Category: Geen categorie

When do freemium apps boost sales of their paid counterparts?

Platform Papers is a monthly blog about platform competition and Big Tech. Blogposts are written by prominent scholars based on their research. The blog is linked to platformpapers.com, an online repository that collects and organizes academic research on platform competition.

This blog is written by Yiting Deng, Anja Lambrecht and Yongdong Liu.

Digital platforms have enabled individuals and businesses to access audience and markets in unprecedented ways. Central to the success in this digital realm is the ability to monetize digital offerings. There are various methods for monetizing digital content, including freemium models, subscription models, pay-per-download or pay-per-view, and advertising.

The freemium business model, a blend of offering a “free” and a “premium” version of a product, has gained considerable traction in the digital space. Well known examples include Dropbox, LinkedIn, Spotify, etc. For mobile apps, freemium apps are those that offer a complimentary version with basic features while providing opportunities to unlock advanced or premium features for a fee. This can be accomplished by providing a separate paid premium version of the app, or more recently, through in-app purchases or subscription plans. Essentially, the free version provides consumers an opportunity to explore the app without incurring any immediate financial cost. The notion of sampling, a fundamental aspect of freemium models, is not exclusive to the digital realm. Well before digitization, merchants would distribute free samples to consumers to allow them to try the product. Similarly, in the software industry it has long been common to offer free trials. These trials typically came in two forms: time-limited trials (providing a free, fully functional version with a limited trial period) or feature-limited trials (offering a free, perpetual “light” version with restricted functionality).

In the context of freemium apps, developers typically opt for feature-limited trials. The free version grants users free access to the app’s core functionalities. For example, a freemium game app might allow users to play a few initial levels for free, while reserving advanced levels for the premium version. While sometimes users can access the premium version as a separate downloadable app, in other instances, these advanced levels can be unlocked through in-app purchases.

Despite the popularity of the freemium business model, a crucial question remains: Does the existence of a free version boost or undermine the sales of its existing paid counterpart? On one hand, the free version provides consumers with the opportunity to try out the product before committing to a purchase, potentially driving up demand for the premium version. However, on the flip side, the free version might cannibalize sales of the paid version. So, does the availability of a free version enhance or undermine the sales of its premium counterpart? How should developers design freemium apps to increase conversion rates?

In a recent paper published in Management Science, we set to answer these questions. We compile a comprehensive and granular data set on mobile game apps from Apple’s App Store and identify apps that offered both a free version and a paid version. In our sample, the majority of freemium apps initially offered a paid version and later introduced a free version. We focus on these apps to analyze how the introduction of the free version affected the paid version’s demand. Interestingly, we find that the introduction of a free version boosts the demand for its paid counterpart, suggesting that such positive effects outweigh any possible cannibalization. It is, however, not immediately clear why the free version of the app increases the sales of the paid version, instead of cannibalizing it. To better understand why this is the case, we explore two possible mechanisms: sampling and app discovery.

Sampling

We first explore whether the free version may benefit the paid version because it allows consumers to sample the paid version before making a purchase. After exploring the free version, if the consumer enjoys the app but wants to gain access to more functions, they would then upgrade to the paid version. In this way, the free version can increase the paid version’s demand through sampling.  

When would the benefits from sampling be most prominent? If, for example, information such as previous ratings indicate that the paid version is of low quality, consumers have such low expectations of product quality that it is not worthwhile incurring the hassle of sampling, and as a consequence, offering a free sample will have little effect on purchases of the paid version. Conversely, if there is information suggesting that the paid version is of very high quality, consumers may prefer to purchase the paid version directly – if they expect that the full and better version justifies its cost, they may consider it not worthwhile to go through the effort of downloading and using an inferior version. The sweet spot therefore lies in the middle: it is only for products where publicly available information, such as ratings, suggests a medium quality level that sampling matters. Indeed, we identify this inverted U-shaped relationship between an app’s average star ratings and whether the free version indeed increases demand for the paid version.

App discovery

With the very large number of available apps, one challenge for apps may be to be noticed by consumers. Statista reports that Google Play had 3.55 million apps and Apple App Store had 1.6 million apps. As a result, it may be difficult for any individual app to be visible to consumers, put differently, it is harder for a consumer to find the “needle in the haystack”. Could the availability of a free counterpart make it easier to find a paid app? We explore our data and compare categories with a smaller or larger number of available apps. Our results show that for a category with more apps, the extra visibility from a second version matters less. This makes sense: if the “haystack” is large, the relative benefit of a second app in helping discoverability is less than if the “haystack” is small. This pattern suggests that indeed the availability of two versions makes it easier to discover the app.

Balancing act: What should be free?                              

The success of a freemium app hinges on striking a balance between free and premium features. App developers must ensure that the free version remains useful and engaging while offering a compelling incentive for users to upgrade. How can developers strike such a balance?

In our data, we observe differences between the paid version and the free version, and analyze which differences are most effective in driving demand of the paid version. There are mainly six dimensions along which the two versions may differ:

• First, the paid version allows users to progress to more game levels than the free version, such as a more complete, advanced, or challenging game experience.

• Second, the paid version offers more modes or themes than the free versions. Although the ability to progress or the difficulty of the game does not change, the user experience can be customized.

• Third, the paid version offers more functions or features (e.g., more powerful weapons, record keeping) than the free version.

• Fourth, the paid version allows for social interactions, whereas the free version does not, such as integration with Game Center, Apple’s social gaming network, or linking to Facebook to post scores and share progress with friends.

