• Alumni

Activity Feed

On November 20, 2015, AP commented on EFL: The psychology behind risk management :

Thank you for your post, Andrea. I think what EFL is doing is laudable both for creating a potentially profitable business model and for potentially increasing credit access for underserved populations. I also agree that data integrity is a key concern. This could manifest not only due to the data collectors, but also because borrowers may figure out how to “game” the system. The nice thing about traditional credit scores is that they tend to be based on more objective criteria such as default histories and income. While far from perfect, they are harder to fake. Psychometric analyses, on the other hand, are subjective by nature – they have optimal answers that potential borrowers can learn. As such, the value of any given set of questions will likely depreciate and the questions that replace them may not have the same predictive power. Thus, I worry that what starts out as a promising tool may devolve into a non-value-added questionnaire.

If it is somehow possible to ensure that the answers to psychometric questions can’t be gamed, I wonder if there would be any regulatory outcry. After all, if a person cannot game the system, that must mean the questions judge something intrinsic about the person. Insofar as these intrinsic characteristics correspond to easily identifiable features (i.e. race, religion, hometown, language, etc.), the tool could be accused of discrimination. Obviously, every type of credit score is discriminatory in some sense, but society is more likely to accept discrimination based on default history than more intrinsic characteristics.

Thank you for your post, Won’t Not. KnowRe certainly appears to be doing interesting work. However, I wonder how useful “big data” is for the elementary and middle school math problems KnowRe is addressing. If the goal is to identify areas for improvement, software can determine where a student should focus (i.e. exponents instead of division) within a handful of problems.

While there are obvious benefits to having a large library of problems, traditional textbook publishers already have extremely large libraries. I wonder how much new value KnowRe can create on a standalone basis – perhaps it’s a simple tool that can be stapled onto a publisher’s existing virtual offerings.

I wonder how extensible KnowRe’s approach is. Most immediately, as you suggest, it would be interesting to see whether the product can deal with higher-level math. Mistakes in these classes are often tougher to decompose into piece-parts like exponents or division – more often, they relate to a misunderstanding of how the piece-parts interact. It would require a complicated algorithm to identify patterns of mistakes in this context – perhaps the “big data” approach you outline would have more value here.

Further out, I wonder if KnowRe can do what it does outside of math. Math is the most analytical discipline and it is easier to work on specific skills there than in other disciplines. However, a tool like this might also have value in strengthening, for example, specific grammar skills. The further it is extended beyond analytical subjects, however, the more likely it is to devolve into a commoditized flashcard solution that simply helps with memorization.

Thank you for your post, Kathryn. Tesco’s ownership of dunnhumby raises some interesting corporate strategy questions. If dunnhumby is truly responsible for Tesco’s meteoric rise to the top of UK retail, then why not use it as an internal tool rather than as a company serving competitors? Although dunnhumby’s value is undoubtedly enhanced by access to competitor data, the net value captured by Tesco is arguably diminished by offering insights from this data to others.

Granted, there are ways to commercialize an internal tool without directly helping competitors. For example, insofar as Tesco is uninterested in competing in the US, then it makes sense to offer the tool to Kroger’s. Similarly, if the company has no plans to integrate upstream (i.e. through private label groceries), then perhaps it makes sense to offer data to farms, manufacturers, and distributors. Indeed, this could even be synergistic for by improving operational integration with the upstream partners. However, sharing its “proprietary analytic approaches” and “wealth of data” with 40 retailers and 50 manufacturers may be ill-advised. At best, it removes an advantage Tesco would have if it one day chooses to expand its geographic or product scope. At worst, it erodes Tesco’s advantage and emboldens competitors in its existing business.

Viewed from this perspective, perhaps Tesco’s strategic decisions with dunnhumby are not so ill-advised. After all, the company’s total enterprise value is ~$37 billion. Sacrificing the potential to realize 1.3 billion additional pounds to protect a $37 billion franchise makes a lot of sense.

