Why no one knows what Uber Drivers make

I drive for Uber and Lyft in Denver, Colorado and no one can agree on what I get paid. A recent study claimed that drivers in Colorado make $5.49 an hour after expenses. Uber, Lyft and Doordash responded to this with claims that drivers average $37, $35, and $25.

Rideshare driver pay, rights and employment classification have received a lot of coverage lately, but this lack of agreement on basic facts is the elephant in the room. One side says $5.49, the other says $37. Imagine if employees at McDonalds claimed they were paid $1 an hour while the management said $25 an hour. There would be a way to verify the truth. How can we proceed with this conversation starting with such contradictory facts?

We don’t know what drivers make because Uber and Lyfts’ pay structure is deliberately confusing and inconsistent, and the companies do not release comprehensive data. My own pay fluctuates wildly, and new evidence shows that different drivers are offered unequal pay for the same amount of work. A study by Professor Veena Dubal coined the word “gamblification” to represent how the mechanisms determining driver compensation mimic casino gambling. We grind for low base pay while getting hooked on big intermittent payouts. The purpose of this article is to use examples from my own Uber driving to show why driver pay is so unknowable.

Engaged hours

When Uber states we make $37 an hour they cite “engaged hours” which is meaningless without context. Engaged hours are hours with a customer in the car. This does not count time enroute to a passenger or waiting for an offer. Drivers return to the population center empty when stranded outside of areas of high demand, and drive home after late night rides that leave us stranded far away.

Engaged hours versus total working hours varies widely between drivers in urban, suburban, and rural areas, but if a driver spends half their day inactive $37 an hour will turn into $18.50. In urban areas with a high density of passengers requesting rides, inactive time may be lower than 50%, however it is disingenuous to say inactive time does not count as part of our shift, especially when the platforms aim to provide rides to customers outside of population centers.

Net/gross earnings

The $37 to $5.49 disparity can further be explained by gross versus net earnings, with Uber citing income before expenses while the other study subtracted average vehicle expenses drivers must cover. When driving a personal vehicle each mile represents depreciation against a car we must replace when the time comes. Gas or EV charging are an easy to measure expense, with maintenance and depreciation being harder to know with certainty. When driving full time even a flat tire instantly eats a big chunk of perceived earnings and we are out of work with no safety net until it is fixed. The IRS mileage reimbursement is $0.65 a mile, indicating they believe each mile of driving costs the average driver $0.65. We started with $37 an hour that could become $18.50 with inactive time included. If each mile costs us $0.65 we begin to see how $18.50 begins to slip closer to $5.49 in true earnings. But even pre expense pay per “engaged hour” is inconsistent and confusing!

Surge pricing

Anyone who has taken Uber or Lyft after a major event knows about surge pricing, when a $10 ride suddenly costs $50 or more. Surge pricing has a massive impact on driver pay. Surges on both Uber and Lyft light up the maps in shades of pink and red that correlate to premium pay. Here is what a massive surge looks like on each app.

Those maps are straightforward. There are more passengers looking for rides than drivers available so each trip pays more. More often surge is less straightforward and predicting it is like watching clouds that may dissipate or build into a storm:

The first could slowly build and swallow the map; it could also vanish. In the middle image downtown is saturated with drivers, leaving lower base pay in the densest area with bonus offers out in the suburbs. On the right it seems to be raining down sweet bonuses on everyone except me, something that sends many drivers into conspiratorial rage when it happens daily.

Conventional wisdom advises new drivers against chasing surge because these red maps bait us in then disappear quickly. However it is very hard to resist giant dollar signs daring us to chase them, and every driver will at some point race towards a bright red $30 a few blocks away only to watch it vanish as we arrive. The gambling metaphor may be too generous as it turns into Squid Games. We race through dangerous conditions in a last ditch effort to pay off debts and find the American dream.

Upfront pricing and trip radar

However the surge map is a piece of the puzzle that seems to matters less and less. The big change that rocked the rideshare community in 2022 was Uber introducing an “upfront pricing model” that uses artificial intelligence algorithms to set driver pay. Lyft then did a mediocre impression of Uber’s system as they do.

Before this change, drivers couldn’t see our destination until we picked up a passenger and began the trip. We were paid according to a standard formula based on miles and minutes, with surge added on top. You can see here how a pre upfront fare to Denver airport was calculated at $.20 per minute and $.61 per mile.

If I had landed on a $10 surge we would add 10 to this formula $45.38. We were paid consistently, but we were in the dark when we accepted a trip: it might be a 5 minutet trip or an hour and we had to show up to know what we were getting.

At the heart of the Uber and Lyft business models is a contradiction that is central to the current debate over drivers rights and classification. As independent contractors we are not offered benefits or job protection. In return we work on a soley at will basis, allowed to work whenever we want with discretion to say yes or no to any job. We are effectively unemployed, then offered a contract for one ride, then unemployed again until we accept our next contract. The contradiction is this: Uber and Lyft need to offer consistent service to customers but are not allowed to force any driver to accept any particular job.

Before upfront pricing drivers lacked basic information in deciding what offers to accept. The change to upfront pay gave drivers information but with a huge catch. With upfront pay there is no standard formula that explains what we are paid.

Instead the price offered for a trip is set by machine learning algorithms in response to supply and demand, and what drivers across the country have observed is constantly fluctuating pay that does not correlate to the amount of driving we do. Drivers get bombarded with trip offers to make decisions on. This is what we see for a few seconds:

In these seconds the driver must look at miles, minutes, and dollars and accept or decline the offer. Once we get good at reading this information quickly we can notice direction, prioritizing trips towards busy areas. You might decline if the passenger has low ratings from previous drivers, and if you analyze all that fast enough you may even notice that both of these rides include stops. Drivers are paid around $.20 a minute for wait time when customers run errands, so added stops at busy times eats into our earnings.

The promise of upfront pay is the promise free market economists make about an invisible hand creating a fair outcome for everyone. The AI will offer a price for a trip and the price will go up or down depending on who is willing to take it. In principal, the price should go up or down until it is considered fair. If no one wants to leave downtown to a far off suburb the price should go up until someone takes it. This key change has been muddled and is crucial so let’s repeat it succinctly:

Pre upfront pay: Drivers knew how pay was calculated but didn’t know where we were going or how much we would earn until we accepted

Upfront pay: We see how much we will get paid and where we are going, but we don’t know how our pay is calculated.

Wildly unequal fares

The key to the controversy around upfront pay is that drivers are finding upfront pay offers to be unpredictable, confusing, and often insultingly low. One Sunday afternoon I was in Boulder and I got these 3 offers to drive to the airport at 2:30, 2:33 and 2:58. The first is the most work but pays the least. The next two are nearly identical offers but suddenly at 2:58 the pay nearly doubled.

The latter is what drivers call a “unicorn offer” because it makes us want to believe in fair pay, something we now consider as rare as this mystical beast. I’ll embrace this terminology, though a unicorn ride is not mystical but instead a deliberately calculated payout like those programmed into slot machines.

Let’s look at various other offers for rides to Denver airport. Look for the unicorn.

What makes the $61.34 trip worth so much more? It is nearly identical in mileage to the $23.32 trip, and the $36.09 trip takes up the same amount of time while costing the driver more mileage depreciation and fuel. Can anyone justify or explain these prices? They are set by our artificial overlords who claim they are paying us fairly; are we expected to take what we get and trust the robots to treat us fairly? Kneel before the almighty algorithm and don’t ask questions!

You may notice the $4.25 includes surge. It does, but surge no longer correlates to high or low pay, as we can see in this offer.

It includes a supposedly higher surge but does not reward the driver with unicorn pay. We talked about how drivers race towards dark red numbers that are supposed to represent higher pay. Yet this trip pays in line with the worst offers above despite the $5 bonus. The unicorn contains a $4.25 surge and pays $61.34 while this one offers pay in line with the worst offers despite a $5 bonus. Under upfront pay Uber incentivizes drivers to waste fuel and time to land smack in the middle of the surge map, yet the pay we are offered no longer relates to the surge we have been promised. Uber refuses to explain to drivers what factors account for this inconsistency except vague statements about supply and demand.

Typical non unicorn offers

Here are 18 regular non-surge trip offers from Uber. I receive a dollar extra per trip for driving an electric vehicle, so most drivers are offered a dollar less than these. Short trips in Denver generally pay $5-$7 for 10 to 30 minutes of work. You can see one 50 minute offer for $26.68, while another offers $17.42 for 52 minutes. I added up the pay for this assorted collection of short and longer trips and they pay $249 for 611 minutes of driving, or $24 an hour. If you subtract the $1 EV incentive it comes to $22.70 per hour. Even without accounting for our expenses and inactive time we are well short of the $37 average Uber claimed.

More unicorn offers

These trips all pay around a dollar a minute or more. But how often do these trips come along? And who gets these offers? Are they equally distributed to all drivers in one market or do some get them all day while others never receive them? With no formula or transparency in knowing how pay is set and surge not correlating to higher pay we are operating in a rigged casino, hoping to make ends meet from bonuses that we cannot predict even by chasing surge.

What accounts for this inequality in pay within one city? With so much unexplained inconsistency we have no idea what the “average” driver truly makes, or if there even is such a thing.

Weekly bonuses

Another factor that contributes to unequal and unknowable pay is weekly bonuses, another example of gamblification. At the beginning of the week and weekend we select a quest, a number of rides we will try to complete to get an all or nothing bonus. Here are 2 examples:

The bonuses offered are inconsistent from week to week and sometimes there are none. The first gives $300 for 90 trips, while the latter gives $190. We gamble before starting the bonus period on how many rides we think we can give. These bonuses are for completing these trip from either Monday-Thursday or Friday-Sunday. Let’s say a driver selects the 80–90 weekend offer. They will need to average 3 trips an hour for 30+ active hours in 3 days. You cannot do 80 trips in a weekend by taking long rides in traffic, so the driver will need to relentlessly seek out quick trips that pay $5-$7, knowing these trips will become $9–11 trips once the $300 extra pays out. A driver who hits their bonus will have higher hourly earnings, but many of us often fall just short. There is nothing more maddening than taking $5 rides all week only to narrowly miss out on a bonus that would have bumped our hourly pay up and there are many factors that can knock a driver off their pace of giving so many rides.

Upfront pay and driver behavior

Drivers have had mixed reactions to this unpredictable upfront pay. On one side you have drivers who claim you cannot outsmart the game and must take whatever is offered. They essentially say “You just have to put in the hours and keep grinding away. You’ll win some rounds and lose others, just accept trips and it will balance out.”

On the other hand you have drivers who take pride in saying no to lowball offers and waiting for unicorn pay. I recommend the Show Me the Money podcast with Sergio and Chris to learn more about driver issues and this viewpoint. They advocate strongly in favor of cherry picking the best rides under the banners #knowyourworth and #declinegarbage. The cherry pickers treat every low offer as a chance for a mini protest, saying no again and again until they find a cherry.

On a systemic level cherry pickers argue that upfront pricing gives us a way to collectively bargain and demand higher pay without actively striking. If every driver in Boulder agrees to never take an airport ride for under $50, the algorithm has to either offer us this minimum or not maintain service. The CEOs of both Uber and Lyft have implied this approach may work by saying that if we decline more rides they are responsible for raising pay until we accept the majority of offers. Most drivers I personally speak to have lower and lower acceptance rates, indicating this is not happening. Here is my declining acceptance rate. But why would I accept $20 to drive to the airport when I am sometimes offered $60 or more?

The cherry pickers take pride in low acceptance rates and refusing to work for crumbs. While I personally support this approach both have their drawbacks. The “just put in the hours” driver must often accept insultingly low pay for long periods of time. But the cherry picker is at the mercy of hoping cherries appear. The $61 Boulder airport offer appeared right before 3 PM one Sunday afternoon, and I returned to the same spot the following Sunday hoping it would come again. The AI trained me to expect a reward there yet I waited and waited and never got a repeat offer. It resembles a trained dog getting a treat at random intervals; I am conditioned to race to a spot where I sometimes get rewarded. When I don’t get the reward I must either settle for low pay or not work at all, a decision obviously impacted by my financial circumstances.

We also now know that the AI is offering each driver different pay, as Sergio and Chris demonstrate in this video that shows 2 drivers on the same couch getting consistently different offers. We are being paid unequally and if this is merit based we certainly don’t know the criteria. Do more selective drivers get better offers because the machine learning eventually realizes we won’t work for crumbs? If so they are rewarding those of us who provide less service while punishing those who accept more offers and provide consistent service. Yet if one driver always accepts $25 for the airport trip why would they be offered an extra $35 when the machine has learned they will always work for less? I fear this is the future of work, with machines rewarding, punishing and manipulating each individual into working for less money, with no criteria shown to us on which we can improve.