• Fifth, the paid version is ad-free, whereas the free version is ad-supported.

• Finally, the paid version might provide better user support, for example, an email contact to address user questions.

We find that the positive impact of introducing a free version is particularly pronounced for apps where the paid version offers distinct additional benefits, such as extra game levels, enhanced functionalities, or a broader scope of social interactions. With such “vertical differentiation”, the free version allows consumers to experience the initial game levels with basic functionality, while the additional levels and/or advanced features provide enough value for consumers to upgrade. Likewise, the opportunity to compete, compare scores and communicate with others provides added value for those who enjoyed the initial version, making a purchase of the full version appealing. When the paid version offers more modes or themes, the free version does not affect the paid version’s demand. Such “horizontal differentiation” only changes the appearance but does not improve the gaming experience to a meaningful degree. Interestingly, if the paid version simply removes ads, an ad-supported free version has a smaller benefit, consistent with the notion that many consumers prefer ads rather than paying for content. Finally, if the paid version only provides more support, then the free version would cannibalize the paid version.

Broader implications

When we initiated the project, most freemium apps existed as two separate versions. However, in today’s landscape, it has become commonplace to have a single app version with integrated in-app purchase functions (an earlier blog covers this model). According to Statista, as of July 2023, nearly 97% apps on Google Play app store and nearly 95% apps on Apple App Store are free to download. This trend is observable in the journeys of many individual apps, such as Fruit Ninja, a popular game released by Halfbrick in 2010. Initially, Fruit Ninja Classic (known as “Fruit Ninja” before 2017) was introduced in April 2010 with a price tag of $1.99. Over time, it transitioned to being free to download with in-app purchases. At the same time, Fruit Ninja (formerly known as “Fruit Ninja Lite” and “Fruit Ninja Free” before 2017) was introduced later in October 2010 and has always been available as a free app.

While our study focused predominantly on apps with separate free and paid versions, its implications are relevant for apps offering in-app purchases and have broader relevance beyond the mobile app market. For digital firms and app developers, the study offers three salient takeaways:

First, it substantiates the effectiveness of a freemium strategy in increasing demand for the paid version of a product. Second, the results indicate that a freemium strategy is most effective for products that prior users evaluated as moderately good. Third, the findings demonstrate that to truly benefit from a freemium strategy, firms need to ensure a sufficiently large difference between the value consumers receive from the free and the paid versions to induce upgrades. In pursuit of this goal, developers should carefully consider which features should be made available for free.

This blog is based on Yiting, Anja and Yongdong’s research, which is published in Management Science and is included in the Platform Papers references dashboard:

Deng, Y., Lambrecht, A., & Liu, Y. (2023). Spillover effects and freemium strategy in the mobile app market. Management Science, 69(9), 5018-5041.

Platform-Paper updates

I am excited about the Platform Leaders Future of Digital Platforms conference later this week at London’s Science Museum. Benoit and Laure from Launchworks asked me to put together a panel on the Future of Platform Research.

In my opening remarks I will document the evolution of platform competition research by pulling some descriptive stats from the platformpapers reference dashboard (for example, mobile app stores are now the most popular empirical context for platform research, replacing video game consoles). It’s an excellent opportunity to start preparations for 2023’s Year In Review post.

My colleage JP Vergne (whose work is included in the references dashboard and who’s written an excellent blog on decentralized platforms and Web 3) and Xu Zhang at the London Business School (whose work will soon be included in the reference dashboard, surely) have kindly agreed to showcase their work at the panel.

I have known Benoit and Laure for some time now and a few months ago I sat down with Benoit for a conversation about network effects. We discussed what network effects are and what firms should take into account when designing their products for network effects. The conversation has been cogently summarized in two separate articles: What are network effects? and The power of network effects. Enjoy!

On top of all of this, several interesting papers have been added to the Platform Papers references dashboard this month. Here are some highlights:

• A paper by Takanori Adachi and colleagues (published in The Journal of Industrial Economics) studies the implications of consumer multihoming in two-sided platforms. Their economic model shows that “the required merger-specific cost reduction is larger if consumers benefit more from multi-homing and that the equilibrium level of platform entry can be insufficient in the presence of consumer multi-homing.” In other words, contrary to popular belief, multihoming does not necessarily alleviate the need for stricter (merger) policy.

• A study by Suzana Varga and colleagues (published in the Journal of Business Venturing) tracks the growth trajectory of Takeaway.com, a successful food-delivery platform, to identify the drivers of ‘platform scaling’—the ability to accommodate growth and add revenues without a commensurate increase in costs. The authors conclude that managers at Takeaway.com “purposefully and repeatedly use and revise a portfolio of decision rules to cultivate indirect and data network effects, which allows them to initially facilitate the growth of their platform and over time support the transition to scaling the platform.”

• A study by Arslan Aziz and Rahul Telang (published in Information Systems Research) looks at the consequences of ratings inflation on platforms. While ratings can reduce search costs, ratings inflation, where the overall average rating increases and the variance in ratings decreases, can dampen the informative value ratings provide. Analyzing data from a quasi-experiment conducted on another food-delivery platform, the authors find that while sales surge following an inflation in restaurant ratings, such inflation decreases user trial and leads to an overall increase in sales concentration among the most popular restaurants.

That’s all for now. See you next month!

Platform Papers is curated and maintained by Joost Rietveld.

