In February we took part in the first Innovation Safari at the "German Remarketing Congress" and won the pitch competition. The challenge was to develop innovative, digital solutions for car dealerships, in our case for a test drive. In a 24h hackathon, we automatically developed a chatbot via Facebook Messenger that arranges test drives for car dealers.

Why do car dealers need a chatbot now?

In cooperation with Autohaus Köpper, we were able to identify the following scenarios when organising test drives:

  1. One of the main problems was the lack of availability (24/7).
  2. Concluding a test drive involves many formalities, which is why there is great potential for digitalisation.
  3. During the process, 4-5 people are needed, which costs a lot of time and resources. The time spent could be better invested in closing deals.

All the above problems are addressed in the following post, which explains step by step how all the information needed about a Chatbot can be collected and how the test drive can be staffed without employees.

Step 1: Selection of a car via Chatbot

Chatbots for test drives

In the first step, the user is asked to choose a make of car and then the model. It makes sense to use special templates for each car in order to name further details such as aesthetics, pricing, years, etc..

Step 2: Scheduling via the chatbot

Chatbots for test drives

Once a particular car has been selected, an appointment must be made for the test drive. There are several ways to do this. You can either ask the user to type ina date in a standard format(dd/mm/yyyy) or even integrate a calendar. Webview integrate a calendar. In our case, the chatbot will process the given date and check with a database which hours are still available. These time frames are presented in quick responses/buttons where the user can choose their preferred time or go back to change the date.

Step 3: Collect relevant documents

Chatbots for test drives

Now it is time to finish the formalities. The user is asked to upload all relevant documents like his ID, driving licence etc. The chatbot can already extract crucial information like name, age, address , by using the Vision and NLP APIs from Google to store the data in a CRM. In case you are wondering how to store the personal data securely. Google offers a Data Loss Prevention API (Beta) which automatically detects sensitive data (text or images) and hides certain elements such as names, addresses, etc. Last but not least, the user can be sent a legal agreement for the test drive, which they have to sign. There is either the option for the user to take a picture of the signed agreement and upload it, or to take it to the test drive and hand it in on the spot.

Step 4: Summary and confirmation

Chatbots for test drives

In the last step, the user has the option to confirm his test drive to ensure that all appointments are correct.

Step 5: Reminder

Confirmation of the test drive

It makes sense to remind the driver the day before of his booked test drive.

Step 6: Feedback

 

Completion of the process

The day after, the user receives an automated message asking them to provide feedback. This is a great opportunity to gather valuable information that can help improve one's service as a car dealer. In addition, this provides an opportunity to remind the test driver, which in turn can increase the likelihood of a purchase. For example, one can send the tester an overview of the car, which points out the unique selling points.

This was a rough draft of the prototype we developed for a car dealer. With this innovation, we believe that not only can the whole process be optimised, but also 70% of the workload can be reduced and up to 50% less downtime can be achieved. However, there are many more use cases for this branch than just arranging a test drive. For example, the chatbot could be used to present all current car models and answer FAQs such as opening hours or product-related questions.

 

If you want to know more about chatbots and the implementation of conversational AI projects, get our white paper: