The big idea: Charge less for low-rated Uber drivers
Uber regularly kicks off drivers that fall below a ratings threshold. What if they made them cheaper instead?
Uber’s ridesharing system is pretty amazing. Through some algorithmic alchemy, it turns a complicated network of individual drivers, car types, locations, and destinations into a single price for transportation.
But you may have noticed that you’ve never seen a 4 star driver on Uber. That’s because Uber regularly culls drivers who fall too low on the ratings. It’s not just terrible drivers that get cut; during my brief stint there a decade ago, a 4.5 rating was enough to risk removal from the platform.1
It makes sense why Uber does this, but it’s still a waste. These drivers passed the background check and are ready to drive; even if Uber wants to keep the quality of the platform high, they’re throwing away a valuable resource.
What if instead of removing these drivers, Uber just charged less for them?
Removing drivers creates deadweight loss
Uber works because both sides know almost nothing about each other. If executing a transaction required both sides to assess things like the quality of the car, driver safety, demeanor of the passenger, or tipping frequency, the number of rides would plummet from cognitive overload.
Instead, all of this information is aggregated into two things: a product tier (UberX, Uber Black, etc.) and a 5 star rating. This transforms a ride into something closer to a commodity trading at a market price. It’s not so different from an econ 101 model of perfect competition, with Uber adjusting prices to make sure the market clears.
If you buy that Uber acts like a commodity, then the removal of low-rated drivers is functionally a supply shock. It might be a meaningful difference.
Of the drivers who do at least 1 trip, 75% reportedly churn within 1 year.2 Some quick math:
With an estimated 1.7M new drivers between 2022 and 2024, a 75% churn rate implies that 5.1M drivers did at least 1 ride before leaving the platform.
If just 5% of those drivers were removed for poor ratings and would have otherwise kept driving, that’s over 250,000 additional drivers.
If those drivers averaged 1 trip a week at $15 per trip, nearly $200 million in annual bookings disappear into nothing.
Sure, not every one of these rides would have been additive. But up to $200 million in deadweight loss sure seems like it’s at least worth looking into.
Moving beyond premium tiers
The idea: Uber Discount, an Uber tier made up entirely of drivers between 3.5 and 4.5 stars. These rides would be discounted up to 20%, in exchange for the rider almost certainly getting a worse experience.
This could be a win-win for drivers, riders, and Uber.
Riders get a cheaper option, allowing price-sensitive customers to take rides that otherwise wouldn’t happen. There’s growing evidence that price hikes have begun to affect rideshare demand — offering better segmentation might bring some of those price-sensitive riders back.
More price discrimination might be a real opportunity for Uber. An experiment in Egypt found that a 50% discount quadrupled demand for rides. Rideshare is an extremely price-sensitive category, which has presumably slowed down growth in cities like New York that face above-average prices. Flexible pricing might let Uber restart growth in mature markets from potential riders that are currently priced out.
Uber Discount also lets drivers make a real tradeoff between their reimbursement rate and improving rider experience (which often isn’t free; think going to the carwash or offering water). For drivers on the edge, the option to fall into Uber Discount means more space to identify what’s getting them low ratings and address it. Uber can even help this process by analyzing poor ratings and giving personalized advice to drivers.3
Of course, these drivers often get bad ratings for a reason. An (admittedly out of date) 2014 leak showed that only 1% of ratings were 1 star, and only 5% were 3 stars or lower. When drivers do get bad ratings, it’s often for reasons like poor driving, being rude, or browsing their phone while going 80 MPH down the highway.
This is a pool of drivers that not only provides a worse experience, but also adds legal liability for Uber as the coordinator. It’s only prudent to require riders seeking a deal to sign a waiver before taking an Uber Discount ride.
The waiver not only shields Uber from legal liability, it signals to the rider that they should expect a worse experience. That means that they’ll either be pleasantly surprised when their driver doesn’t run a red light or more accepting of a driver offering them a cigarette.4
Ok, maybe a checkbox isn’t enough. Even if we assume that covers legal risk (a big if), Uber relies on a massive commercial insurance policy for its drivers. I’d expect that the insurer would still require that Uber remove drivers that regularly get pulled over for reckless driving, even if they let them keep the drivers that just provide a bad experience.
Maybe that’s a blessing in disguise. Taking an Uber Discount is probably an easier sell when the main risk is a dirty car vs the possibility that your driver will run red lights.
Riders could face ratings consequences too
Uber uses a two-way rating system, so it’s only fair that riders also have their experience affected by their rating.
This already happens informally. Anecdotally, drivers often refuse to pick up riders with a rating below 4.6, causing longer wait times and difficulty getting picked up. Instead of forcing them to trade on wait time, why not let them trade on price?
Much like drivers, riders that fall below a 4.6 rating could have a 20% Painful Rider Surcharge (PRS) added to their fare estimates. This surcharge compensates drivers for the pain of taking these specific passengers, sort of a hazard pay for “this guy is going to watch YouTube videos without headphones during the ride.” This might actually be a better experience for these riders vs having drivers repeatedly cancel on them.