On November 1, 2015, AP commented on A/Z Testing: Crowdsourcing the ad creation process :

Thank you for your post Asaf. I agree that crowdsourcing advertising research offers interesting opportunities when designing very targeted ad campaigns on the Internet. However, companies can simplify and deepen the effort for broad-based campaigns – they can crowdsource the ad creation process altogether.

Doritos was a pioneer in this effort. In fall 2006, the company launched the “Crash the Super Bowl” contest, whereby it solicited full-length television ads on its website and asked visitors to vote on the best ones. It committed to airing the ad with the highest vote-total during the Super Bowl. The result was an absolute coup for the company – Doritos became the first company to air a consumer-generated Super Bowl ad (it actually aired two of them during the 2007 Super Bowl – http://www.prnewswire.com/news-releases/doritos-surprises-super-bowl-television-audience-airing-two-consumer-created-commercials-live-the-flavor-and-check-out-girl-54272512.html) and received very positive feedback. In fact, one of the two commercials it aired was rated among the top 5 Super Bowl commercials that year. This was all achieved at minimal cost – Doritos simply offered the top 5 entries a cash prize of $10,000 and a trip for 2 to the Super Bowl. While the size of the grand prize has dramatically increased since 2006 (it’s now $1 million – http://ktla.com/2015/02/01/usc-graduate-wins-doritos-super-bowl-commercial-contest-1-m/), the company has been able to save millions of dollars in ad agency expenses by using crowdsourcing.

The extensibility of Doritos’s strategy to other advertising venues is, of course, questionable. The Super Bowl occupies a unique place in American pop culture and the prestige of potentially having your ad aired during the game is a reward in itself. It would be far tougher to generate as high-quality productions from the crowd for ads that will be aired during a typical weekday television program. However, the viability of a crowdsourced advertisement should establish a price ceiling that companies are willing to pay ad agencies. After all, in a worst-case scenario, a company of sufficient scale could run a similar contest just among its employees and likely generate content that is nearly as creative as that of an ad agency. In other words, as companies tap “the crowd” to generate creative advertising content, the “creativity premium” should decrease.

On November 1, 2015, AP commented on 50% + 1 Rule :

Thanks for the post, AlonKremer. Your idea reminds me of a case study on Loyal3 that we analyzed in Investment Management. Loyal3 is a financial technology company whose aim is to promote stock ownership among companies’ consumers. The basic hypothesis, as explained by the company’s Chief Creative Officer, is “If you like the company, you’re loyal to the company and you use its products. You become even more fanatical once you become an owner. That locks you in.” He points to a study that found that consumers who also owned a company’s stock “spent an average of 54% more per year at that company, visited its stores and Web sites 68% more often, and referred twice the number of new customers than consumers who did not own stock in the company.” If ordinary consumer/retail companies achieved such stunning results among their consumer-shareholders, I imagine the impact on engagement would be even greater for an explicitly fan-based company such as a sports team.

Loyal3 has created various digital offerings to fulfill its aim. The most differentiated is a Social IPO platform, which partners with companies going public to allocate a portion of shares to the companies’ consumers. Sports teams could actually use this very platform to achieve the goal you’ve outlined. The next time a team’s owners sell a stake via an IPO (as Manchester United did in 2012), they can leverage Loyal3 to allocate a portion of the offering to fans. Another one of Loyal3’s offerings that could be used to implement your idea is “Stock Rewards,” which offer small chunks of equity as rewards for consumer behavior. Perhaps sports team could offer micro-portions of equity to fans who attend a certain number of games, who buy a certain amount of merchandise, or meet some other consumer spending threshold.

One criticism of Loyal3 is that it enables and persuades individual investors to allocate their savings to single stocks without conducting due diligence. Individuals would generally achieve superior risk-adjusted returns via a mutual fund or index fund where they cede individual stock-picking responsibility. This concern is relevant to the sports team crowdfunding idea as well. For instance, a particularly passionate fan may irresponsibly invest his/her entire life’s savings into a team that turns out to be a bad investment. A platform that encouraged and facilitated this decision would be partly culpable in the fan’s financial loss and would have perpetrated a social wrong.