We need to start with transparency

When I first began driving I got addicted to the game, getting an exhilarating rush whenever I landed in a red bonus zone or finished a high paying trip.

However I have now met the reality that base rate does not pay a living wage in our expensive city, and surge pay cannot be relied on since it comes and goes with little predictability. I have gone out on snow days and made $60+ an hour, but I have also driven in dangerous snowy conditions with surges never appearing, taking additional risks to play the lottery and come up empty. Urban driving in snow storms is dangerous and scares most drivers off the roads; one would think driving into this to get people safely home merits consistent premium pay rather than a slot machine shot at it.

Some unequal pay is merited. Uber needs to get drivers to areas of high demand, and picking passengers up at an NFL game should be worth more than driving in low traffic suburbs with low demand. But the current structure offers insultingly low pay for services that are still needed. People traveling at non peak times still need transit to the airport, but the low pay offered will not entice a driver who expects 2 or 3 times more and knows the trip may not even cover their costs. I feel as though their goal is to chew through new drivers who take low pay without knowing what they should be making, only offering fair pay to experienced cherry pickers when there are too few drivers available who will work for less. As I have seen the inequalities in pay increase under upfront I no longer feel like my job is about providing service to customers who need it but rather is about outgaming the system, analyzing my timing with increasing obsessiveness to chase the unicorn.

I will be out late at night in bad weather and will sit and wait, declining ride after ride with mocking disdain at lowball offers. I will reflect on how distorted my job has become. As a green newbie I took pride in helping everyone get to their destination safely, now I say no again and again hoping to outsmart the AI by timing my driving impeccably to catch the high payout. I believe waiting for high pay is rational given the system, but I sometimes reflect on how each lowball offer represents a human in need of a ride. Perhaps they are leaving a bar and can wait outside with friends while I battle AI for a fair price, but perhaps they are stranded in the cold, alone, waiting for a driver to accept a trip that the AI underpriced. Their phone dies before a driver accepts this race so they cannot access the app, all while we sat and said no to garbage offers awaiting fair pay. Perhaps they are in a domestic abuse situation and desperately need transit away from the perpetrator, or will lose their job if their daily ride are delayed due to long wait times. In what neighborhoods are trips priced higher and lower? Uber sometimes deactives drivers because their AI believes we are discriminating against low income neighborhoods, but their algorithm is itself creating the inequality. If it pays me 60/hour to drive in a wealth area and 15/hour to drive in a poorer area, why would I make the irrational decision to work there? People rely on these apps to meet so many needs, and Uber’s AI alone sets the price we will accept or reject, deciding who gets a ride quickly and who does not. Uber and Lyft have become an indispensable part of our transportation infrastructure in a car-centric world, and it has morphed into a strange battle of working class drivers trying to outsmart a mysterious robot overlord to eeke out fair pay. In the long run I fear how much worse this can get if not reigned in soon.

In discussing driver rights and pay we must start with transparency and accurate facts about driver pay. Pay can fluctuate some based on market forces, but we deserve to know why we are sometimes paid fairly and more often not. We deserve to know why a trip that usually pays $35 goes down to $25 one minute and up to $60 the next. We also deserve to know the spread and level of inconsistency. Is one driver getting $60 an hour while another gets $25 for the same work? Is this correlated to race, gender, acceptance rate, the car we drive, or how desperate the AI believes us to be? What should the driver making $25 do differently to earn $60? Uber’s reluctance to release detailed information on pay and bonus structures indicates they are hiding something. The companies have shown that if left alone they will leave driver compensation as an indecipherable mystery. They will not share information willingly, so labor rights advocates and government officials need to demand it. We deserve the right to choose to drive or not based on knowing the rules of the game, and not base our livelihood on a casino game that sometimes pays and sometimes doesn’t. I call on Uber, Lyft, Doordash, and others to stop hiding behind vague statements and mysterious algorithms and be honest about what is going on, so we can make informed decisions about when, where and how to work without being at the whim of a robot overlord’s whims and manipulation.