I was an Uber driver for a few days and this is what I learned

A few weeks ago I decided to do a little experiment and become an Uber driver for Uber Argentina. “Why?”, you may ask. Well, having worked with a lot of startups as investor and advisor for many years, there is one area where I think Argentina (and Latam in general) lacks, and that’s how products are developed and how we think about creating great user experiences. The US is a clear leader here, and its one of the reasons (not the only one, obviously) why they are so successful creating global products that we all use.

One of those successful global products is Uber, so I decided to embark on a research trip (literally) to find out more about this magical service that has been able to expand to over 100 cities worldwide.

The experiment

Between October 3 and October 15th I drove for a total of 11 rides. To do that, I had to go through the Uber driver sign up process, download the Uber Partners app and go chase passengers. The following is the summary of my experience:

Overall service and app functionality (good and not so good)

• Let’s face it: Uber has managed to disrupt a pretty established industry worldwide. Their service for the user is pretty amazing, and I would say that for the drivers (Uber Partner, as they call it, although I was not offered a % their equity, unfortunately), the app just works, which is actually pretty impressive. If you have ever been involved with digital product development, you know how hard it is to get a full experience to work flawlessly. Uber achieves that (almost, see below) and it does so many many million times a month. This is very hard to do, especially when you’re dealing with a real time marketplace and you’re trying to connect people in the offline world. Too many things can go wrong, and Uber has figured out most of them.
• The Uber Partner app is generally simple and intuitive, but I did find a major flaw for this market: 80% or more of the content in the app (and in emails that Uber sent me) is in English! I am pretty certain that a huge % of Uber drivers in Argentina have no English knowledge whatsoever, and it seems hard to understand how a company that is already operating in dozens of Spanish-speaking cities around the world hasn’t solved this yet.

The Home screen has some Spanish, but only because they send some messages in Spanish. Note how all the menus are in English, and this is even worse in the rest of the app screens.

• I did like the “gamified” approach to picking a ride. I found myself checking the Uber Partner app quite often while I drove “empty” through the city, and every time a new ride request popped into my screen, I felt good! I am not sure if this is purposely designed that way, but there’s a rush of adrenaline every time you get someone’s name on the screen and you have like 7 seconds to respond. If the Uber designers did this on purpose, kudos. But maybe it’s just me and I need to talk it with my shrink… :)

I “stole” this screenshot from Google Images since it was kind of hard to capture the screen while driving, but it looks exactly how the Arg App looks.The app gives you just a few seconds to answer the request, generating some sort of impulse that’s hard to resist.

On-boarding and using the app as a driver

• The on-boarding process to become a driver was incredibly (I would argue dangerously) simple. It took me literally 4 minutes to sign up and upload the 3 pictures of my driver’s license, my car insurance and my car registration. Moreover, I was able to do this whole process and start driving without ever talking -either in person or virtually — with an Uber representative. This lack of friction between the thought of “let’s try uber driving” and actually be able to drive is good for Uber because it increases conversion rates. Yet, I have to say that being able to pick up passengers without ever showing to Uber that I was a trustworthy, law-abiding individual is a little creepy. OK, is VERY creepy! And while one of the documents that are “required” is a police-issued background check that ensures that drivers are not accused of committing any crime, getting this document requires to show up at a police station and loose time, so I decided not to upload it and I was still able to sign up and start picking up passengers (I promise you I am a good guy and did not kill any of my passengers… or did I???) — more about this later.

The “Missing” background check certificate did not prevent me from riding Uber and picking up passengers. It did not even prevent me from receiving weekly payments!