On November 1, 2015, AP commented on Crowdfunding in Agriculture: Feeding the World :

Thanks Charles, very interesting post. I like your idea of an operator using lease contracts to piece together larger plots of land and to improve yields through investment. However, I wonder whether crowdfunding is the best source of capital. The types of investments you’re speaking of require very patient investors who are willing to stomach illiquidity. Moreover, they would benefit from investors providing technical know-how (about agriculture, local real estate regulations, etc.) in addition to capital. As such, why wouldn’t the appropriate organizational structure for the operator look more like an agriculture-based private equity firm? For instance, Black River Asset Management, the private equity affiliate of agriculture giant Cargill, could be the right investor given its ultra-long-term capital and significant agriculture domain knowledge (i.e. they already know what investments should be made).

Alternatively, a vendor of agriculture seeds/equipment/services could operate a leasing program of the type you’re proposing with meaningful synergies versus an independent crowdfunded operator. For example, Monsanto, a publicly traded seeds/agrochemical company, would not only capture the ~30% ROI you suggest any operator could, but would also benefit from an expanded addressable market for its core products.

Even if we suppose that private equity firms and agriculture vendors can’t or won’t pursue your idea for whatever reason, why is crowdfunding superior to standard partnerships? If, as you suggest, the operator only needs 20-30 investors, he/she wouldn’t need a digital platform to find investors and it wouldn’t be worth the investment to create one. Alternatively, you could argue that someone could set up a digital platform to pair potential operators with investors. However, this presumes there’s a critical mass of operators interested in this business model but who can’t access funds in more traditional ways – it’s not clear that there’s a large enough network to merit an independent digital platform.

Thank you for your post, Jason. While I agree Apple and Android-based devices beat BlackBerry on the basis of their indirect network effects, I disagree that BlackBerry had strong network effects to begin with. On the direct side, although BlackBerry Messenger (BBM) only worked on BlackBerry devices and thus created a closed “network,” it is not clear that the network “effect” was meaningful. That is, people did not buy BlackBerry phones so that they could use BBM. On the indirect side, BlackBerry made minimal efforts in building a 3rd party developer ecosystem and people certainly didn’t buy BlackBerries for the small handful of 3rd party apps that existed. Instead, they bought BlackBerries because they were the most technically superior smartphones (until the iPhone) that were prized by corporations and consumers alike. The technical superiority was established with the best batteries (that lasted almost a week – amazing by today’s standards!), the best keyboard (at least until touchscreen devices were improved), the best email application, the best Internet interface, and the most secure communications architecture (via the Network Operations Center, which connected to mobile carrier and corporate networks). In short, BlackBerry sold a premium hardware product as opposed to a network or a platform.

The genius of Apple and Android, of course, was in creating a platform around what was previously just a hardware purchase. BlackBerry failed to do so. As the platform became an increasingly important purchase criterion for both individual and corporate smartphone buyers, BlackBerry’s competitive advantage eroded. Eventually BlackBerry lost even its technical superiority – for instance, the same Network Operations Center that was considered the most secure communications architecture in the mid-2000s is now considered an insecure single point of failure outside the control of its users (http://blogs.reuters.com/great-debate/2013/11/12/what-we-learned-from-the-blackberry-era/).

It is interesting to consider scenarios where things would have ended differently for BlackBerry. The first and most obvious scenario is if BlackBerry had encouraged 3rd party app development earlier on. If it had done so in the mid-2000s, perhaps it could have actually established indirect network effects and blocked out Apple from successfully entering the market.

