I spoke on “Overcoming the Barriers to Building Great Products” at this year’s Silicon Valley Product Camp. In this presentation, I look at how great products generate superior financial returns even though they have equivalent or lesser functionality to their competition. I then present a hypothesis that building a great product requires making decisions that run counter to the quantifiable ROI requirements that almost every business endorses.
Uber’s surge pricing causes a ton consternation among for its customers and while surge pricing is a theoretically correct economic response to a shortage, in practice it makes for service that doesn’t meet the all the needs that Uber is trying to fuflill. The idea behind surge is pricing is sound on paper: when there’s more demand than supply, the price should rise so that market clears. The alternative is a shortage which would mean long waits for an Uber, similar to how its difficult to get a taxi in a period of high demand. The problem with surge pricing is that it forces the customer to be in a spot market for transportation. A spot market is where a transaction is made for immediate fulfillment or in simpler terms, an Uber ride can only be bought when you need it and the price fluctuates so there’s no way of knowing in advance (say a day or even an hour before) how much you’ll pay.
Consumers do not like spot markets in general and most of the time, their transportation needs are predictable rather than of the moment. The number of commodities that the average consumer buys at spot prices is quite small. Ones that come to mind are gasoline and food products like milk, fruits, vegetables and meat. The amount of complaining when gas prices rise is a good indicator of how much people dislike spot markets. There’s less complaining about food because there’s lots of substitutes. If the price of beef is high, one can always buy chicken or pork instead.
When it comes to transport, people avoid spot markets. When was the last time you walked into the airport and bought a ticket for plane departing in a hour? The problem with walking into the airport and buying a ticket is that the price could be very high. If there are few seats left, a flight that is usually $400 could be $1000. Since most of us rarely need to fly on short notice, we avoid the spot market and make a reservation in advance. If fares are high the weekend we want to fly to Vegas, maybe we go the next weekend or decide to take a road trip instead.
With Uber, we have none of these options since we don’t know what the price is going to be till right when we need the service. Say I bought an airline ticket for 4 weeks from now. I know 4 weeks in advance that I need transportation to the airport on a particular day and time, but if I want to take Uber, I have no idea what that will cost. If I open the Uber app and see it’s 3x surge pricing, I likely don’t have enough time left to take the bus instead. The same problem exists if I need to be a work at 9am tomorrow and my plan is to take is to take Uber. For much of most peoples transit needs, they know in advance when they will need transportation. That foreknowledge should be useful in making sure that supply and demand are balanced but with Uber, it’s not used at all.
With Uber, the riders are only one half of the equation. If surge pricing worked as intended, drivers would see the the high prices, start driving and increase the supply. For the supply to meaningfully increase, the number of drivers has to be elastic meaning that high prices need to actually significantly increase the supply. Uber drivers have a similar problem to riders in that they don’t know what the rates will be in advance. If an Uber driver is sitting at home in their underwear and watching re-runs on TV then 3x surge pricing might motivate them to put on some pants, get in the car and starting driving. However, if they made other plans or just decided to sleep in, then 3x surge has no effect on their willingness to provide rides. If they had known that it was going to be 3x surge pricing maybe they would not have agreed to go to brunch or stayed out late the night before but with Uber’s current system it’s all guess work. An experienced driver might know when surge pricing is likely and plan accordingly but if it’s unpredictable (either in time or amount), then potential increase in supply is limited to drivers who happen be sitting around doing nothing.
I don’t have access to data on how surge pricing affects supply. Uber certainly has this data but even under heavy criticism, they’ve never (to my knowledge) made any specific claims about how effective surge pricing is at increasing the number of drivers. Even without the data one could presume that if surge pricing was effective at bringing in drivers that surges above 1.5x would be rare as most people would not pass up the chance to earn 50% more than usual. In some geographies, 2x and greater surges are a common event.