When apps have network effects, platforms orchestrate their markets differently

Platform Papers is a monthly blog about platform competition and Big Tech. Blogposts are written by prominent scholars based on their research. The blog is linked to platformpapers.com, an online repository that collects and organizes academic research on platform competition.

This blog is written by Shiva Agarwal, Cameron Miller and Martin Ganco.

Platform owners need to manage third-party complementors to create value for customers and capture value for themselves. However, managing product categories on digital platforms can come with unique challenges not typically faced in traditional retail settings. One such challenge that has emerged on digital platforms over the past decade is the presence of products that have direct network effects. When third-party complements like social networking apps or massively multiplayer online games have direct network effects, the value they create for a user depends on the number of other users adopting and their level of engagement with the product. This creates a challenge for the platform owner because, in the presence of direct network effects, the value the products create for users and the performance the category creates for the platform owner can be very sensitive to the market structure; thus, the platform owner will want to manage such complementary products carefully.

To illustrate this challenge, consider a product category on a digital platform with products that do not have meaningful network effects. For instance, these could be single-player games that could be played also offline. In such a category, the platform owner will be agnostic as to the category’s market structure if users find the products that they want, the costs of offering those products through the platform are negligible, and no complementor has excessive bargaining power. Under such conditions, a category with 100 users each adopting a separate product could produce as much value for users and for the platform as a category with 100 users all adopting the same product.

Now consider a category on a digital platform for which products have meaningful direct network effects. For instance, this could be a category of massively multiplayer games. When the category has a low concentration of adopters, each product may fail to build enough installed base to create value for its users. You do not gain much value from a multiplayer game that lacks other players. Therefore, in the presence of product-level direct network effects, the platform owner does not want the share of users to be too diffuse across products because no product will have enough installed base to create sufficient value for users.

High concentration can also create problems for the platform owner because the market could tip towards a dominant complementor who may stymie innovation in the category or challenge the platform owner’s rules. Unlike the product without direct network effects, a large installed base in the presence of direct network effects becomes a strong competitive advantage that can deter competition and enhance the bargaining power of the complementor. Complementors may even begin to challenge the platform owner’s rules, as in the case of Epic’s lawsuit against Apple.

In a recent paper published in the Strategic Management Journal, we examine how Apple manages game developers with direct network effects in the App Store. We find that in the presence of direct network effects, value creation in a product category is indeed sensitive to the category’s revenue concentration. Specifically, when category concentration is low, user satisfaction falls as demonstrated by significantly lower app ratings, but as the category becomes more concentrated, new product innovation and entry of new developers declines and user growth stalls. Successfully managing the categories means balancing these sets of tradeoffs.

How Apple manages product categories in the presence of direct network effects

One way in which Apple manages product categories is through the selective promotion of apps via the Editors’ Choice Awards. These awards, which Apple uses sparingly (only 0.04% of games in our sample receive the award), are typically given shortly after an app’s launch and are highly sought after by developers because they are associated with a significant increase in adoption.

Using awards makes an app more visible and enticing and allows Apple to nudge users towards the app and help it build a user base. The returns to such awards are higher for games with direct network effects as the initial boost can create a virtuous cycle that drives adoption and growth, which is why we find Apple is more likely to award a game with direct network effects than those without. Apple also pays attention to the developer’s track record, favoring those with strong prior performance. No use stoking network effects if the developer lacks the means to grow the installed base. 

Apple pays attention to the market structure of the category and appears to strive for balance. Apple is more likely to award a game with direct network effects when the category’s concentration is low. Pushing users to the same game(s) builds the installed base and creates value for players. When the category becomes concentrated, Apple uses awards to curtail the dominance of the leading developer(s). We find that Apple is significantly less likely to award a dominant developer in a concentrated category, in fact, Apple is more likely to award a challenger’s new game. In essence, Apple strives for a “Goldilocks” market structure—not too diffuse but not too concentrated. For Apple, the “porridge is just right” when multiplayer games have enough players to create high utility but enough competition among them to provide enough choice for players and keep complementor power in check.

Managerial and policy implications

Our research provides a novel perspective on how platform owners manage complementor products when these products have their own network effects. In the presence of network effects, it can be difficult to strike a balance between sufficient adoption and a market that tips toward a winner-take-all scenario. When given rarely and early in the product’s life cycle, using awards or other selective promotions can be an effective mechanism to shepherd users to products so to achieve the platform owner’s preferred market structure. This departs from their traditional use as a means of highlighting the best products or awarding good performance.

While such tactics may create value for users, micromanaging market structure can draw the ire of regulators and policymakers, especially when the platform owner is viewed as undermining successful complementors. However, using soft nudges like awards may be viewed with less skepticism than other, more draconian measures that overtly limit the performance of a dominant complementor. Importantly, our work highlights the difference between networked products and non-networked products. Regulators need to appreciate these differences when assessing the effect of platform owner tactics on consumer welfare.

This blog is based on Shiva, Cameron and Martin’s research published in the Strategic Management Journal, which is included in the Platform Papers references dashboard:

Agarwal, S., Miller, C. D., & Ganco, M. (2023). Growing platforms within platforms: How platforms manage the adoption of complementor products in the presence of network effects? Strategic Management Journal, 44(8), 1879-1910.

Platform-Paper updates

Today’s blog has several touchpoints with prior Platform Paper blogs. For example, for an excellent overview of how platforms can orchestrate their ecosystems through effective governance, see this blog by Melissa Schilling. Or, see last month’s blog by Chiara Farronato on when winner-take-all dynamics are and aren’t likely to occur in platform markets (bonus: it features dogs!). Finally, for further discussion on network effects and freemium online multiplayer games, see one of my earlier blogs.

