Revolutionizing the Porsche Experience: AI-Powered Porsche Configurator
Want to buy your dream Porsche but can't afford the wait?
The Porsche Configurator is a simple mobile and online web application that makes choosing alternatives as smooth as its ride — www.porsche.com/germany
Porsche, is a German automobile manufacturer specializing in high-performance sports cars.
Project overview
+ Results
the Goal
Increase data quality for configuration analysis in the worldwide sales network of Porsche AG. Optimize customer preference predictions where little or no data is available.
My role
Project Manager, UIUX Designer designing user flow, prototyping, user testing, evaluation, user interface design
Results
-
57% regional improvements in user experience in the locating, customizing and ordering experience for Porsche’s luxury cars.
-
43% improvement in customer preference predictions
responsibilities + reception
I was a Project Manager for a team of 12 students from HdM Stuttgart. The team consisted of 1 Project Manager, 5 developers, 1 data scientist, 3 UIUX Designers, 1 Visual Designer and 1 Business Analyst. To be sure we're addressing the right problems, we worked closely with Simon Dürr and Robin Seyock from Porsche AG, designing our process around regular cycles of prototyping, proposals and revisions. We also regularly hosted face-to-face user sessions, validating every step of the way with the input from real users.
We designed Porsche Configurator mobile app with AI trained chat bot to increase data quality for configuration analysis in the worldwide sales network of Porsche AG.
● Obtained user data with analytics reporting, task analysis, and participatory design techniques and leveraged that data to create iterative improvements, optimizing user experience and functionality of online configuration websites. Proposed site decreased the bounce rate by 57%.
● In a Scrum environment, strategized and implemented recommendation engines, improving prediction of consumer trends and desired configurations by 30%.
● Launched a technical prototype of AI-driven chatbot to solve existing “cold-start problems”, to analyze and gather existing market and public opinion, improving the quality of results of recommendations for the Porsche dealership by 55%.
● Responsible for managing and coordinating resources, and cross functional teams through progress measurement activities
the cold start problem
issues faced
Porsche's sales cycle information is not fully integrated, with disparate data from dealerships and online data records. This results in low data
availability for generating configuration recommendations for low volume derivatives (e.g. 911 Turbo S) as well as for product launches (e.g. Taycan).
There is no historical data available.
Therefore forecasts can't be made based on historical order data.
Predictions must be made in other ways.
background
In the past, Porsche sales ecosystem has mainly focused on reselling through individual dealerships. This prevents a clear understanding of market needs and customer feedback in the sales journey. Users are also unable to get their exact configuration in dealerships. Over the years, Porsche has tried transitioning into online direct sales, allowing complete customization and control in the car ordered. However, over 70% of Porsche's customers prefer to buy immediately available vehicles online or in dealerships, which results in purchasing worse equipped vehicles and poorer satisfaction.
so why can't porsche get customers to buy online?
challenge 1
Customers report a lack in confidence in online purchases, as they do not feel a personal connection. They also report decision paralysis when overwhelmed with online options. Most choose to purchase at local dealerships as they are most convinced by in person interactive and tactile experience.
lack of interaction
challenge 2
Customers feel lower levels of product assurance when they are made to wait 4 to 6 months for their personally configured vehicle. As cars are practical products, many customers worry that they may make a irreplaceable mistake that only reveals itself 6 months later. Thus, many simply cannot wait and choose to make the purchase at local dealerships
do not like to wait
so what does it mean for porsche?
Dealerships have to pre-order cars with configurations expected to be desirable to resell. Due to the interim financing, the warehouse vehicles ordered are usually much worse equipped than vehicles desired by customers, leading to long term customer dissatisfaction towards the Porsche Brand.
dealership guesswork
challenge 1
lack of data
challenge 2
When the sales process is influenced by multiple factors, Porsche AG lacks perspective in its sales performance and customer outlook. Therefore it is unable to use the full sales and turnover potential in marketing or selling storage vehicles.