The second scenario is if BlackBerry had produced phones with the Android OS earlier on. BlackBerry has clung to only producing phones with its proprietary OS. Now that it has already lost the battle, it is finally considering producing Android-based phones en masse (http://www.cnbc.com/2015/08/20/why-blackberry-could-ditch-its-own-os-for-android.html). Perhaps if BlackBerry had adopted the Android OS shortly after it had begun exhibiting network effects, BlackBerry could have evolved into a key hardware partner for Google and the key competitor to Apple, much as Samsung has become.

The third scenario is if the mobile internet had developed differently. At present, the supermajority of mobile activity is on mobile apps – rather than visiting Facebook.com on their mobile web browser, users just log in through the Facebook app. This is why app ecosystems are so critical to the value of smartphones. However, in an alternate universe, consumers could have done everything they do on smartphones via mobile browsers using HTML5 technology. At various points over the last several years, many smart technologists have argued that HTML5 websites should displace apps as the primary medium of mobile activity. Indeed, the debate still rages: http://www.itworld.com/article/2912880/mobile/has-the-native-vs-html5-mobile-debate-changed.html, https://www.mobilesmith.com/html5-vs-native-debate-is-over/. If HTML5 had developed more rapidly and consumers could indeed access games, entertainment, productivity applications, etc. through the mobile web as seamlessly as they can through mobile apps, perhaps Apple’s and Android’s app ecosystem wouldn’t be as important. In such a world, consumers would decide what phone to buy based purely on technical superiority, just as they did in the mid-2000s. Perhaps BlackBerry could have competed more effectively in this alternate universe.

On October 3, 2015, AP commented on Tinder: The dark side of network effects :

Thank you for your post, Jennifer. It is certainly interesting to consider that network effects can turn negative beyond a certain point. However, it is worth noting that this phenomenon is unique to platforms where (1) the user relies on the platform to make recommendations, and (2) the platform’s technology struggles to do so in a satisfactory way. For instance, there is no congestion-based downside for WhatsApp because users decide who they want to message with on their own. Moreover, there is no negative network effect for Google from the relentless expansion of the Internet because its algorithm is sufficiently sophisticated to produce the most relevant results for users’ searches. Perhaps, therefore, Tinder’s problem is less structural and more that it purports to offer relevant results without having the necessary data or technical capacity to do so.

I was also intrigued by your analysis of the nature of network effects for dating apps vs. other platforms. You are correct to point out that because dating is primarily a local activity, dating networks are built locally. This is a limitation on the scalability of a dating network. Another limitation is that there is much greater churn in the addressable user base vs. other platforms. A typical person might register for a dating app in college and the average age for marriage is in the late 20s (27 for American women and 29 for American men – http://www.huffingtonpost.com/2013/11/14/married-young_n_4227924.html). As a result, the time period over which a dating app is relevant is quite limited for the majority of users (though, of course, there will be a long tail of people who take longer to find their significant others, who never do, or who get divorced and re-enter the pool). Indeed, it is ironic that the more value a dating app creates for its users, the more churn it will have – that is, a dating app that successfully matches two compatible people will take them off the market. This is in stark contrast to a platform like Uber, which can be used for a lifetime and for which a positive user experience is likely to reduce churn. The practical effect of high user churn for dating apps is that it significantly weakens the network effect. Because dating apps are constantly losing large portions of their user base, they must constantly attract new users – they have to run just to stay in place. If the next generation of users perceives them to provide less value, network effects will not be enough to save them.

Karthik, thank you for your thoughtful post about Ola Cabs. You pointed out that Uber entered the market by offering high-end cars for premium prices while Ola entered with lower-cost hatchbacks. However, the difference in how the two companies approached the Indian market extends further than cost. Whereas Uber focused on attracting independent drivers to its platform, Ola Cabs began by simply allowing consumers to book ordinary taxis. In a sense, Ola Cabs’ offering was “backwards compatible” with the predominant existing form of hired transport. This is analogous to the approach of new PC operating systems. Retaining backwards compatibility with apps developed for the prior generation encourages consumer adoption of the new OS because it comes with an existing ecosystem (albeit one that is not as “up-to-date” as the OS itself). Similarly, working with existing taxis and taxi companies allowed Ola to rapidly scale the supply side of its platform. This, in turn, has allowed it to more rapidly add users and take a lead in Indian tech-enabled transportation. The downside of relying on taxis is that it’s even easier for taxi companies to multi-home than independent drivers – insofar as Ola’s initial network is built on taxis, it offers a relatively weak network effect. Indeed, Uber has started to also attract taxi supply through its “UberTaxi” offering.