New Year’s Eve is surge pricing at its most extreme. On NYE 2013, some riders were paying $500 for rides. Every business that is part of people’s New Year’s Eve festivities, especially bars and restaurants, raises prices on New Year’s Eve which makes sense because demand is so high and supply is fixed so it certainly makes sense that an Uber ride will cost more. Personally, I’m not a huge fan of New Year’s Eve and there’s an amount of income that would at least get me to consider driving for Uber on NYE. However, with the current surge pricing system, I don’t know what I’d make. While some revelers wouldn’t want to pre-plan what time they go home, I suspect many would prefer to book their ride in advance especially if it gave them certainty on the cost. I’m not going to give up my NYE’s because the earning potential might be good but is unknown and I suspect there are many people like me. Predictability would increase supply and make consumer better off.
Allowing reservations and still keeping an element of dynamic pricing so that supply and demand balance isn’t an easy problem. However if ride sharing services are going to fulfill their vision of enabling people to not own cars, Uber or its competitors will need to create a method where prices are known in advance for transport needs known in advance. While surge pricing might be economically efficient in a purely economic view, that’s too narrow a way to look at the problem. From a product point view, there’s customer needs not being met. If Uber can’t figure out how tell people what it will cost them to get from point A to point B a week now, somebody will come into the market and meet that need.
Google cut prices dramatically last week on it’s Infrastructure as a Service (IaaS) offering, Google Cloud Platform. Amazon followed suit on Amazon Web Services the next day and cut prices to match Google’s price cut. What business strategy lessons can we learn from these prices cuts and what can we predict about the future prices?
Commodity Products Compete On Price
Even though it takes both access to a large amount of capital and a high level of technical capabilities to run a large scale cloud provider, the cloud computing services especially compute and storage are a commodity as one providers offering is equivalent to anothers in the purchasers mind. There’s little difference in a virtual machine from Amazon, Google or Microsoft Azure.
As cloud resources are undifferentiated products, it’s not surprising that Google is choosing to compete on price. As a smaller player (in terms of cloud market share) hoping to take share from Amazon, Google needs to compete on price. Other attributes such as reliability surely matter but there’s not enough difference for Google to effectively compete on those.
As is frequently the case when when one player cuts prices, the other players respond with their own price reductions. Amazon responded by cutting prices to match Google for on-demand instances. Interestingly, there’s a significant difference in pricing structure for heavy usage. Google gives automatic discounts at 70% and 100% usage in a billing cycle. Amazon is cheaper when using reserved instances but they require up front payment and thus the buyer loses the option to terminate or reduce usage with out penalty. The reserved instance approach helps Amazon when comes to capacity planning but is less buyer friendly.
The other piece to note is that Amazon cut prices only on services where Google competes with them. Amazon left prices alone on services that do not have a Google equivalent such as Dynamo DB or the RedShift data warehousing service. Services where the competitors do not have competitive offering are differentiated and not a commodity.
What to Expect For Future Prices
For both compute and storage, we can expect costs to fall due to Moore’s law for processing and that hard drives get denser over time. Even if data center costs stay flat, providers being able to get more yield for a given amount of rack space and power means costs will continue to fall. The other question is will competition force lower margins over time? If so, prices could fall faster than costs as competition drives margins and thus prices provider lower.
As providers show no signs of being able to differentiate offerings in cloud storage and compute, I expect prices to continue to fall. With prices likely to fall, Amazon three year reserved instances become on a gamble since as realized price for Google’s offering may fall over enough over the next 3 years where it’s a better deal to use Google’s pay as you go price especially considering the flexibility to change or reduce consumption as needs change.
I spoke on ‘Business Strategy for Product Managers’ at this year’s Silicon Valley Product Camp. I was thrilled to see that there were many product managers interested in the topic as it’s not often talked about in a Product Management context.
Whenever one is evaluating an action to improve or maintain their business’ competitive situation, thinking through the competitor’s likely response is a critical element. I frequently see companies taking actions and not thinking about competitors response. A classic move to increase competitiveness is to cut prices. It’s a huge mistake however to evaluate your offering with lowered prices against your competitors offering with their existing prices. There will be a new equilibrium once competitors respond to your action. There maybe several rounds as some initially follow suit, and some wait and see and take action later. But assuming there will be no response and predicting the outcome using that assumption is foolhardy.
A recent example is Google Maps on iOS. According to many published reports, Google refused to offer Apple turn-by-turn directions in the Google powered iOS mapping application that was bundled with iOS 5 and before. Google’s refusal was so they could keep turn-by-turn directions as an Android only feature. In hindsight, this plan was a strategic blunder as Google lost a 23 million iOS users as a result. Even in foresight, this was predictably a blunder since it failed to consider Apple’s likely response. Was it really even possible that Apple would just accept this deficiency in their most profitable product line, the iPhone? It wasn’t necessarily predictable that Apple would create their own mapping app as there were many companies they could have chosen another partner instead. But what was predictable was that Apple was going to quit using Google as their mapping provider costing Google many million map users. Even though Apple’s switch to providing mapping was rocky to a say the least, Google still suffered. And, Google’s response to Apple moving to their own mapping system? To introduce that same turn-by-turn direction feature they originally refused to provide on iOS. Google had no choice but do so once Apple introduced their feature as losing the entire iOS user base greatly decreases the value of their local content.
It’s entirely possible there were other factors that lead to Apple’s decisions to no longer rely on Google maps which make it difficult to judge how big a blunder this was on Google’s part. Apple may just not have been comfortable relying on a competitor. But what is clear is that Google was never going to succeed in gaining any competitive advantage over Apple in the smartphone war by withholding any mapping features as Apple had alternatives for getting these features. The moral of the story is think through the likely response from your competitors when taking any action to boost your competitiveness. If someone else is proposing a plan, ask them what the competitor’s likely response will be. If you get a blank look, you know that they haven’t thought it through and their touted benefits are likely fantasy.
Tonight I saw Adam Lashinksky speak about his book Inside Apple: How America’s Most Admired–and Secretive–Company Really Works. As he was describing how Apple would erect walls inside its offices to keep projects secret , I realized this extreme level of secrecy solves a core part of the The Innovator’s Dilemma. It becomes impossible for one part of the company to impede the disruptive innovations of another when the first part has no idea what the second part is doing. Other pieces of what Lashinksky described about how Apple operate,s including having being functionally organized rather than having multiple business units, helps keep Apple innovating rather than getting mired in protecting its current businesses. But I’m wondering if the secrecy that was designed to keep information from getting outside the company also prevents Apple from getting it in own way.
Reaction to Netflix’s decision to spit streaming video and DVD rental into separate business has ranged from awe to bafflement. Clayton Christen who is the oracle when it comes to disruptive innovation says:
And while it’s always worrisome to disagree with a luminary like Clayton Christen, I’m not buying that streaming video is a disruptive technology for Netflix. The key element of a disruptive technology is that it’s performance is much poorer on some attribute that the incumbents dismiss it. Let’s looks at attributes of streaming vs. DVD:
|Time to content||30 seconds||minimum 2 days||Streaming|
|Content Consumable in Month||More than anyone should watch||Limited by DVD mailing time||Streaming|
|Video quality||HD – depends on bandwidth||Full 1080p with Blu-ray||DVD|
|Equipment Needed||Streaming enabled TV or Blu-ray player, Broadband internet||DVD or Blu-ray player||DVD|
|Payment model||Subscription (or pay per at competitors)||Subscription (or pay per at competitors)||Draw|
|Cost||$7.99/month||$7.99 and up, depending on number of discs, Blu-ray||Streaming|
Streaming’s instant access alone means most people find it superior. Given the choice of the same content library and same price, most consumers (at least the ones with solid broadband connections) would opt for streaming.
Many would argue the content available on DVD is far greater so that’s a strong advantage for DVD. However, that’s not an advantage of DVD technology itself but rather an artifact of US copyright law. As a matter of technology, DVD and streaming are both capable of delivering the same content. And the limitation on catalog is not an issue for content producers who want to enter the streaming video market.