• Back to the product review, as a driver I would have liked to have an internal messaging system within the Uber app: Since a large % of Uber users are tourists from abroad, the option to contact them via phone or SMS is not very reliable. In one case, I arrived at the pick up point (a very busy street where it was hard to wait for too long) and the passenger wasn’t there. When I tried to contact her via SMS, I noted her cellphone was from the UK. I sent her a message to let her know I was outside waiting, but of course I later found out (once she finally showed up) that she had not received it since her roaming was off. A very simple (and useful) addition to the app would be an internal messaging system that requires only to be online (while not many have data roaming, it is more likely that they will have access to wifi since they need it to use the Uber app to request the car!). It could also be as simple as sending the messages through multiple channels (SMS, WhatsApp or internal messaging and even email). That way, there’s a high likelihood that they receive the messages on time. And this feature should work both ways: as an Uber rider when traveling in other country, I many times had difficulties to contact the driver because I did not have a local chip and couldn’t easily call or SMS using my Argentina chip. Having an internal messaging system that didn’t rely on my data or SMS plan would have been very useful as a user.
• Another problem, which is also related to the “meeting” moment, is that while the rider sees in her app exactly where they are standing (the little blue dot in the map), the driver does NOT know where the passenger is. So, going back to my previous example, when I sent the SMS I had no idea if the passenger was in the vicinity or maybe she was a block away and coming. It would have been very helpful to see that she was close. Again, this would be an easy fix to the app and would provide the drivers with valuable information about the passengers and help them lower the anxiety that the “encounter” brings. 90% of the times I arrived to the pick up location and the passenger was not visible. And while the app showed me a message that “the user has been notified”, I sometimes had to wait several minutes for her to show up. If I were a real Uber driver, I would be pissed because all this time you wait you can’t really “start the ride”, which means that the driver is not being compensated for that wait at all. Not a huge issue, but given that the economics for the driver are not that solid, I think they should be compensated from the minute they reach their destination and are ready to start the ride (more on Economics later).
• Uber maps worked horribly wrong. Not sure if they are pulling the data from Google Maps, but after a few rides (and very bad directions), I ended up opting for Waze as my guide through the city. The Driver app gives you the choice to use Waze as the main GPS, and its integration is quite reasonable, but it had 2 small problems: 1) At the end of the ride, I had to close Waze and switch back to the Uber Driver app to conclude it, which was a little counterintuitive and cumbersome; 2) The name of the passenger remains in the Uber Partner app, so once I arrived to the pick up place, I had to switch back to the Uber Partner app to see the passenger’s name. None of these were big deals, but there’s room for improvement there (or just make a deal with Waze and replace the maps within the app!).
• This one really pissed me off, and I don’t understand how Uber hasn’t figured out a better way: There should be a way for the drivers to set up a “destination zone”. Let me explain: At the end of the driver’s “shift”, there should be a way to set up that you will only pick a passenger that is going to your “home’s zone” (or whichever zone you want to go next). That way, you don’t risk being sent to the opposite side of town (it happened to me 2 of the 3 days I rode, including a last-minute ride to the international airport which meant driving back home for over 1 hour without any passengers!), and the 3rd day I actually got a ride request 10 blocks from my home and I didn’t want to risk it so I didn’t accept it, which was bad for me, bad for Uber and bad for the rider, who maybe was needing a short ride, we’ll never know). I believe that drivers would even be willing to be paid LESS in that last ride, if they can end closer to their home base. With Uber’s technology and a few simple algorithms, this could be a win-win-win for Uber, the drivers and the passengers an have a huge (huuuuuuuuge!!) impact in the service availability, especially in new markets.

Conclusion#1: Safety concerns a big issue

• I am 100% certain that Uber needs to work harder on regulating itself to avoid lack of control and increase service safety. I am pretty shocked that there aren’t any (visible) controls in place to ensure that there’s a clear knowledge of what’s going on with its drivers and their cars. Even if they do have some controls in the background (and I suspect they do), I think it would be good for them to make them visible and explicit. It’s not just about being good, it’s also about people thinking you are good.
• It is sheer luck that we haven’t yet seen a murder or a serious crime involving an Uber ride in any emerging market. I mean: Uber is in hundreds of markets, many of them in emerging economies where it is pretty easy to forge (or steal) the paperwork that its required to sign up as a driver. Granted, you could argue that if something ever happens, Uber will have the data of where the driver was, etc. etc… but I doubt that will be enough for the person that gets mugged or attacked. And also, I could in theory find (or steal) someone else’s papers and drivers license, pick up a passenger, mark the ride as “complete”, go offline to stay off the grid, murder him, dump the body and run. (Ok, maybe I’ve seen too many Dexter episodes but you get the picture…). The fact that on-boarding is so frictionless is a little scary.
• Similarly, I believe there should be a mandatory review of each car that becomes part of the Uber fleet. As a rider, I had a near-death experience when I was taken to the airport in a car that had a wheel about to fall off and the driver was doing over 70mph as if he was driving a brand new car.
• Another big #fail relative to safety: While driving (luckily, without passengers) the app sent me a push notification that there was an available update. I was able to do the whole App update process (which involved several steps and touches in the screen) WHILE in the moving vehicle! Needless to say, this is NOT the best safety policy (using the phone while driving?) And again, these kind of things can easily be fixed (block the updates if a driver is moving, you KNOW that!!!) and it is quite amazing that no one in Uber has paid attention to it. It makes you think that there’s not much effort put in the overall safety of their product…

Not a very smart idea to allow drivers to update the app while driving. C’mon Uber… you could do much better!