You may have noticed that the PRS perfectly offsets the Uber discount fee. The PRS serves a secondary purpose of seeding the Uber Discount market. Passengers subjected to the PRS will be more likely to seek out Uber Discount vs a traditional Uber, seeding demand for the new service.5
This system has real synergies with Uber’s other lines of business. It’s not so hard to imagine Uber Gold freezing customer ratings, or giving them a PRS discount. Letting users prepay the PRS at a discount with Uber Cash is a great way to lock in purchases on the platform. Uber Eats can offer PRS waivers if you order enough delivery.
Most importantly, the PRS needs revenue sharing with the driver that’s being asked to stop at the McDonald’s drive-through in the middle of a 30 minute ride. It’s only fair.
There could be some unexpected behavioral changes
If the discount pool really takes off, the population specifically seeking discounted rides may begin to cannibalize today’s UberX population. That could reverse the value of a rating; with enough demand, getting reliable rides could be worth a 20% price cut. In extreme locales, drivers might start intentionally speeding, watching NFL on their phone, and coughing aggressively to kill their rating and stay in the cheaper price tier.
Or, if falling into the discount pool is too rough, drivers might more aggressively push for higher ratings. Optimistically, this could lead drivers to be extra intentional about the riding experience they’re giving. But it could also mean drivers asking riders to give them 5 stars immediately when the ride ends, using social pressure to artificially inflate their ratings.
On the rider side, charging more to low-rated riders means that they’ll be more protective of their reputation. That’s good in some ways — they’ll be more polite — but it may stop some socially desirable behavior. I’m still a bit sour over paying for my drunk friend’s Uber home in 2014, then waking up to my perfect 5 star rating being ruined.6 If it got me into the PRS pool I might have thought twice about it (or at least Venmo’d her for the cost of the Uber).
This has some negative effects on Uber’s business too
There’s theoretically a lot to like about Uber Discount. By allowing more dimensions of trade, Uber creates more space to find unserved rides and make them happen.
But it also might have some unexpected consequences for the platform.
I imagine that Uber lives in fear of bad customer experiences getting into the public consciousness. While Uber Discount could get a lot of positive attention — really crappy rides with Uber discount could go viral in a fun way — it’s more likely that the cost of negative experiences will outweigh anything else. While a story about someone getting yelled at by their Uber driver might push users back to UberX, it’s more likely to push them to Lyft.
More concerningly, pricing discipline becomes significantly harder once a discount tier is introduced. Some of the purchasers that move to Uber Discount would have otherwise just sucked it up and paid for the UberX. A discount tier anchors users to see UberX as relatively more expensive (I imagine this is why Uber Pool never really recovered after COVID).
But the biggest impact might be on the rest of the market. Uber relies on a small number of highly engaged, savvy drivers that do thousands of trips to make sure that general demand is covered. Even if Uber Discount brings net new rides, it’s likely that it will reduce demand for UberX. If these high-value drivers start to see their earnings drop, Lyft will be ready to swoop in and take them.
It’s a tricky balancing act, and likely means that Uber Discount is only feasible in markets with a general shortage of drivers. If it launched, it’d come to Omaha before New York.
Official idea rating
2.5/5, maybe worth trying in a test market. I think there’s real potential in expanding the supply of drivers, and giving people a little more optionality on the low end. Uber specializes in turning data into price signals; it only makes sense that they could lean further into their strength.
Still, I can’t quite see this launching in a mainstream market. People have already anchored on the 4.6-5.0 star system. Even if the legal, network, and pricing issues weren’t a problem, changing that consumer behavior is a tall lift.
But maybe there’s value to de-anchoring ratings. It’s probably not ideal that all of our platforms have 5 stars as the standard, and anything below a 4.7 as an implicit failure.
Platforms compensate by adding new layers on top, like Superhosts on Airbnb or Top Rated on eBay, but it seems silly to create endless sub-tiers within the 0.4 star variance we limit our transactions to. Maybe Uber Discount would make it easier to just say that sometimes the experience will be cheap and mediocre, and that’s ok.
I was in a second-tier city with driver supply issues, I imagine the cutoff is even more strict in places like San Francisco with lots of drivers
Like a lot of our data, this is from 2017. Definitely a little out of date, but the best we have
Back in the early 2010s we would sometimes do coaching at office hours for drivers to help them get their ratings back up, but it was questionably effective
This did legitimately happen to me once; I turned down the cigarette and gave him four stars
This might actually not be so different from the world today. Uber’s original matchmaking patent seems to take both parties’ ratings into account in the matchmaking process.
I haven’t forgotten, Lydia
Prices could even go negative on the worst drivers -- they pay for willing riders who could rehabilitate their ratings. Survive a trip to LaGuardia driven by a cannibal with one eye and three DUIs and get $10.