Profile photo of Richard White

On Uber: Big Data and Blind Spots

Recently Travis Kalanick, the CEO of Uber, took to Facebook to defend his company against complaints from long-time users that the quality of service is degrading in San Francisco. It’s a fantastic display of transparency which you don’t often see from CEOs so I highly recommend you check out his post for that reason alone. One of his more interesting nuggets of information that Travis drops in that thread is that they’re own internal metrics don’t reflect any changes in the quality of the service.

As a long-time Uber rider myself, I couldn’t help but think that this is a common problem we see, especially with “big data” companies, whenever there’s qualitative feedback that isn’t supported by quantitative ones. The crux of the problem is twofold:

  1. It’s too easy to believe your own metrics. Even when they may not accurately describe what’s the end-user UX is.
  2. Many companies don’t give enough thought to how they get customer feedback. The fact that they get any feedback is often proof enough that they have a solid, representative system in place.

Obviously I have a horse in this race, given that UserVoice focuses on drawing out what I call explicit feedback, both qualitative & quantitative, that’s gathered directly from users rather than using implicit feedback, event tracking and behavioral signals which are all the rage these days.

The tricky thing about explicit feedback, which Uber relies heavily on with their star ratings for drivers, is that it’s easy to get biased information if you’re not careful in how, and when, you ask for that feedback from a user. In Uber’s case I think there’s a lot of opportunity for improvement but let’s first look at what happens when you call for an Uber.

Current Experience

1. You order a car.

Note that if you cancel because the driver is taking forever to get to you or is going the wrong way there’s NO way to give Uber feedback on that driver. This is a common complaint among Uber users recently. I’ll give Travis the benefit of the doubt here and say that maybe looking at cancellation rates alone would be enough here

2. Car comes and you jump in

Note that while you’re en route there’s no way to leave feedback on the driver. This is the most frustrating part to me. Whether you’re in a web app or playing a mobile game it’s crucial to allow users to give feedback at the point of pain. When my UberX driver drove 0.5 miles out of the way the other morning there was no way for me log that feedback until I got out.

3. After you’re dropped off, you’re prompted to rate the driver.

However would you open Uber again once you’re out of the car? No, you wouldn’t. You open the app when you’re looking to get your next ride (when I was making these screenshots I first had to rate a driver from 3 days ago). At which point you’ve likely either forgotten the bad experience you had before or, even more likely, you just hit 5 to get it out of your way because you want a car RIGHT NOW DAMMIT!

Another issue with this screen is the fact that it’s impossible to separate out feedback on the driver vs feedback for Uber itself. I live on a corner and not a single Uber has picked me up less than a half a block away from me. I always end up walking down the street. But I have no way to tell Uber this. Or tell them that it NEVER remembers my home address.

A Better Way (with Minimal Effort)

This is why of all the bad Uber experiences I’ve had in the last couple months I’ve officially reported none of them. However there’s a couple easy improvements that could be made to change that:

1. Add in a persistent way to get help or give feedback on Uber itself anytime I’m in the app.

If you leave your phone in an Uber or something worse happened you want an easy way to contact Uber quickly. To do this they should have a persistent button that you can always find (and ideally it goes to something more sophisticated than a Mail composer. Check out PicCollage’s success story). Showing users where it is before there is a problem is key to it being used when there is an issue.

But there’s a greater opportunity here and that’s to get feedback on how the Uber app or service itself could be better. That’s because you usually have a few minutes to kill when you’re waiting to be picked up on en route. That’s when all the great ideas I’ve had for how to improve Uber have come from. If only there was an easy way for me to tell them.

2. Don’t wait until the ride is over to get feedback.

Once I’m in the Uber and en route there should be a big button to give feedback on the driver. Is the driver taking a weird route? or am I doing all the navigating? I can click that button and enter in my rating right then. Now may be the ride starts off rocky but the drivers redeems themselves. That’s cool. I can change my rating any time before my next Uber ride.

If you really wanted to get better quality you could probably even text me or do a push notification to prompt me to rate (or confirm my existing rating) when I got out of the car. That way I don’t have that screen getting in my way next time I go to open the app.

Don’t get me wrong, I love Uber. And while these may sound like quibbling annoyances with an otherwise amazing service, it’s this last mile of UX, that last 5% of functionality and polish, which really separates great products from just really good ones.

And though Uber has the pole position now we’ve seen time and time again that if you’re no longer the product that’s nailed that last mile best you run the risk of a swift decline to irrelevancy (see Digg, Blackberry, Nokia, etc). Though customer feedback gets a bad rap because “people would have wanted a faster horse” it’s this last mile of UX where customer feedback can really shine and help get your product over the hump.