Driven to succeed

Increase how likely new drivers find passengers

About BlaBlaCar

BlaBlaCar is an online marketplace for carpooling. Its website and mobile apps connect drivers and passengers willing to travel together between cities and share the cost of the journey. With 75 million users, including 10 million active drivers the service is available in 21 countries including France, Russia, Mexico, Brasil, India, or Turkey.


Drivers who offer their first ride on BlaBlaCar aren't familiar with the experience of carpooling. Their probability to find passengers depend on a lot of criteria (country, axis, date & time, stops, options, profiles…). As of today, we don't guide them to increase their chances and become a successful active driver…

User problem

"It's my first time on BlaBlacar, I want to offer some seats for my next trip but I don't know how it works."


Onboard new drivers the right way so they feel comfortable and find passengers.


Newbie driver success

Starting from a data 

Everything started from a data, the newbie driver success. It was telling us that we were doing a really bad job onboarding new drivers and making them successful, no matter the country. From this information, we were asked to investigate and understand what was going on there. So we started our user research. We called newbie drivers (successful or not), we carpooled with some of them, we invited some in our office to do longer interviews. We also did some usability tests on the old publication flow with non-BlaBlaCar users. In the meantime, we were investigeting the data to know quantitatively the impact of each option on the driver success, options that the driver has to choose during the publication.  

Keeping in mind our objective of improving the newbie driver success, we then synthesized, prioritized and shared our findings. After iterating on the design of the flow and we converge to the one that have the best chance to solve our user problem. We tested three times the usability of this flow with both curent BlaBlaCar users and non-users. Then, knowing that the success of the driver can't be tested due to too many other factors (Destination, Date and Time, User profile…).

This article is a summary of the main challenges we encountered.

Asking one thing at a time

Offering a ride on BlaBlaCar is pretty simple, but if you want to make sure that passengers book your ride, as a driver you need to understand and get some empathy to the other side of the marketplace. Thanks to the analysis of our data, we discovered that newbie drivers were offering rides of poor quality (e.g. Offering a ride without photo can have huge impact on the booking rate). So in other words to improve their chances of success, our main challenge was to improve the quality of the ride they offer.

With our qualitative research we confirmed our assumption that most of the newbies don't understand the value of some options we were offering. BlaBlaCar being developped as a web service the old publication flow was a three pages form, the user had to complete arround 10 fields per page. Redesigning these three pages was not an option, we needed way more space to explain clearly the value of each options. So in order to guide them the right way, while minimizing their cognitive load we took the approach of asking only one action per screen. So we redesigned the entire publication flow, rethinking questions by question… 

The first thing to do was to remove from the publication (almost) all the features that didn't have any significant impact on our KPI, the driver's success. Following the same logic arround our KPI, we decided to add two more steps linked to the profile completion directly in the flow. So at the end of the day, we still had a lot questions for their first publication. That's a lot, so we had to find a way to make it as smooth as possible. That's where the content played a key role.


Having a conversation

One other challenge was to keep the same conversion rate of the flow than before, or even increase it. Knowing that, multiplying the number of pages by 6 may sound silly, but we were convinced that the less action the user has to do in a page, the more he will complete it. But we really had to find the perfect way to ask 20 questions while staying attractive, clear and keeping the user motivated. Our solution was to design a real conversation with the user, as BlaBlaCar was someone. Each page was like BlaBlaCar speaking. The order of the flow was directly impacted by the conversation, it has to feel natural, like if the user was speaking with a friend. 

We worked very hard to make sure everything was fitting, with the right tone of voice and with continuity… At first, we were designing the conversation without thinking at all about the pages, knowing each time BlaBlaCar will speak, it will be a page. It was like designing a script for a movie, knowing that the story will be different for everyone (different cities, different countries, contexts, cars…). 

At one point we had to create the screens and adapt the content to make fit all these scenarios. The question had to be at the heart of each page, the objective was to allow the user to answer the question and make the action easily, without any thoughts on about the screen. We had to find a way to make the UI almost invisible in order to reduce its cognitive load. When we couldn't shorten a message enough and we had to strengthen it, we used illustration.


At the end of the day, we were really confident about this flow. Actually, maybe a bit too much, because it almost didn't impact the newbie driver success on BlaBlaCar's marketplace.

But I definitely won't say this project was a fail. By doing it, we isolated each question in unique screen, which created a good fondation to further iterations. Plus, while keeping the same conversation rate, this 1 action per screen flow gave us a much better adoption rate than before on many features and options. 

This project helps the product and design team to feed the design system that was under construction at this time. For example, the one action per screen approach was integrated in the system and reused many times in the product since then. On the content side, we also keep the approach of the conversation as much as we can, and align the tone of voice with our brand. We also create a template of pages for these kinds of flow, 1 title, 2 option, and nothing more.

Modular and beyond

In the meantime, we realized quickly that this project was only a small part of what we had to redesign. Creating the newbie driver publication was one thing, but we had to think bigger and design for all our drivers. So based on the 1 action per screen, we did create a modular paradigm.

Having 1 choice per page was a bold decision because the user will enter into a 22 pages flow. For a newbie, it's perfect cause you can explain clearly the choice to make and actually you don't have any information about the user, so you have to ask all the questions.

But for some of our most engaged users, that publish sometimes 4 times a week, it can quickly become real nightmare… So our strategy was to create a publication experience that learns everytime you use the platform. And by comparing the data from different rides of the same drivers, we discovered that most of our users stick to their preferences and their axis.