More generally, it is not surprising that Ola has been able to successfully compete with Uber in light of how transportation network effects work. The vast majority of a typical consumer’s transport needs are within the city he/she lives in. As a result, the typical consumer doesn’t care much about the supply density of drivers 50 miles away, let alone the density of drivers in other states or countries. Similarly, the typical driver doesn’t care about user density anywhere other than the confines of the city in which he/she drives. Therefore, the cross-side network effects for an app like Uber are uber-local. In fact, they are among the most local of any tech-based business. Online commerce, for example, has at least national network effects because merchandise can be shipped between merchants and users anywhere in a country with ease. Social networking arguably has international network effects because many users have connections to multiple parts of the world. However, the fact that Uber is the top player in the US is nearly meaningless when it comes to competing in India, at least from a network effect perspective.

Of course, Uber’s global scale may offer it traditional scale economies on the cost/capital-side. For instance, Uber can amortize its tech development costs over a broader revenue pool. However, returns to scale from tech development quickly peter out because tech development costs are minimal compared to the variable costs of operating networks / paying drivers. Uber can also subsidize the development of new networks (like the one in India) with cash from already-profitable locations or externally raised capital. However, a “cost-of-capital”-based advantage is among the least valuable – Uber is no more differentiated on this basis than a Google or even a Ford would be in entering tech-enabled transportation in India.

Given the lack of a transferable network effect and the lack of a clear cost advantage, there isn’t any reason why we would expect Uber to have any advantage in India vs. Ola. If anything, given Ola’s market-specific features, we should expect Ola to do well. However, as you rightly point out, it is unclear whether the network effect will ever truly be “monetizable” in light of the ease of multi-homing for both drivers and consumers…

On September 12, 2015, AP commented on OnePlus One, Redefining the Math of Smartphones :

Thank you for the post, Erika. It’s always fascinating to hear about a new competitor to the behemoths that dominate mobile hardware. However, the real value in iPhone and Android devices come not from the hardware (impressive though they are), but from the accompanying app ecosystems. While the iPhone is certainly valuable in its own right, it would be far less coveted without Uber, Venmo, Facebook, and countless other apps that transform it from a communication device into a productivity/entertainment/financial platform. If OnePlus intends to rely on its internally developed OxygenOS, it will struggle to build the ecosystem that Apple and Google have developed. After all, the world of smartphone operating systems is a classic two-sided network – a large and growing user base attract developers and a dedicated developer base producing must-have apps attracts more users. The market has already “tipped” toward the iPhone and Android ecosystems – they each have hundreds of millions of users and mobile developers have no reason to focus on any other platforms. The only way to disrupt such a strong network effect is to develop an entirely different type of hardware with a different ecosystem. After all, the iPhone and Android ecosystems were only able to develop because they focused on the mobile devices that were disrupting PCs (where Microsoft benefitted from an incredibly strong network effect with developers). However, while OnePlus’s phones are interesting, they are nowhere near as revolutionary as the original smartphones and tablets were.

Indeed OnePlus’s strategy reminds me of Blackberry’s strategy circa 2010. At that point, many tech critics agreed that Blackberry produced superior communications devices vs. the iPhone or Android phones. Moreover, the OS was more stable, messages through Blackberry’s Network Operations Center (NOC) were more secure, and the keyboard was easier to type on than touch screens. While Blackberry produced excellent hardware and had a technically superior OS, it chose not to focus on attracting 3rd party developers. Thus, while the functionality of iPhones and Androids continuously improved and gained in popularity, Blackberry remained static – the rest is history.