Incumbents like Amazon and Apple embraced streaming video quite early on. The TV studios have gone so far as to create their own service, Hulu. Television networks like ABC, NBC and HBO, all offer their content via streaming through their websites and iPad apps. Cable companies like Comcast offer extensive video on demand services. I’m hard pressed to think of an incumbent in the entertainment space who did not see streaming video coming and enter the market. Retailers without the resources and technical know-how of Amazon are the only ones I can think of.
Looking at the combination of streaming technology not seriously lacking in any performance attribute and most incumbents in the entertainment space already entering the streaming market, I can only conclude that streaming video is a sustaining technology and not a disruptive one at least when it comes to Netflix’s existing DVD business. The cable and satellite TV business models of expensive packages of channels will experience declines as streaming grows and the reluctance to give up the large revenue streams that channel packages provide may cause the TV providers to move less aggressively into streaming content which is a disruption story.
Netflix’s decision to break streaming and DVD into separate brands services is a risky gambit due to the impacts I outline in my last post. The rational for separating the pricing makes plenty of sense as content licensing costs will grow as Netflix expands it’s content library and as Bill Gurley points out, licensing costs may be driven by the number of subscribers. But given that the disruptive technology story does not fit for streaming vs DVD, it’s hard to see how side-lining the DVD business is good move for Netflix’s customers or investors.
Ever since Netflix announced its new pricing structure, I thought the structure odd as there’s no discount for subscribing both to DVDs and streaming. The resulting subscriber loss erased 2.8 billion is market value between Wednesday’s market close and Friday’s close after new subscriber projections were announced. 25.6% of Netflix market cap evaporated over two days even as the S&P 500 rose 2.3%. Today, Netflix announced that it’s renaming and separating the DVD business. This move is a stunning gamble as it will likely lead to further subscriber erosion and is the underlying reason for the new pricing structure which is leaving money on the table in the short run.
The lack of bundle pricing is costing both revenue and profit today. A quick thought experiment illustrates how: First, figure out the highest price you would pay for Netflix’s unlimited streaming service. This what economists call your reservation price. It’s the price at which one penny more would cause you not to buy the service. Next, figure out the highest price you would pay for Netflix’s DVD service. Now assume that you already have Netflix’s streaming service. Either someone gave it you or you purchased it. Now what’s the highest price you would pay for DVD service? For most people, the reservation price for DVDs falls if they already have streaming. People already have so much demand for entertainment and once they have some options, the value they place on additional options falls.
This substitution effect between Netflix’s two services is easily addressed by bundle pricing. A discount for subscribing to both services would have meant less canceling subscribers as some customers would have found the bundle to provide enough value even though buying both separately did not. The customers who are closest to their reservation prices are likely to the DVD customers who turn over their DVD’s the least often and are thus the most profitable, as costs in the DVD business primarily driven by postage. Not offering bundle pricing loses customers and reduces both revenue and earnings though higher subscriber loss.
The Netflix brand has enjoyed a very strong reputation. The DVD business being renamed means that DVD business will lose the benefits of the brand. Building a new brand from scratch is a difficult, long-term, and often expensive endeavor. In the customer’s mind, offering both DVD and streaming both fit within the brand. Separating the brands, doesn’t provide any business value today.
An Impaired Product Experience
The integrated streaming and DVD website provided a great experience customers who subscribed to both services. Ratings a movie improved recommendations for both services. Searching for a movie shows both the streaming and DVD options. Separating the two services means users have to rate on two different sites to get recommendations and search for movies they want to see twice. This is a degraded user experience and one customers have been quick to point out the Netflix blog.
A Huge Gamble
Given all the immediate negatives for the business Netflix has today, the move to separate the streaming and DVD services places a huge bet on it’s streaming service. And while it’s clear the growth opportunities are in streaming business, what’s not clear is where to capitalize on the streaming opportunity required forgoing a substantial immediate value in the DVD business.
