Startup Idea: A Better Uber

Uber and Ola have failed at providing a reliable service, because they have too many problems:

  • When you book, drivers call and ask where you’re going, but you may be in another call. Even otherwise, it’s a waste of your time.
  • Drivers often cancel.
  • It almost always takes 10–20 min for a car to arrive, and when the driver cancels, another 10–20 min.
  • In one case, the driver didn’t cancel, but deliberately kept driving away. I couldn’t book another one till I canceled this, and I was penalised for it. Uber’s customer dis-service didn’t help.
  • Drivers sometimes refuse to stop in front of my office, or go into my apartment complex, instead leaving me at the gate.
  • The AC sometimes doesn’t work.
  • One driver, as soon as I got in, checked the destination and said that he’s not going there, forcing me to get off.
  • One or two drivers abused me.
  • Uber sends me marketing spam, so I opted out. But now I can’t log in because it tells me that I have to opt back in to receive the verification code. But I never opted out of receiving verification code SMSs, and I won’t agree to receiving marketing trash just to use their service.
  • I subscribed to Ola Select by paying more than ₹400 for a month, which is supposed to give me priority booking, but I still couldn’t get a single cab in two weeks.
  • Customer service is not available on phone when you need it.

I usually drive nowadays because it’s just less hassle than all the above. Uber and Ola have failed at their job.

A startup to deliver reliable cabs

We can build a better service that works the way customers want. We’ll kick out drivers who engaged in the above behavior, even if there’s one confirmed case of, for example, abusing the customer. Or deliberately driving away from the customer’s pickup point without canceling, which can be detected since we’re tracking the driver’s location. Or refusing to go 50 meters further than the location shown on the map because the location was shown inaccurately. As for cancelation, we’ll have a strict policy of say only 1 ride can be canceled per 100 rides completed. If the AC isn’t working or the windshield is cracked or the car is dirty or otherwise in a bad condition, we’ll refund the customer, and further offer the same amount as credit for the next ride. We’ll charge the driver a penalty of 3x the fare, or ₹1000, whichever is higher.

Since these policies are not in drivers’ favor, we’ll have to pay them more to compensate. Which is passed on to customers.

Target Users

If our high prices means the majority of Uber and Ola riders won’t use our service, that’s fine. We’re competing on quality and service, not on price. I’d rather run a company that makes a profit of ₹10 crore a year than a loss of more than ₹2000 crore a year in case of Ola. Or $5 billion in a single quarter in case of Uber! I’d rather be a small success than a big failure.

Upmarket is where you make a profit. Not the mass-market where people aren’t willing to pay what it costs to run the service, let alone generate a profit.

Uber and Ola have 3 million rides per day. If we assume that 5% of users are willing to pay more for a reliable service, that’s still a lakh rides per day.

Is this viable? Not at Uber or Ola’s scale, but certainly at a smaller case. Taxis have existed for decades before being apps came along. It’s a viable, proven business model, just as restaurants are a proven business model. And adding an app doesn’t make it unviable, because a tech team is a fixed cost.

Getting the Network Started

Any marketplace or network has a chicken-and-egg problem: drivers won’t join the service because there aren’t enough customers, and vice-versa. Even those who join will leave because the service won’t work properly without enough volume, generating a bad reputation for the service, and detering future users from using it.

The only way to beat this is to subsidise the network to get started. Assume we’re launching in Mumbai [1]. It has an area of 4300 sq km. Say we want to station a taxi in each square kilometer block so that customers get a taxi quickly. That means 4300 taxis. Considering that we want 24/7 service and a driver can’t drive all the time, that means say 10,000 drivers need to be available for round-the-clock service [2]. If each earns ₹40,000 a month, that means $6 million a month, which is not a problem for a VC.

For our first batch of 10,000 drivers, we’ll guarantee that they get paid 8 hours a day as if they’re fully occupied.

We won’t offer discounts to customers, let alone offer free rides like Uber did in their early days. Any subsidies will be given directly to drivers. Because discounted customers take more rides, preventing other customers who are willing to pay the full price from finding a cab. This will create dissatisfaction and defeat the point of the service in the first place. Why will anyone pay more than Uber for a service that isn’t more reliable? Companies often give discounts in a way that backfires on them, like Rapido’s discounts making it impossible for me to find a ride, which contributed to me stopping using their service. You can overcome this by having more cabs, but that requires a higher investment. Which doesn’t make sense till you’ve validated that the business is viable. Startups should validate that they have an economically viable and sustainable business with as little capital as possible. So subsidise drivers, not customers.

Besides, if you don’t give a discount, you’ll know whether customers are willing to pay what it takes to run the service. If they’re not, you’re better off failing fast, burning less capital in the process. And if they’re, that’s a strong case for an investment from investors, stronger than “Give us billions of dollars and maybe we’ll build a sustainable, profitable business, or maybe we won’t”, which is effectively the pitch that companies without proven unit economics make to investors.

We’ll taper off subsidies as soon as possible. Subsidies are necessary only to get the network started, not to buy growth, which is anyway fake growth. As the saying from the US goes, any fool can sell a dollar for 90 cents.

To begin with, we’ll have only one class of service, unlike UberGo, UberX and Uber Premium. We’ll allow only those cars that are spacious enough for one tall person like me to stretch his legs fully. Such a car doesn’t need to cost much. My Marurti Ritz meets that criterion if you sit in the front passenger seat and move the seat all the way back.

[1] Because it has the most crorepatis. The number of well-to-do people matter, not the raw GDP, because tons of people who won’t be able to afford our services also contribute to the GDP.

[2] But another driver can drive the same car when the first is sleeping. Unlike Uber and Ola, we’ll explicitly support this in our system rather than assuming one driver per car. After all, it makes sense for a taxi driver who took a huge loan to buy a car to make the most of the car by renting it to his friend when he’s sleeping.

Consulting CTO. Earlier: Google | Founder | CTO | Advisor