Moreover, several interesting papers have been added to the Platform Papers references dashboard this month. Here are a couple of highlights:

• Platforms have various tools in their toolkit when it comes to managing their ecosystems (including badges and awards as illustrated by today’s blog). Besides setting prices on both sides of the market, platforms can make investments in a platform’s standalone value, add social features to take advantage of same-side network effects, or develop integration tools and boundary resources to facilitate third-party content creation. In a paper published in the Journal of Management Information Systems, Edward Anderson, Geoff Parker and Burcu Tan develop a simulation model exploring which of these tools platforms should focus on at different points in their life cycle. While pricing is something platforms must constantly tweak, the importance of the other tools varies over time. Their importance also differ with the platform’s monetization model (who’s getting charged, who’s being subsidized and to what extent). The authors tweak the model for different types of platforms; mobile platforms, social media, the sharing economy, and business-to-business platforms.

• There’s plenty of evidence regarding digital platforms’ disruptive forces on traditional industries. Empirical studies have mostly looked at the cross-elasticities between ridesharing platforms and taxi’s and between Airbnb and hotels (consistently finding that such cross-elasticities exist). A recent paper published in Marketing Science adds to the body of evidence by studying the effects of market entry by for-sale-by-owner platforms (e.g. people selling their own homes online) on real estate agents’ listing prices. The author, Qiyuan Wang, finds that the introduction of a Chinese FSBO platform leads to a 2.8% decrease in listing prices, a percentage comparable to the commission rate typically

  charged by agents. Moreover, the impact of reduced listing prices charged by the vendors extends to the final sales price of the properties.

That’s all for now. See you next month!

Platform Papers is curated and maintained by Joost Rietveld.

What a merger between two pet-sitting platforms can tell us about network effects

Platform Papers is a monthly blog about platform competition and Big Tech. Blogposts are written by prominent scholars based on their research. The blog is linked to platformpapers.com, an online repository that collects and organizes academic research on platform competition.

This blog is written by Chiara Farronato

How many times have you heard managers, regulators, and academics talk about network effects in digital platforms? Network effects arise when the value a participant derives from joining a platform is an increasing function of the number of other participants on the same platform. Network effects are as old as your grandparents’ home telephone. Back then, there was no use in a phone if none of the people you wanted to talk to also had it. And the more friends and family had a phone, the more you benefited from buying one to communicate with them.

Network effects

Fast forward almost 150 years, and many of the most valuable companies today benefit from network effects. Meta, Facebook’s parent company, is currently valued at USD 780 billion thanks to its ability to attract almost 3 billion active monthly users, which in turn make the platform attractive to millions of advertisers. The same is true for Amazon Marketplace, where millions of sellers and hundreds of millions of buyers exchange products globally.

When network effects are strong, having market participants join a single dominant platform rather than spreading across many competing platforms maximizes value creation. This implication was not lost on Theodore Vail, the fourth president of the Bell Telephone Company, who in the early 1900’s, used network effects to argue that Bell Telephone should have a monopoly on telephone networks. Since then however, network effects have been used to justify launch strategies and acquisition decisions across a variety of industries, not all of which in reality exhibit strong network effects.

How can we quantify the strength of network effects? The truth is: it’s hard. In theory, one would want to measure participants’ utility and how it varies as the number of platform participants changes. However, changes in the number of platform participants are often correlated with other events that may affect participants’ value from a platform. Think about Amazon Prime Day, when many deals are available to Prime members. On those days, you’re likely going to see a sizable increase both in the number of sellers listing products and people shopping on Amazon. But the increase in transaction volume—a clear sign that buyers and sellers benefit from Amazon—is due to the presence of deals and discounts rather than network effects.

Dog eat dog: The merger of Rover and DogVacay

In a Management Science paper that I wrote in collaboration with Jessica Fong and Andrey Fradkin, we found an ideal setting where a platform experienced a sudden increase in the number of its participants. Rover, the largest pet sitting platform in the US, acquired DogVacay, its largest competitor, in the Spring of 2017. By the summer of the same year, Rover had shut down DogVacay, and asked pet owners and sitters to migrate to rover.com.

Because pet owners typically look for pet sitters close to their home, the geographic variation in the size of the two competing platforms implies that Rover experienced larger increases in the number of platform participants in some cities—those where DogVacay was large—compared to others. Thus, if network effects were real, we would expect greater benefits in cities where Rover experienced a larger influx of participants from DogVacay. This is indeed true in our context, where we find clear evidence of network effects. But our results do not end there.

As much as pet owners and sitters benefit from being able to find a trading partner with whom to transact, and the likelihood of that match increases when there are more pets and sitters to choose from, there’s a reason why not all pet owners and sitters hadn’t already converged to a single dominant platform. Academics call that reason “horizontal preferences.” The fact that some consumers may prefer red shoes while others may opt for blue shoes is an example of horizontal preferences. If people have horizontal preferences over Rover and DogVacay (perhaps because of their web design, or brand reputation), eliminating DogVacay may hurt pet owners and sitters for whom DogVacay was their favorite alternative. And if this effect were too large, it could end up offsetting the network benefits of the merger.

If horizontal preferences were strong, we would expect that pets and sitters from DogVacay  would have a harder time finding matches on Rover, and would leave as a result. Our analyses strongly support this hypothesis. But, why would DogVacay users prefer DogVacay? After all, both platforms were designed to match pets with sitters, with similar pricing, search algorithms, payment systems, and review mechanisms.

