Why did I get to drive this car for €0.01 per kilometer?
April 13, 2021
I’m not sure but I have an idea.
This article was originally published on CodeX.
Disclaimer: I am in no way associated with any of the companies mentioned in this article.
A friend, whom I had not seen for months, invited me over for dinner. Neither of us had met with others in a while so it seemed safe to do so. The only thing left for me to do was to get there.
Living in the Netherlands, I mostly ride my bike to get around the city. Living in the Netherlands, I also frequently encounter terrible weather. Sometimes this is too much, for example, in this situation, where I did not want to cycle twenty minutes in a crazy downpour.
I don’t own a car, but luckily, a car sharing app called Juuve operates in my neighborhood: I can rent a car at any time for the combination of an hourly fee (typically €2.5) and a distance based fee (typically €0.25 per kilometer) without any additional cost (fuel is also included). Doing some quick math I estimated that visiting my friend would cost me €8–10, which, looking at the weather, seemed like a good deal.
When I opened the app to reserve car, I was surprised by what I saw. The hourly fee was €1, which, although lower than normal, was not surprising, as Juuve is currently running a marketing campaign with discounted hourly fees. The unexpected part was the distance based fee: €0.01 instead of the usual €0.25.
At first, I though I must have encountered a bug or a human error by Juuve, but made the reservation and then the trip hoping that they would actually honor the advertised price.
After getting home and finishing the ride I got the email which summarizes my reservation. The offered price was indeed honored. Intrigued, I took another trip the next day, and the price was still the same. When I checked the price of other cars in the neighborhood, they had their usual pricing, only this specific car had the low price.
“This does not make sense. Such an error should not occur. I should not be getting this great deal,” I thought. I was still thinking about this on my second cheap trip, then it suddenly hit me.
I should have figured this out faster. I am a data scientist; I analyze data, build models, run experiments. That is what this must be. An experiment.
A new product
My hunch is that Juuve is planning to launch a new product. It would be a subscription where you pay a relatively high fixed monthly price and get a number of hours and/or kilometers that you can drive for free or at greatly reduced cost. Similar to a personal lease setup, which are popular in the Netherlands, but in this case you would not have exclusive usage of the car, but could use any of Juuve’s cars.
Why would Juuve want to do this? One of their competitors, Snappcar Snappcar’s business model is somewhat different: you typically rent cars for a full day and get a number of kilometers, typically 150, at no extra cost, but you have to refuel the car before returning it. , offers regular and short-term lease contracts, so it might make sense for Juuve, too, to explore a similar direction.
As mentioned, I think that you won’t have your “own” car in Juuve’s setup. If they know how many cars they need on the streets so that everyone has one available when they need them, they will probably still need fewer cars than customers. After all, cars are parked 95 percent of the time. Having fewer cars than customers would mean that they can offer the service at a lower price than traditional lease contracts, which, in turn, would give them a strong competitive advantage.
From a personal perspective: for some reason, fixed costs are less painful to me than variable costs. I can easily imagine that paying for a €100 monthly subscription could feel more comfortable than paying €5–10–15–20 for each trip, even if I end up paying a bit more, since I would have some certainty that I can plan around. However, I cannot justify the cost of buying or leasing a car as I would not use it that much.
In addition to the forces outlined above, there are two direct clues that may also suggest that Juuve is moving in this direction. First, they recently launched subscriptions, which offer a discounted hourly fee (but no discounts on the distance based charge). Second, googling “juuve lease” shows a (mostly empty) page that says “Private Lease » Peugeot 108” — which happens to be the same model that I rented for this low price.
Why charge only €0.01?
If I were a data scientist at Juuve, and a stakeholder approached me with the idea of launching a lease-like subscription product, I would probably have also come up with a similar approach.
If we want to have a subscription model, we need to know how many cars we must have on the streets if we want our customers to always find a car when they need one. If they don’t find a car, they will be dissatisfied with the product and might stop using it (and use a competitor, get a regular lease or simply buy a car). Remember, they are paying a high monthly fee and expect constant availability. In order to plan for peak usage and know how many cars we need, we need to know how much and at what times our customers would drive them.
Sure, we could ask them about this potential product, for example, in a questionnaire. There are two problems with this. First, we might not want to really advertise this product until it’s ready to launch. Second, could you answer the question “How much would you drive a car per month if you had access to one and at what times?” if you don’t already have one? I, for one, can not give a reliable answer. I would say: “I don’t know, I’d need to see given the costs. And the weather. And my mood. But I’m really not sure.”
If you want to gauge the maximum potential demand for a product, you can give it out for free and see how many people take it.
By handing out kilometers essentially for free, Juuve gets very detailed and accurate data for the maximum potential demand for their cars. Obviously, they don’t want to do this with all their cars at the same time as they would lose too much money — but a small scale experiment, with just a few cars, might still be affordable and deliver priceless information.
This experiment tells Juuve the maximum amount a specific car is used, but it does not paint the full picture. If I were a data scientist there, I would also run an experiment where a small sample of users get this low price for all the cars. This would tell us the expected monthly usage of a user, which can be a very important input the right monthly price we should charge. Admittedly, Juuve might already be doing this — I am just not one of the lucky users who are in the treatment group.
If Juuve knows how far, how often, and for how long people drive if it is essentially free, they can build a very good estimate for the absolute maximum of demand. Combining this with other data, such as weather, time of the day, day of the week, etc. will allow them to get started with proper supply and pricing for this lease-like product.
Everything that I have written here is purely speculation. I might be just projecting my desires: affordable access to a car at a predictable monthly cost. I might have just discovered a bug in Juuve’s system. I might have accidentally stumbled upon their next big step. I am not sure. I just know that I felt really lucky and happy to have gotten these (almost) free trips and that I really enjoyed thinking and writing about this experience.