Most great companies can handle multiple related lines of business. Apple didn’t marginalize the Mac business to capitalize on the iOS opportunity. Other than the management challenges of managing related lines of business, there’s no external reason that Netflix could not have kept multiple integrated services under one brand. It will be quite interesting to see how this gamble turns out in the long run. I suspect the Netflix stock will be in for a rough ride in the short-term.
LinkedIn stock’s large price jump following it’s IPO last Thursday has caused many to comment that LinkedIn was “ripped off” by its underwriters by pricing the offering too low. LinkedIn used the standard offering process where the its underwriters, Morgan Stanley and Bank of America, offer the stock to their clients and set the offering price. Most of the commentary has missed the underlying economic fault in the traditional offering process which is having individuals price anything especially when they have competing incentives is asking for a poor outcome. The underwriters benefit by giving their best clients a big instant return on a hot IPO as they extract part of that gain through high fees on those clients. Pricing too high will lead to a failed offering which the underwriters will be blamed for.
There’s a better way as Google demonstrated in its 2004 IPO. A modified Dutch auction will set a market price for the IPO and maximize the value the company and the shareholders participating in the offering receive. For those not familiar with the procedure, a modified Dutch auction is pretty simple. Everyone interested in buying shares makes a bid saying they will buy X shares at a price of Y. The auction prices at highest price, called the clearing price, that will sell all the shares on offer with everyone biding the clearing price or higher getting their bid filled. The US Treasury uses the process to sell Treasury Bills and Bonds so it’s empirically proven in addition to being a theoretically sound.
To compare the two alternatives I normalized the offering prices of LinkedIn and Google to 100 and looked at the relative returns over the first 3 trading days. Google retuned 18% over its offering price versus 109% on the first day for LinkedIn. Much more money could have ended up in LinkedIn coffers if they had priced as close to the market price as Google did.
There are too few examples of high demand IPOs being Dutch auctioned to do a statistically significant comparison. However, the available evidence suggests it’s a much better pricing mechanism and economic theory is solidly behind that conclusion. I doubt Dutch auctions will become anymore common though. For non-hot IPOs, it’s not clear would be enough bidders. The underwriters will be opposed since it both marginalizes their role in the process and reduces the amount of value they can capture for their clients and ultimately themselves. Underwriters can not be eliminated from process since the securities laws require them. Corporate executives aren’t likely to clamor for it either. While a Dutch auction will benefit their shareholders, doing something unconventional always carries more risk of negative personal repercussions if it goes wrong. It’s lot easier to blame the underwriters for a unsuccessful IPO if their advice for traditional offering is followed rather than actively discarded.
Jack Welch had a famous rule that GE would only be in businesses where they can be number one or number two. In the industrial businesses GE was in, the scale that biggest players could achieve gave cost advantages that allowed for much higher profitability than the smaller players could achieve. Software is a world away from GE’s businesses but for platforms and operating systems, the number one or number two rule applies but for very different reasons. Operating systems with small share can not attract the third party developers needed to make the platform successful.
The personal computer era showed how minor platforms do not make it. Windows dominated the landscape for two decades with the Macintosh being the only viable competitor. Many others came and went like OS/2 and BeOS. NeXT which would later show itself to have the right attributes to be successful didn’t perform well in the market till Apple bought it and transformed it into OS X.
Android and iOS already have huge leads as application platforms. One of them could stumble but with the momentum both have, this looks unlikely. Even though the spoils are likely to be meager, there are several competitors for third place: Blackberry OS, WebOS (Palm/HP), and Windows Phone 7. Nokia had the good sense to take Symbian out of the race but made the peculiar choice of of Windows Phone 7. Both RIM (Blackberry) and Nokia would be be better off switching to Android as it would would put them on a platform where there’s already a wealth of third party apps. Neither RIM nor Nokia has ever proven themselves to be world class at mobile software. RIM’s entire success was based on email and they could salvage their market position by bringing their email system to Android.
These kind of strategic mistakes always puzzle me. I wonder if RIM, HP, Nokia and Microsoft have all convinced themselves they can become number two or if they believe there’s a profitable path as number 3 or 4.