How Product Psychology Could Stop Uber Drivers From Stealing

Did you know that Uber Eats drivers openly share strategies on Reddit about which orders to cancel, prioritise, or pass onto less savvy drivers?

There's almost a pseudoscience for detecting likely tip baiters (don't worry, I'll explain what this is later).

Call me naive, but I assumed that when drivers stole food, or cancelled an order that's in progress, their wives were having a baby or something.

It's nothing like that.

So what's this got to do with UX?

This is the second time I've published an analysis on being an Uber Driver. The first study looked at how the driver's app was poorly designed and difficult to learn.

This time we're tackling a bigger problem: how product psychology could reduce stolen (or delayed) orders, increase tips and make drivers happier, before the event occurs.

This kind of preventative design is sometimes labelled "Upstream Thinking".

And in about 10 minutes, you'll have this in your arsenal to try for yourself.

Ping. "Your order has arrived". 

You rush to the front door. But there's nothing.

For exactly this scenario, drivers are asked to upload photos as evidence.

They'll even suggest that images include the entire door in the frame, with identifiable house numbers.

You can take a picture of a Kinder Egg, a brick wall, or a bag in a bush.

It's not just that these images are shown to customers, it's that this could have been flagged before the driver had walked away.

Consider the "wait, that's not my door" problem with a preventative lens.

You could ask the user to upload a picture of their door, which is then auto-magically cross-referenced to whatever image the driver takes as evidence.

If I have a blue door, and my food is photographed sat outside of a red door, the driver should be notified, obviously.

Let's think about the customer experience for a moment.

Whilst studying the world of online Uber Driver game theory, I learned about a chess move that all drivers fear.

Customers can pre-approve a tip before an order is collected, and Uber will include this in the value of the order, to the driver.

This motivates drivers to approve their order quickly.

Then, after the driver has completed a flawless delivery, the customer can reduce the tip.

Yes, after the food has been delivered.

The driver sees a message about why they've not been paid what they were expecting.

You'd hope that Uber would at least try to discourage customers from doing this, right?

Well, they don't. Like a driver cancelling an order, they've made it easy.

There's no contextual framing about how far the driver has driven, or how much of the total fare this tip represents.

There's no attempt to humanise the driver as a real person, being paid (sometimes less than) minimum wage.

The customer isn't then warned about this practice. There are no rammifications.

The flow is optimised for clicks, not for action or outcome.

Drivers don't get paid to wait around.

As a compromise, there's a mechanism where if a PIN is required to complete a delivery, but the driver can't find the customer, a countdown will begin.

After this, they can just ditch the food and be paid. Photos are optional.

Their food is about to be abandoned.

The customer gets an automated call (which I genuinely missed as I was doing this), but aside from that, the app does nothing.

No sense of urgency. No alerts. No timer.

The entirety of the customer's app should be screaming.

You've likely noticed the trend: Uber makes almost no attempt to stop users doing bad things—accidentally or intentionally.

Every punishment or resolution is downstream and reactive. This means higher customer service costs, and drivers resorting to "street tactics".

If an ounce of prevention is worth a pound of cure, then what's the real cost of not experimenting with product psychology in those moments?

P.S., if you wanted to see how much progress Uber has made in 3 years, revisit this study.