We find evidence for repeat exchanges and switching costs to play an important role in explaining why DogVacay users left after the merger with Rover. Owners often prefer their pets to stay with the same sitter over time. The migration to a new platform likely made finding the same sitter harder, especially if the sitter left altogether rather than migrating to Rover. Owners may also find their previous sitters offline, making disintermediation an additional driver of attrition. Surprisingly, new users had similar experiences, suggesting that horizontal preferences do not simply originate from experience gained while using a particular platform.            

Managerial and policy implications

Overall, we show that platform differentiation is in practice an important factor in offsetting network effects. Our results imply that a single dominant platform may not be as effective as multiple platforms, both from a strategic and antitrust perspective. The antitrust perspective is more obvious. Regulators are interested in ensuring that consumers have multiple options to choose from, because that creates healthy price competition and incentives for companies to keep innovating. Even though antitrust authorities have historically been hesitant to get involved in regulating digital platforms, the tide has already changed in Europe with the passing of the Digital Markets Act, and may soon be changing in other countries. After all, even the Bell telephone monopoly system was eventually dismantled.

The strategic perspective is more subtle. Network effects are often assumed to be large enough to warrant mergers and acquisitions strategies that progressively concentrate activity on a single dominant digital platform. However, the simple presence of network effects is not sufficient to justify these strategies. Instead, it may be beneficial for a company to operate multiple platforms rather than combining them. In fact, there are many instances of acquisitions where the acquired platforms remained operative—e.g., Zillow and Trulia, or the many online travel sites within the Booking Holdings group.

Beyond mergers, our study calls into question the importance of a first-mover advantage and the likelihood of a winner-take-all equilibrium, which have historically pushed platforms to invest heavily to achieve scale fast and deter competitive entry. In fact, despite network effects, entry and competition are likely in equilibrium, where multiple platforms can coexist and new platforms can successfully enter by identifying niche consumer preferences. This may be why in the ride-sharing market, Uber and Lyft are still competing, and why Alto may end up becoming a thriving alternative for the luxury segment.

In this evolving narrative of network effects, the true strength lies not solely in the number of participants, but in the variety of choices. The finale of this story is yet unwritten, but it’s clear that a balance of platform size and platform variety will guide the next act in the ever-changing digital landscape.

This blog is based on Chiara’s research forthcoming in Management Science, which is included in the Platform Papers references dashboard:

Farronato, C., Fong, J., & Fradkin, A. (2023). Dog eat dog: Balancing network effects and differentiation in a digital platform merger. Management Science.

Platform Papers is edited and published by Joost Rietveld.

The Role of Superstar Exclusivity

Platform Papers is a monthly blog about platform competition and Big Tech. Blogposts are written by prominent scholars based on their research. The blog is linked to platformpapers.com, an online repository that collects and organizes academic research on platform competition.

This blog is written by Elias Carroni, Leonardo Madio, and Shiva Shekhar.

In the dynamic world of digital markets, platforms play a crucial role in facilitating interactions between various groups of agents with varying degree of market power. Some of these agents significantly influence consumer participation on a platform and can be considered superstars. Critical questions are: how does the presence of superstars shape competition between platforms, and what are the effects of platforms’ acquisitions of superstars?

The role of superstars

In many digital markets, platforms host a mix of professionals and “complementors”, where the former, especially superstar artists or developers, tend to be more attractive to consumers. For instance, music streaming platforms like Spotify and Apple Music feature established superstars such as Beyoncé and Taylor Swift, along with a long tail of emerging talents. Similar examples exist in mobile app markets, where popular applications like WhatsApp and Instagram coexist with independent ones and in the video-games market, where well-known gaming studios, such as Activision Blizzard and Ubisoft, among others, produce and distribute AAA+ games (e.g., Destiny, Battlefield, and the Call of Duty). The relevance of superstars on platforms extends to other domains such as open-source software, app-stores, news, and sports broadcasting.[1]

The value of exclusivity

Superstars can provide a competitive edge to platforms, particularly when they offer exclusivity to one platform over other(s). For example, in the music industry, Taylor Swift released her album 1989 under time-limited exclusivity conditions on Apple Music before making it available to other streaming platforms. Similar deals have been signed by other major artists and podcasters, and the practice has also been adopted by Apple Music’s competitors, Tidal and Spotify. In the gaming industry, Sony has several AAA games exclusive on their PlayStation consoles (e.g., God of War, The Last of Us).

Exclusivity can enhance a platform’s market reach and allow consolidation of its competitive position because the presence of cross-group externalities can also spur within-group externalities. For instance, within-group externalities can be easily spurred by exclusivity in the video game industry due to multiplayer games, which results in the formation of communities. Indeed, many users would migrate to a platform hosting the superstar exclusively. Therefore, superstar exclusivity jump-starts a “network effects feedback loop” (see Figure 1) where more complementors also associate with the platform favored by the superstar, further attracting additional consumers. Thus, consumers and complementors tend to agglomerate on the platform favored by the superstar, resulting in direct and indirect gains and market capture for the favored platform.

From the perspective of a superstar, the exclusivity choice involves a complex trade-off. Specifically, on the one hand, by offering its product exclusively, the superstar can benefit from a more lucrative fee charged to the favored platform. On the other hand, exclusivity implies a lower consumer reach (i.e., only those consumers who patronize the favored platform) resulting in lower engagement and ancillary revenues. For instance, ancillary revenues can take the form of royalties from streaming, tickets and merchandize sales, and in-app purchases. This complex trade-off determines whether a Superstar decides to be exclusive on a platform or non-exclusive.  

