Augmented Reality Framework (ARKit) and new Mobile Machine Learning (CoreML) framework, as well as Capital One's own models for identifying and pricing cars. The feature is also compatible with Android, using mobile ARCore and Tensorflow.
ARKit allows us to detect surfaces and planes and create a 3D scene that we can then use to position our AR assets, while CoreML allows us to capture images from a video stream and pass them through our convolutional neural network to make vehicle predictions. We've improved the experience so that customers can scan multiple cars in one session and get more information about the car they've already scanned (even after scanning several). We then use Capital One's proprietary forms to customize the offer so they can see them