Beyond the problem with OnePlus’s proprietary software approach, it seems highly unlikely that they can cost-effectively develop superior hardware than Apple and Samsung. As you point out, OnePlus shipments are miniscule compared to the behemoths. Given economies of scale and technical learning that comes from producing so many phones (even if much of the actual production is outsourced to ODMs), Apple and Samsung likely have such a cost advantage that OnePlus couldn’t compete even if it was willing to accept lower profit margins. You might argue that economies of scale diminish beyond a certain point and that OnePlus can achieve a similar cost structure even at much smaller scale. I don’t know enough about the smartphone supply chain and manufacturing process to know whether that’s the case. However, if it is, it’s tough to see how OnePlus will be building a good business. Without the brand premium commanded by Apple and Samsung (or the network effect they get from their app ecosystems), then economies of scale are the only barrier to entry. If they are tapped out at a relatively small scale, OnePlus will be vulnerable to any other upstart hardware manufacturer.

On September 12, 2015, AP commented on Zillow: Bringing Data and Transparency to the Neighborhood :

Thank you for the post, Noah. I agree that Zillow has done an incredible job attracting 77 million users in such a short time period (it was at just 17 million users in early 2011). I also agree that it is disrupting a huge market – there are ~$1 trillion of home sales each year, real estate broker/agent commissions are ~$60 billion per year, and real estate brokers spend ~$10 billion on advertising per year. Zillow’s revenues for 2015 are expected to be ~$665 million, suggesting that it has meaningful growth runway in penetrating the $10 billion advertising opportunity. The opportunity is even greater if Zillow can eventually circumvent real estate brokers altogether and target the $60 billion of commissions. However, the company faces several challenges to realizing its promise.

First, it is unclear what share of Zillow’s users and user growth is comprised of consumers actually looking to buy or sell real estate. Its pitch to brokers is that its website’s visitors are “transaction ready.” If the supermajority of Zillow’s users just have fun with the site’s slick valuation tool without any intent to transact, they users are much less valuable to brokers. If, in fact, its users are “transaction ready,” Zillow must find more effective ways to prove it than they have to date.

Perhaps even more concerning is that brokers who are Zillow’s primary customers are equipping their own websites with more relevant data and easy-to-use interfaces (http://www.barrons.com/articles/SB50001424053111904329504580071503076700616). For example, Realogy (which owns real estate brands such as Century 21, Coldwell Banker, and Corcoran Group) is growing leads from its own sites at almost twice the rate of third-party sites (46% vs. 27%) and converting them into purchases at almost triple the rate (5.6% vs. 2.0%) [source: Realogy investor presentation: http://ir.realogy.com/phoenix.zhtml?c=198414&p=irol-presentations%5D. Granted, this is only possible for the very largest real estate brokers who have the scale to add sophisticated features and drive traffic to their websites. However, the largest brokers such as Realogy and ReMax have been taking share from smaller ones – if they continue to do so and/or if there is increased M&A activity in the space, their proprietary platforms could pose a challenge to Zillow. Indeed, this dynamic played out in the airline reservations space. While online travel agencies (OTAs) like Expedia, Travelocity, and Priceline were initially consumers’ websites of choice for online air bookings, the airline industry consolidated and airline websites improved to the point where it became more convenient for consumers to book direct. As such, although OTAs are still very relevant in the hotel space, online air bookings have been shifting away from OTAs to airline websites.

Furthermore, while Zillow’s business model may lend itself to network effects, it is unclear whether the market has “tipped.” That is, the market may be nascent enough that it has not yet entered the virtuous cycle whereby the existing leader becomes the sole winner. Leads from the internet still produce less than 10% of property buyers and 5% of property sellers, including those who find real estate agents online (http://www.barrons.com/articles/SB50001424053111904329504580071503076700616). Given the low existing penetration, there is plenty of room for a competitor with meaningful housing data to develop an attractive alternative. Indeed, Realogy is working on just such a tool replete with listings from other brokers as well as a pricing tool comparable to Zillow’s. Realtor.com and others have also introduced similar offerings. If one can emerge as superior to Zillow in the eyes of users for whatever reason, Zillow may lose its early lead much as Myspace ceded its commanding perch atop the social networking world to Facebook.