In our study titled “Superstar exclusivity in two-sided markets”, forthcoming in Management Science, we develop an economic model to analyze exclusivity decisions made by superstars and their effects on competition between platforms. In addition, we investigate how exclusivity decisions change after a superstar has merged with a platform.  

Main findings

Given that exclusivity requires the superstar to limit its consumer reach, the exclusive fee charged to the favored platform must compensate for the foregone revenues achievable under non-exclusivity. We show that a superstar finds it profitable to be exclusive on a platform only if it can substantially enhance consumer participation. This jump-starts the network effect feedback loop attracting more complementors and, thus, more consumers. As a result, the superstar’s ability to charge for exclusivity increases while minimizing losses from limited market reach (e.g., foregone ancillary revenues). In contrast, if the superstar cannot attract a substantial base of consumers by signing an exclusive contract, non-exclusivity is more profitable because it allows for a wider market reach and higher ancillary revenues. We find that exclusivity can lead to welfare gains when platform market participants find it highly valuable to interact with each other, i.e. when network effects are strong.

The above discussion has important real-world implications.

In markets where consumers are not very “mobile” due to lock-in effects, a superstar has a lower ability to attract significant masses of consumers and thus jump-start the network effects feedback loop. This makes it unprofitable for the superstar to offer exclusive deals in the first place. For example, these lock-in effects exist in markets where consumers must pay upfront the cost to participate on a platform such as buying expensive electronic devices (e.g., smartphones or gaming consoles). In such markets, the likelihood of exclusive deals is expected to be lower since a new exclusive content would not trigger immediate consumer switching. Note, however, that one may still observe exclusivity in such markets. Often such exclusives are the platforms’ first party apps (e.g., Apple’s suite of apps).

In contrast, in markets where consumers are very “mobile” and have low entry and switching costs, one would expect a higher likelihood of exclusive deals by superstars. For instance, it is quite common for superstars to offer exclusivity in the music and video streaming markets, where subscription fees are low and cancellation is easy. For instance, Spotify, the largest music streaming service by volume, allows cancellation of subscription anytime.

Does superstar acquisition by platforms enhance the likelihood of exclusivity?

In recent times, the acquisition of a superstar by a platform has been a topic of intense discussion and concern in policy circles. A notable example is Microsoft’s proposed acquisition of Activision Blizzard, which many competition authorities initially challenged and cleared afterwards. One primary concern was that Microsoft would shut-down access to Activision Blizzard’s games for competing consoles, in particular Call of Duty.                                      

When a platform acquires a superstar, the new merged entity (platform-superstar) decides whether to withhold access or license the superstar content with rivals. If the superstar content is licensed to the rival, the merged entity benefits from the largest market reach of the superstar content, making the platform market less competitive. Indeed, the two platforms offer the same value to consumer and there are no incentives for consumers to switch and jump start the network effects feedback loop. If, instead, the superstar content is withheld, the merged entity becomes more aggressive in the market, and this intense competition reduces the rival’s profit. Against this backdrop, the merged entity can negotiate better terms and conditions for licensing the superstar content to the rival platform. As a result, the merged entity may have fewer incentives than before the acquisition to engage in exclusivity (i.e., withholding the superstar content).  

Implications for managers

There are several implications for strategies that platform owners, managers of superstars and complementors can take into account when deciding which platform to affiliate with. First and foremost, in markets with cross-group externalities, exclusivity is the most profitable choice in platform environments with sufficiently intense inter-platform competition. One way to measure the competition’s fierceness is to assess consumer preferences and their incentives and abilities to switch from one platform to another. Securing a large number of switchers is key to ensuring the profitability of such a strategy.

Second, the decision of superstars to go exclusive to one platform should be monitored closely by small complementors, who can, in turn, appropriate some of the benefits generated by the platform’s decision. Our analysis suggests that those complementors with high costs to adapt their products to the platform’s specificities can benefit the most from following the superstar and going exclusive as well. Superstar presence can help complementors break into the market and increase variety and differentiation.

Final thoughts

Understanding the dynamics of superstar exclusivity in platform competition is crucial for navigating the ever-evolving digital market landscape. To drive effective strategies and ensure balanced antitrust policies, managers and policymakers should carefully consider the power of exclusivity, the trade-offs involved, and the implications for various market participants.

This blog is based on Elias, Leonardo, and Shiva’s research forthcoming in Management Science, which is included in the Platform Papers references dashboard:

Carroni, E., Madio, L., & Shekhar, S. (2023). Superstar exclusivity in two-sided markets. Management Science.

[1] A more detailed discussion on the presence of Superstars in various market contexts can be found in the online appendix of our paper.

Platform-Paper updates

Several interesting papers have been added to the Platform Papers references dashboard in the last month. Here are some highlights:

• A study on eBay’s Top-Rated Seller badges offers a nice illustration of how a platform’s selective promotion of certain sellers can spill over to the rest of the platform. In the journal of Quantitative Marketing and Economics, Xiang Hui and colleagues find that when eBay offered financial incentives to sellers with the Top-Rated Seller badge for offering fast handling and generous return policies on their listings, other, non-incentivized, sellers also were more likely to adopt the promoted behavior. This suggests that a platform’s targeted actions can have wider, ecosystem-level implications.

