Testing and iterating a “car-on-demand” service

I’m a believer of agile methods, testing and experimenting. In this post, I’m writing of a very small ongoing real-life piece of “action research” – testing alternatives for putting in place an “car-on-demand” service that would enable people from our apartment building to use a car without owning one. Since our family and many neighbors do not own a car but would (maybe) occasionally want to use one, it might make sense to have an arrangement where there would be a car available somewhere nearby “on demand” ie. with a pay-per-use model.

There are three main untested hypotheses I identified.

  • Demand. Would I and others be willing to pay to use an “on-demand” car, and how often? Those who do not have a car currently have typically structured their life around not having a car, so maybe there is actually very little need for a car given the excellent bus, tram, metro and train network in Southern Finland & the opportunity to use taxi’s when needed.
  • Supply. How many cars are needed to satisfy the demand? Is one enough, or would there need to be a way to scale up the capacity to accommodate e.g., for the fact that everyone wants to use it on the same sunny Saturday?
  • Booking and payment system. What is a fair and easy-to-use way to determine who gets to use the car when, and how to split the costs of the car?

In terms of the competitive landscape, there is currently an EkoRent electric car rental location roughly half a kilometer away, from which one can rent an electric car for 11 EUR / hour without any other costs. Another car rental company 24rent.fi has similar pricing and multiple available cars, with the closest one being one kilometer away. The service kortteliauto.fi has a very attractive cost structure especially for long-term rental (at roughly 0.4 EUR / km for overnight trip of 240 kilometers). And of course people have friends and family ready to share their cars for free, in exchange for gifts, or for a fee.

So we started the journey of determining what would be the best way to go forward. And given the agile & lean mindset, of course the best way to start is by making simple experiments to learn about the real issues, and then optimize and change the concept as we learn more.

Idea 1: Low cost car test
First idea was to buy a used car in the price range of 2000 EUR, spend 400 EUR on insurance and taxes for first 6 months, and then divide the cost to 6 shares of 400 EUR each. Use Whatsapp channel where people send message to book the car, report mileage after their drive and the sums with which they have fueled with. Balance the books (split fuel & other costs per kilometer driven) every few months. End the experiment after 6 months, sell the car (if it did not break down earlier) and gather lessons learned. Did anyone use the car? What was the cost per kilometer? Did we experience shortages on popular days? Would it make sense to somehow automate the booking system or was it just a small pain to balance it manually?

Cost of the experiment? Max 400 EUR (assuming zero value of car after 6 months – very unlikely) + any true cost per kilometer you drive. If nobody ends up using the car, the only cost was the insurance and taxes for 6 months.

“Waterfall” counter-idea to Idea 1
A competing “large scale / waterfall” based model was also proposed by others. First, it was argued that a specific car costing 17 000 EUR would be more economic in the long term given repair costs etc., and more enjoyable to drive. Second, a more “efficient” way of splitting the costs based on fixed 100 EUR fee per month was proposed on grounds of simplicity and preditability of costs. Third, kilometers would be limited to 1000 per quarter (over that, extra costs) and that limit would be tracked based on a written “driving log” in the car. Fourth, a specific credit card would be bought to pay for the fuel.

In this model, the decision to join (assuming 6 families buy shares) would set you back roughly 3400 EUR for 6 months even if you do not drive at all. And the lessons learned on demand (and thus supply) will be skewed, as people basically have the incentives to drive the car for the minimum monthly allowance each month. So you would replace a bus trip with a car trip just because your marginal cost is zero for the car option. And third, you would lack unbiased estimates for the amount driven by each family (again due to the minimum quota) and consequently not learn much about the fair ways of splitting costs. This idea was rejected.

Idea 2: Give space to external renting company
Parallel to discussion between the agile-testing and waterfall-building options around Idea 1, we learned about a company that is willing to put a car to an empty space in the yard of the apartment building for zero cost, and then rent it out all persons living nearby for a variable fee of roughly 11 EUR / hour (excluding fuel), with a 1 month cancellation period for the contract.

So I proposed we try this out: we commit to a 1 month test period and then decide if we want to continue. We would learn about demand and supply, and the booking system would be automated. Cost of joining the system would be zero. Of course the company would be making a profit, so this is not an optimal long term solution assuming demand would exist. On the positive side, this system would be very easy to join and leave (unlike the share-based models) so we would get a good view on how many families actually wanted to join a system like this at all.

Idea 2 meets the waterfall
This idea was was met with a number of very good reasons why it would not be a sustainable solution from here to eternity (waterfall-mindset of “deliver” being the final permanent stage). The argumentation was based on noting possible worst-case scenarios if the system was decided to continue from here on indefinitely without any changes. Example criticism included

– Is this the right way to decide on this kind of a an idea, or do we need to involve the board of the building or some other way of making sure the decision is made in the right way?
– Can we find alternatives as to which rental company to use?
– Shouldn’t we make the company pay rent for the parking place?
– Is this particular place on the yard the best place for placing the car?
– Should we change the yard plan officially to make it possible to have a parking place?

While these questions are interesting, they are not helping us answer the main uncertainties related to the project – supply, demand, booking & payment. Debating for example the rent of the place is a bit theoretical if we plan to do a one month test: at maximum, we are talking of 100 to 200 EUR lost rent for a month, and that money would definitely not come either if we did not launch the test!

Idea 3: “General terms” for borrowing a neighbor’s car
A third idea for a lean and simple solution to “on-demand car” is now brewing. The starting point it the observation that there are actually in any given point in time multiple cars in the garage that are not being used. So why buy / obtain a new car if existing capacity is not utilized to the maximum?

So what is preventing people from borrowing cars from each other in a flexible, self-organized way? Barriers are roughly

  • Social barriers – is it OK or not to borrow the car? How can I repay the favour? Is my car fancy and clean enough to borrow?
  • Lack of price level visibility – what is the right charge for borrowing a car? Free? Or X per km as it does cost money to own and maintain a car?
  • Lack of efficient communication – how do I ask for a car to borrow? How do I advertise for a car being available?
  • Transaction costs – is it worth the hassle to agree on the pricing and other terms (e.g., liability for damage) for this short borrowing, or should I just go with a commercial operator?
  • Trust – all parties have to have sufficient trust in each other to enter into agreements. In our apartment building, trust is higher than on average as people know each other very well through numerous common activities

So what if we could devise “general terms” and a system for car rental, ie.

  • Defined pricing that is fair for both parties (e.g., 0.40 EUR / km covering fuel)
  • Way to ask and offer cars (e.g., Whatsapp channel, first to reply to ask gets the deal)
  • Basic agreement on who covers risks (e.g., all cars are insured and owner covers risks whereas borrowed pays the possible co-payment in case of accident) and other basic conditions (e.g.,  car keys returned to mailbox, no need to clean car for short rental periods etc.)

By launching this type of a system, we would learn about demand, supply and cost systems. We could easily tweak the “general terms” based on feedback, and if necessary, upgrade from a Whatsapp system to a more robust platform. And, if there is sufficient uptake of this model, we could consider jointly buying a dedicated car for this.

I’ve now started testing with Idea 3. Instead of opening the idea to general debate within the apartment building, I’ve actually written a draft version of the “general terms” and asked a few car owners if they would be willing to borrow their cars with those terms next Saturday for 6 hours. And the first one has said yes!

One thought on “Testing and iterating a “car-on-demand” service

Leave a Reply