On September 12, 2015, AP commented on Netflix in the ever changing home entertainment space :

Thank you for the post, Pipedreamer. You articulated the drivers of Netflix’s success to date very well. While I am a fan of Netflix as a consumer, I worry about its future in a world where over-the-top content is becoming increasingly prevalent. Content owners are trying to dis-intermediate distributors like Netflix by offering over-the-top options themselves. For example, all four major broadcast networks now offer live streams and on-demand content online for consumers who have pay-TV subscriptions (http://www.wsj.com/articles/nbc-to-live-stream-network-shows-1418706061). Moreover, an increasing number of content owners, including CBS (CBS All Access: http://www.cbs.com/all-access/), HBO (HBO Now: https://www.hbonow.com/), and Showtime (http://www.showtime.com/?i_cid=int-default-1010), are offering live streams and on-demand content a la carte. In addition to the content owners’ direct efforts, many other services, such as Amazon Prime and Hulu Plus, are trying to replicate Netflix’s online distribution model. Yet others, including cable companies themselves, Sony, and potential entrants like Apple (http://www.wsj.com/articles/comcast-launches-streaming-service-for-cord-cutting-customers-1436788422) are offering live TV streaming bundles.

Given all these recent and forthcoming developments, the supply-demand dynamic doesn’t look great for Netflix. On the supply-side, Netflix’s content costs will increase at ever-faster rates as content owners have more online distribution options and begin to favor their own offerings. On the demand-side, Netflix’s pricing power with consumers will decrease as they have exponentially more ways to consume content online. And these cost and pricing pressures start from an undesirable status quo – Netflix doesn’t generate any meaningful profits today. You might argue that Netflix benefits from network effects that could mitigate or invert this supply-demand dynamic. In theory, because of its first-mover advantage, Netflix could attract so many users that content owners need to distribute through it, and it could attract so much content that users need a Netflix account. However, competitors like the cable companies, Amazon, and Apple have sizeable user bases of their own (albeit for different purposes in the case of Amazon and Apple). The cable companies also have existing relationships/contracts with content owners (or, in some cases, as with Comcast’s NBC-Universal, produce content of their own) that should ease their negotiating process. Moreover, consumers are not locked into Netflix – because switching costs are de minimus, content owners will attempt to drive traffic through their own online channels. In an unbundled world with low-cost a la carte options, it will be quite easy for consumers to “multi-home” and create their own content packages rather than being forced to rely on Netflix. In short, Netflix’s theoretical network effects may very well fail to generate the profits it expects.

If Netflix’s content costs rise inexorably, the only way it can make money is to distribute content that it produces itself. Indeed, Netflix has had success in “original programming” with shows like House of Cards and Orange is the New Black. An optimist might argue this success is replicable going forward. After all, given its large existing user base, Netflix could leverage the insights it develops from analyzing content consumption patterns to create shows that people will predictably like. I’m skeptical – it is extremely difficult to translate data analytics into an inherently creative process with predictable results. Moreover, existing content owners with decades of production experience are not without data of their own. I believe content production will remain a hit-driven business and Netflix is no more likely to produce hits than premium content providers (e.g. HBO, Showtime), the broadcast networks (ABC, NBC, Fox, CBS), and the major studios (e.g. Disney, Warner Brothers, NBC Universal).

Perhaps it’s not such a bad outcome if Netflix becomes another content producer (albeit one that happens to have a lot of direct subscribers). However, Netflix trades at 80x 2015E EBITDA, a far cry from the 10x EBITDA multiples more common among content owners. While Netflix has been a darling of the digital revolution to date, its shareholders might not be so quick to label it a winner when all is said and done.