A Deeper Look at Uber’s Dynamic Pricing Model

The following is a guest post by Bill Gurley, a General Partner at Benchmark and Board Director at Uber, which originally appeared on his blog, “Above the Crowd.”

Over the course of the past year, many writers have offered their perspectives on Uber’s dynamic pricing strategy. Perhaps the only consistency is that people have deeply passionate views on this topic. However, there are still many misperceptions about how the model works, and the purpose of this post is to clarify some of those misperceptions.

I am an Uber investor and board member, and therefore expect that many will dismiss these thoughts as naked bias. But consider that as a result of my role I have access to more information that might enable a deeper perspective. I also have quite a bit on the line, and as a result have spent a great deal of time contemplating the policy as well as the potential alternatives.

CLARIFYING CERTAIN SPECIFICS REGARDING UBER

Before diving into Uber’s dynamic pricing model, it is important to clarify some of the key elements of how Uber is structured and operates.

1) Uber is a marketplace and Uber’s drivers are all independent agents. Uber’s drivers are independent agents that are either self-employed, or work for someone who owns multiple cars. Uber does not own cars and does not employ drivers. Each day, and each hour for that matter, these drivers decide whether or not to open the Uber application and accept requests for rides from Uber customers. These drivers are not bound by exclusivity. Many of them work on multiple services, and many have “regular customers” that they engage off the Uber platform.

2) The majority of Uber fares go to these independent drivers. On average, over 80% of gross fares end up in the hands of drivers. What’s more, of the percentage that is retained by Uber, a large portion goes to cover variable expenses within the service. These expenses include payment processing, payment fraud, refunds, customer support, dispute resolution, cellular handsets and service fees for the drivers, and local regulatory efforts. The bottom line is that this is a low margin business — much more akin to Amazon than Google.

3) Uber is a committed to being a low-price leader.  Some have suggested Uber is a “luxury brand.” This confusion is understandable, as Uber’s initial focus was on traditional black-car services. However, since its launch of the low-priced uberX brand over 18 months ago, Uber has been laser-focused on leading at all price points. This goal is to expose Uber’s amazing customer experience to a much broader base. UberX is now the company’s most strategic and fastest growing offering, and has become the company’s largest offering in many cities passing UberBLACK in number of daily rides. The company has also intentionally worked to lower price on uberX as often as it can (in some markets it has already lowered price four times). Despite this, some competitors still hold on to the convenient delusion that Uber is solely a high-end service, perhaps akin to CDNow assuming Amazon was only a bookstore.

4) Uber’s dynamic pricing (“surge pricing”) affects a tiny minority of all Uber rides, less than 10% of trips. Dynamic pricing is most common on peak times on Friday and Saturday nights, on certain Holidays, such as Halloween and New Year’s Eve, and during particularly big events and bad weather conditions. All told, it’s a fraction of the time that Uber drivers are operating. The vast majority of the time, Uber’s increasingly low basic rates (uberX is often 40% cheaper than the local taxi alternative) are the primary price points for the service.

5) Uber is remarkably transparent about its dynamic rates. Ever since the company first encountered feedback about its pricing model, the company has gone out of its way to make sure that customers are aware of the policy and how it works. They have inserted special splash screens where the customer has to key in a specific confirmation of the increased rate. Additionally, the company has gone out of its way to publicize how and why the program works. This past New Year’s Eve, the CEO even published a how-to video encouraging riders NOT to ride at certain times.

THE ORIGINS OF UBER’S DYNAMIC PRICING

So why does Uber’s dynamic pricing even exist? The answer lies in understanding that Uber is fundamentally a marketplace, where supply is controlled not by the company but by the legion of independent contractors and transportation providers with whom they work.

Back in early 2012, Uber’s Boston team noticed a problem. On Friday and Saturday nights, around 1am, the company was experiencing a spike in “unfulfilled requests.” The root cause was that drivers were clocking off the system to go home, just before the weekend partygoers were ready to venture home themselves. There was a supply-demand imbalance, and the result was a lot of very unhappy customers. So the Boston team had an idea. What if they offered the drivers a higher price to stay on the system longer (until around 3AM)? Would more take home dollars for drivers increase supply? In just two weeks they had a resounding answer. By offering more money to drivers, they were able to increase on-the-road supply of drivers by 70-80%, and more importantly eliminate two-thirds of the unfulfilled requests. The supply curve was highly elastic. Drivers were indeed motivated by price.

Based on the results from the Boston experiment, Uber implemented its dynamic pricing policy to be used solely when demand is materially outstripping supply. Dynamic pricing changes are driven algorithmically when wait times are increasing dramatically, and “unfulfilled requests” start to rise. In essence, there are two functions of the increased price model. One is to increase supply. The second function of the price increase is to temporarily intentionally reduce demand. Through these two mechanisms, the company is able to (a) increase supply, (b) assure reliability, a key tenet of the company, and (c) maximize the number of completed rides.

ECONOMICS 101: SUPPLY AND DEMAND CURVES

If you were to pick up a copy of any introductory economics textbook, in either the first or second chapter you will find a description of the supply-demand curve. It is the key operating model for economic analysis, and is as fundamental to economics as DNA is to biology. If you have never been exposed to these topics, I highly suggest Khan Academy’s Microeconomics courses. As one might expect, the demand and supply curves are the first two sections.

There are a few things to note about Uber’s marketplace and the supply-demand curve. First, Uber’s analysis and research has shown that both the supply curve and the demand curve are highly elastic. The Boston experiment, and every effort since then, confirms that higher prices increase supply, all things being equal. This is true in every market the company has entered.

On the demand side, the company has confirmed price-elasticity in two different areas. First, when prices surge, they see an immediate reduction in open-to-order ratios. As expected, higher prices do indeed reduce demand. Moreover, as mentioned in this video interview with Uber’s CEO Travis Kalanick, the converse is also true. Uber’s numerous price decreases have all resulted in materially increased demand. Basically, Uber’s marketplace is highly efficient, and operates in the exact way that your economics professor would expect. When you consider that both sides of the model – Uber’s riders and drivers – are both large groups of fragmented independent agents, this is no surprise. The market should operate efficiently.

Using the supply and demand curve as a model, Uber’s dynamic pricing model is rather straightforward. When demand outstrips supply, dynamic pricing algorithms increase prices to help the market reach equilibrium. Of course, these situations are always temporary, eventually supply outstrips demand, and the price falls back to normal. If demand were to spike with no resulting price increase, you would have what is known as an economic shortage. Without a price increase, Uber’s unfulfilled rate would skyrocket, and most customers would be left without a ride. With dynamic pricing however, the variable Q on the graph is further to the right than it would be without.  More absolute rides are fulfilled precisely because supply increases.

COMPARING UBER TO HOTELS, AIRLINES, & RENTAL CAR COMPANIES

Many of the articles about Uber’s pricing model have noted that airlines, hotels, and rental car companies frequently use dynamic pricing, and typically at ratios (10X differentials in these industries are common) that are very similar to Uber’s highest peaks on New Year’s Eve. For a hotelier, the demand for a room on New Year’s Eve is dramatically higher than a random weekday two weeks later. With no ability to increase supply, they are left with the alternative of selling to the highest payer. This is a relatively well-understood and accepted practice. No one appears even the least bit emotional about it, as it is well understood and expected.

There is, however, one key difference that materially increases the need for dynamic pricing in Uber’s case. With hotels, airplanes, and rental cars, supply is relatively fixed. One cannot build more rooms for New Years Eve, and then take them down. Uber has a problem these companies do not. At the exact time that riders want more availability – Friday and Saturday night, in a bad storm, on New Years Eve – drivers would rather not be driving. You see, while hotel rooms are fixed, Uber’s supply actually shrinks at these times, because the drivers would prefer not to be working at those times either. The exact events that increase demand for needing a driver also cause supply to shrink. In these cases the supply curve is moving left at the exact same time that the demand curve is moving right. As a result the need for a price catalyst to increase supply in the Uber case is vital.

Another factor that impacts driver supply is substitute opportunities. Drivers have lucrative alternative opportunities on event nights like New Years Eve. Some party-goers are willing to book a single captive driver for a flat rate which could be well over $1000 for the night.  And in this case, the driver enjoys quite a bit of downtime.

THE ONLY REAL ALTERNATIVE – “NO CARS AVAILABLE”

Some have argued that they understand the economic underpinnings of Uber’s dynamic pricing, but they suggest that the “PR hit” is so great that the company should reconsider its policy. This single-step analysis fails to consider the real alternative to dynamic pricing – tons and tons of unsatisfied customers complaining about a lack of availability/reliability (two of the key tenets of the Uber customer proposition).

Screen-Shot-2014-03-07-at-6.41.40-AM

The bottom line is that the only real alternative to dynamic pricing is a ton of customers staring at screens that read “No Cars Available.” This is the fact that is least appreciated by Uber’s critics. Remember thePR hit that UPS took this Christmas when they could not deliver packages as customers expected? The idea that Uber could make its network available in a normal state and at a normal price while demand is increasing and supply is shrinking is quite simply unfeasible. And the argument that Uber should keep prices flat in moments of peak demand is a de facto argument that Uber should be comfortable stranding the majority of its customers with the disappointing message: “No Cars Available.”

If this sounds to you like an apology for the price increases, try the following experiment: the next time you see a message indicating that Uber’s surge pricing is in effect: immediately try an alternative other than Uber. In other words, try to hail a cab, call a traditional black car service, find a rental car, or jump on a bus or subway. You will find that availability and reliability for all forms of transportation are under stress at that same precise moment in time. At these times, a fixed price taxi will be highly unavailable, and a fixed price subway will be remarkably over-crowded. Relative to those choices, Uber prefers to be reliable and available for the maximum number of customers it can serve, and believes that the customer dissatisfaction from being unavailable would be way worse than the limited customer dissatisfaction with their dynamic pricing model.

DRIVERS ARE PEOPLE TOO

Another point to keep in perspective is that the operator of each and every car on the Uber service is a human just like all of the passengers. Why should we expect that individual to be excited about working precisely when we want to be out of the town? Do you enjoy working on Friday and Saturday night? What about Holidays? What about New Year’s Eve? Nurses and doctors routinely receive 2-3X overtime pay for work at those times; is there a reason that a driver should not? What about during really bad storms? Should the independent driver be more concerned about your needs, or those of their own family and friends? This is not a plea for you to be overly empathetic for the independent drivers on Uber’s system, but rather simply asking you to consider the basic human reasons why they may chose not to drive at the exact same time that you are most interested in not driving. Is it unreasonable to expect higher fares if they are sacrificing their own time at the least convenient moment?

UNDERSTANDING WILL IMPROVE WITH TIME

Uber is indeed sensitive to the perspective that some customers have about dynamic pricing, and this is precisely why they work so hard to message awareness to their consumers. Just last week, the company announced a new feature called Surge Drop that will notify the customer when surge prices have fallen, enabling someone to wait until the supply and demand are more in balance, and avoid the higher fees.

As many have noted, dynamic pricing policies are well understood in hotels, airlines, and rental cars. This awareness makes the changes expected and less surprising. Increased awareness will help to reduce the element of surprise that some customers may encounter. It will also cause certain customers to steer clear of time windows when prices are higher, and when they would then prefer not to transact. Basically, as more people understand the model, they will become more comfortable with what to expect, and can make informed choices.

Fundamentally, most critics of Uber’s pricing model fail to recognize that Uber is a true marketplace. The majority of leading Internet marketplace companies use dynamic pricing as a solution when confronted with a scarcity of supply. This was the fundamental premise behind Ebay’s original auction model. It is also exactly how things work on StubHub (ebay’s ticketing engine), as well as Airbnb and Homeaway. Additionally, it is the key pricing algorithm behind Google’s core Adwords offering — a service that has over one million customers, and takes in close to $50 billion in revenue. Its usage is tried and true which is partially why Uber’s CEO has confidence that it’s the right model here as well. Uber has no intention of abandoning dynamic pricing precisely because it is in the consumer’s best interest, especially when one understands the true alternatives.