• Buyers often rely on a platform’s search algorithm to reduce search costs and help them make decisions. In the International Journal of Industrial Organization, Yangguang Huang and Yu Xie study the implications of such search algorithms not being fully equal—as they often are. “If a platform adopts a highly unequal search algorithm, buyers are likely to obtain repetitive information about a small group of sellers, which causes buyers to consider fewer options and suppresses competition.” Analyzing data from food delivery platforms, the authors provide evidence that markets with less equal distributions of search results have higher average prices and more concentrated sales. They point to the design of platforms’ search algorithms as a potent avenue for antitrust regulation.

• Notwithstanding the increased risk of disintermediation, allowing buyers and sellers to communicate with each other can increase the chances of a successful transaction on a platform. In the context of a peer-to-peer platform for long-term real estate rental properties, Xia Zhao and colleagues, in Information Systems Research, find that direct communication between the renter and the host can lead to the renter choosing a relatively more distinct property compared to the average property under consideration. Moreover, renters tend to be happier with their choice following their exchanges with renters, suggesting they make more informed choices.

• In Management Science, Alan Kwan and co-authors study the dynamics of crowd-judging on two-sided platforms. Crowd-judging is a novel crowdsourcing mechanism whereby buyers and sellers volunteer as jurors to decide disputes arising from transacting on the platform. Studying data from Chinese e-commerce platform Taobao, the authors find that jurors tend to suffer from in-group bias: buyer jurors favor the buyer in disputes and vice versa. The extent of bias reduces with juror-level experience and when the platform dynamically allocates disputes to a diverse pool of jurors.

Enjoy the rest of your summer and see you next month!

Platform Papers is curated and maintained by Joost Rietveld.

“The crowd” can be a valuable resource for platforms

This blog is written by Johannes Loh and Tobias Kretschmer.

The internet is, and always has been, shaped by content produced by “the crowd”, that is,  individual contributors providing input freely. For instance, Wikipedia has rendered traditional encyclopedias obsolete by drawing on the huge variety of knowledge provided by (more or less) regular users who spend their time writing articles, seemingly out of the kindness of their hearts. And the content they provide is of high quality: As early as the year 2000, a crowd of “clickworkers” helped NASA classify images from Mars to determine the planet’s age – they did as good a job as trained scientists. Since then, platforms have tapped into this resource in many ways: Firms like Starbucks and Doritos have launched high-profile crowdsourced promotional campaigns, Lego and Threadless let users design new products, and Uber or Airbnb – as well as virtually all e-commerce platforms – rely on user-written reviews to establish trust between service providers and consumers. Clearly, the idea of involving the crowd in a platform’s value creation activities is intriguing, as it can constitute a source of content that is both of high quality and variety. In addition, users typically volunteer their time and effort for free, making it a low-cost resource as well.

Leveraging “the crowd” for competitive advantage is challenging

Research has looked into what drives these volunteers to become active in the first place, and it has identified a wide range of non-pecuniary sources of motivation. For example, some contribute to open-source software development because they see the need for a particular solution which is not provided by firms or because they want to hone their skills. Others become active for the social benefits it provides: They enjoy working with other like-minded volunteers and attaining a position of high social status within their online community. While these motivational sources can substitute for financial incentives – platforms do not have to pay their contributors – they also make it challenging to both attract and manage the crowd. First, crowdsourced platforms are usually subject to strong network effects – the larger the community, the stronger the social benefits. This generates “winner-takes-all” dynamics and platforms will have to fiercely compete for participation. And second, because volunteers participate for their own enjoyment, it is challenging to direct them towards activities the platform owner has in mind. Put simply, they alone choose if, when, and to what extent to contribute, not beholden to the interests of the platform that hosts them. Hence, while the crowd has the potential to be a valuable resource, it is highly uncertain that platforms are able to actually leverage it for competitive advantage.

An open question: How are platform competition and contributor activity interrelated?

Which, then, are conditions that can lead to the crowd actually being a source of competitive advantage for platforms? This is the broad question we seek to answer in our article “Online communities on competing platforms: Evidence from game wikis” published in the Strategic Management Journal. We study two competing platforms, Fandom and Gamepedia, that host several crowdsourced game wikis. Similar to Wikipedia, community members write and maintain articles about different games that can be read by the public. However, instead of being one single repository of knowledge, wikis in our setting are distinct entities that cover a single game and that are maintained by distinct online communities. In addition, both platforms have a strong interest in hosting large, well-maintained wikis to attract an audience which they monetize via advertising. A key feature of our study is that their ability to attract such communities varies substantially across different games. That is, Gamepedia has an advantage in the coverage of some games, and Fandom in the coverage of others. To answer our research question, we investigate how the activity and contributions patterns of these online communities differ between more and less successful domains (as represented by games coverage).

Activity and contributions patterns between more and less successful crowdsourced platforms

In our study, we identify four conditions under which the crowd can provide a platform with a competitive advantage:

1. Success breeds success: Not unlike other types of platforms – such as e-commerce or social media – having a strong competitive position today will help in attracting and motivating subsequent volunteer activity. There are two reasons for this: First, new contributors are more likely to join a more active and well-maintained community. And second, a better competitive position comes with greater non-pecuniary benefits for contributors – they reach a larger audience with their content, and they can collaborate with a greater community.

2. “Personpower” matters most: Community size is the single biggest factor that determines a platforms’ competitive advantage. Having more community members contributing means more content on the platform – it is as simple as that. Further, it outweighs any advantages arising from network effects or non-pecuniary benefits which can drive individual contributors’ activity.