• Finally, while I totally get the value of a quick and easy on-boarding to make sure that you don’t loose too many potential drivers with paperwork or admin stuff, I would feel much more comfortable if, somewhere during the process, someone from Uber contacted me to make sure that I am who I am and that my car is relatively safe to drive. There are many ways of doing this, including a remote Skype call (or asking drivers to send a picture holding their Drivers License to make sure that it is actually them, for example). And they could also do it in a way where you are able to start driving but you have X days to stop by the Uber office to validate your ID and review your car, holding payments until you do so (not that this would prevent a murderer to operate but it would at least be better than the current situation)

Conclusion #2: Contrary to what I expected, Unit Economics seem to favor Taxi over Uber for a driver

• Using the info provided by the app plus some additional data points, I made a quick calculation and concluded that, at most, an Uber Driver could make about ARS180–200/hour of revenue on average (and that is extremely optimistic). 1 USD = 15 ARS approximately, so that means drivers make on average less than USD 12–13.5/hour. That’s not too bad for a country where the minimum hourly wage is just USD 5. Yet, when you add the gas costs, car amortization, insurance, etc., the net amount gets down to between USD 7.5 and 9 (among other things, depending on how much milage you get from your car and how far “home” is related to where you have to pick up the passengers).
• Using the scenario, a driver that works 8 hours per day, 5 days a week would take home (net of direct costs) about USD 360 per week, or roughly USD 1,400 per month (8 hours x 5 days x $9/hour of net profit).
• Now let’s look at Taxis: according to some taxi drivers I interviewed, the taxi owners charge them ARS 1,200 / day for the car (and use of taxi license). This also includes covering any issue with the car, insurance, etc., so let’s assume that all direct and indirect costs for the driver are included in that amount.
• Taxi rates in BA are approximately double Uber’s (this varies in some cases but for simplicity let’s assume that)
• Considering the same amount of hours AND the same % of utilization/efficiency, this means a taxi driver would bring home USD 320 per week, or USD 1,480 per month ((8 hours x 5 days x $18/hour of revenue) — $80 / day of car payments x 5 days)
• This math might make you think that drivers are better off driving Uber, but this is misleading because we have not considered the real cost of ownership of the Uber car for the driver (capital investment, depreciation, insurance, out of pocket expenses, etc.). If we added all these extra costs for the Uber driver, driving a cab becomes the better deal.
• However, a big assumption we made here is that Taxis and Ubers are equally efficient. I really don’t think this is true, but can’t prove it either. I think more research should be done, but right now I don’t have any plans to start driving a Taxi, so that will have to wait.

Final thoughts

Overall, I really enjoyed doing this experiment, and I actually learned things that I was not expecting to learn. I also hope that this somehow helps the Uber product team to work on some of the issues I mentioned here and hopefully become a better company (hey, I am ok with a few free rides in compensation!!)

More importantly, it reminded me of a lesson I actually learned many years ago, as a fresh college grad at Exxon Argentina: we were about 30 recently graduated economists, engineers, lawyers, recruited from the top universities in Argentina. No matter where you were from, or what position you were heading to, the first week at Exxon consisted pumping gas and working at the convenience store at a gas station. I never quite understood why would the company spend money sending highly paid (well, not really, but to me it seemed like it!) employees to perform basic duties. Well… as my experience as an Uber driver hopefully showed: There is no better way to improve your product than making ABSOLUTELY ALL YOUR EMPLOYEES actively use it.

So my parting advise to all of the product entrepreneurs is simple: use your product. Many times. In many different ways. Make your new employees become users of your product. And watch them using it! That’s the only way you’ll note the flaws and be able to fix them to create a unique, magical experience. And that’s sometimes all it takes to separate a great, successful company from the rest of the pack.