To get time-sensitive HR data and learn more on relational analytics within an organization, we created the Porsche Configurator as a comprehensive solution to resolve pain points faced by both Porsche AG and Porsche Consumers. The Porsche Configurator uses prioritization tools to propose and evaluate alternative configurations for the customer, creating clear user journeys to reduce decision stress on the users end; and increasing Porsche's understanding of user preference.
understanding the user
-
User Research
-
Persona
-
User Journey Map
-
User Flows
To create the architecture of the configurator, we asked our users to define the most important objectives they have in mind when they are looking to purchase their Porsche.
desire 1
desire 2
Customers wish to use the Porsche Finder
database to search for immediately available
new vehicles or similar alternatives near them.
varied overview of configurations+ alternatives
Customers wish to receive support during the purchasing process by being offered alternatives based on statistic predictions.
personal touch
in offering suggestions
research
We looked into the benefits and drawbacks of the current web based platform, and looked into how integration of 3d immersive gamification can enhance the online purchasing experience. As a starting point, we did some competitor research to investigate the current market offerings and draw inspiration from specific features that we liked about each platform.
current product analysis
Resolving solution- prioritization tool
Previously, the online Porsche finder configurator only showed the immediate availability of one model within a limited radius. It lacked comparative capabilities for customers to compare the differences in availability and configurations of its different models. Thus the navigation process becomes time consuming for potential buyers.
Competing brands like Audi had a similar method of comparison, presenting vehicles with similar configurations. However, it do not seek to rank the priority of importance of the configurations. This method offers too many alternatives, complicating the choice.
Customers use the tool to search for new cars in the area. In the case that their desired configuration does not exist, by prioritizing the features of their configuration, our tool will find a vehicle that still fits their needs. The tool collects the prioritization data. This allows us to calculate a qualifier from the collected data and thus evaluate the relevance of the available data sets.
Offers many alternative configurations. But does not seek to rank the priority of importance of the configurations, complicating decision making.
weakness 1
Strict locational limitation causes bad user experience, causing very limited buyers perspective. It also eliminates hope of finding the perfect vehicle 'a little further away'.
weakness 2
calculate importance based on user ranking feedback
product goal
Using prioritizations the existing data can be qualified.
A qualifier was assigned to existing configuration data. An automated script evaluates and prepares existing data for use in the Recommendation Engine.
hypothesis 1
Customers will prioritize options to find a suitable car. Willingness and flexibility to travel further is expected.
Use AttrakDiff user testing to test usability.
Use testers to review in Porsche fair.
hypothesis 2
PERSONA
Using the quantitative and qualitative data from interviews and survey results, I defined the three target group profiles Peter (Teacher, 55), Sarah (Student, 16) and Tobias (Student, 13) to better empathize with my main user groups and prioritize goals according to their needs.
.jpg)
porsche finder
Find Your Porsche
The customer uses the improved Porsche finder and defines
his own locational scope to search for an available car.
Responsive filter
Ability to see flexible locational range of +5km. Visual indication of suggested configurations located further away.
ranking prioritization
Rank Choices
The customer lists and ranks his preferred qualities in a car, prioritizing his options.
Similar Recomendations
If customer's exact configuration is unavailable, he can expand the search results by prioritizing the options and finding the most similar configurations.
Tailored Prioritization
If the customer prioritizes getting his car directly over waiting for a Porsche with his exact configurations, alternative vehicles which are immediately available and fit his preferences will be recommended.
Based on the insights gained from the initial content audits, competitor analyzes and Card Sorts with potential users, I defined the sitemap for Porsche and then evaluated it via tree tests with potential users.
paper prototyping
.jpg)
.jpg)
Wireframing,
Prototyping
& Usability-Testing
With low-fidelity prototypes, the planned dashboards and the general structure of the application could easily be tested in usability tests. Without much effort, adjustments could be made before going into the much more costly digital implementation.
Clickable prototypes
Building a demo we could do user testing and refine the user experience process.
Testing with AttrakDiff revealed that our idea offered a "genuine", "understanding" and "enjoyable" approach to choosing a car. It sped
up the process by 45% and made the choice
"less confusing" and "more straightforward".
Apart from the ease of use, the simplicity of the configurator was crucial, Analyzing the funnel, we were able to identify the biggest pain point and streamline the flow for the least amount of steps and the highest completion ratio
For the best viewing experience, please visit the site on Desktop