3. Social benefits also matter: Each individual volunteer contributes more in a larger wiki community. This can be attributed to social benefits, that is, they enjoy collaborating with like-minded peers. Hence, crowdsourced platforms are subject to direct network effects arising from these non-pecuniary sources of motivation.

   

4. Contributor heterogeneity matters: Not all community members are equally active. In fact, most content is produced by a fairly small group of high-productivity contributors in each community, and failing to attract (or retain) them will jeopardize a platform’s competitive standing. In addition, this very important group of core contributors does not seem to care much about the community aspect of crowdsourcing. Instead, they are driven by the satisfaction and sense of accomplishment that comes from contributing to the “winning” platform. Moreover, not only are high-productivity contributors the greatest content producers, but they also engage in policing and maintenance activities to ensure the high quality of their wiki.

What can owners of crowdsourced platforms do to attain a competitive advantage?

These four conditions have implications for potential strategies that platform owners can pursue to attain an advantage. First, community size and the presence of high-productivity contributors are the most important factors. However, neither are easily attained or imitated. Instead, acquiring a competing platform may be a more promising avenue. As a case in point, Fandom acquired Gamepedia in early 2019 following years of coexistence and competition in the area of video game wikis. In addition, Microsoft acquired GitHub, the most popular platform for open-source software development, in an effort to tap into crowdsourcing. Second, because size in itself is more important than individual contributors’ non-pecuniary benefits, strategies aimed at growing the community fast are more important than those aimed at motivating its members. Consistent with this, interviews with wiki community members revealed a focus on search engine optimization to increase website traffic and, in turn, community growth. And finally, the reliance on high-productivity contributors is a double-edged sword. They are an important determinant of competitive advantage on the one hand. But on the other hand, interviews with community members confirmed that they do not respond well to heavy-handed attempts at directing their activities, which makes managing them challenging. Instead, platform owners have to trust their judgment and consider involving them in some decisions that affect their community.

This blog is based on Johannes and Tobias’ research published in the Strategic Management Journal, which is included in the Platform Papers references dashboard:

Loh, J., & Kretschmer, T. (2023). Online communities on competing platforms: evidence from game wikis. Strategic Management Journal, 44(2), 441-476.

Platform-Paper Updates

In this month’s Platform Papers update, some platforms for good, some platforms for bad… All outstanding scholarly contributions though:

• With all the regulatory scrutiny big platforms are facing these days, we can easily lose track of the societal benefits that platforms bring. In a recent paper published in the Journal of Product and Innovation Management (JPIM), Paavo Ritala outlines how platforms can help to address ‘Grand Challenges’ by way of their coordination structures, instigation and maintenance of collective action, and generativity potential. Clearly, platforms can be a force for good!

• Transitioning to the ‘bad’ then, in Management Science, Liu, Lou, and Li study the unintended consequences of advances in matching technologies. The authors argue that platforms may have an incentive to obfuscate high quality matches between buyers and sellers on their platform. High-level match quality can reveal information about the thickness of the market. If demand is thin, platform suppliers might decide to leave the platform resulting in a loss of revenue.

• Continuing on this theme, Long and Liu argue in a paper published in Marketing Science that “a platform may manipulate an inferior seller’s product to appear more attractive to intensify sellers’ competition to bid for advertising and also manipulate a superior seller’s organic placement to either compensate or penalize the superior seller.” Put differently, it’s not only buyers and sellers that can at times be considered as bad actors, but also the platform itself… Yikes!

• Finally, in The Journal of Industrial Economics, José Ignacio Heresi looks at platform price parity clauses (e.g., contractual clauses that stipulate platform suppliers to not offer lower prices or better terms on other platforms or their own (digital) storefronts than on a focal platform. Such clauses have been observed on Booking.com and Valve’s Steam platform). The main takeaway from the study is that such clauses aren’t great for consumers: “when price parity clauses are endogenous, they are only observed in equilibrium if they hurt consumers.”

These and several other papers were added to the Platform Papers references dashboard in the last month.

Management without managers?

Platform Papers is a monthly blog about platform competition and Big Tech. Blogposts are written by prominent scholars based on their research. The blog is linked to platformpapers.com, an online repository that collects and organizes academic research on platform competition.

This blog is written by Ying-Ying Hsieh & JP Vergne

In 2009, a new organizational form, which we call the decentralized platform, began to diffuse without relying on managerial authority and without having to employ anyone. The most well-known decentralized platform, Bitcoin, has millions of users, thousands of contributors, and a market valuation never achieved in history by an organization without a CEO nor shareholders. 

Based on this novel blueprint, Web3 has quickly emerged as a new way of interacting with the Internet. Unlike Web 2, where coordination between digital platform participants is orchestrated centrally by a corporation, Web3 is rooted in the idea that users should own and exchange their data without relying on corporate intermediaries. To achieve direct coordination, Web3 requires decentralized platform infrastructures that can facilitate exchange activities without the intermediation of centralized platforms (e.g., Gmail, Uber.com, Instagram).

How do decentralized platforms operate?

Decentralized platforms provide an alternative to corporate platforms for coordinating the exchange of digital assets, but without the “visible hand” of corporate managers, how does such coordination happen exactly?

Our study “The future of the web? The coordination and early-stage growth of decentralized platforms” demonstrates, based on in-depth qualitative and quantitative analyses of twenty blockchain platforms, that Web3 platforms combine three components for operating in the absence of managers: 

(1)   Decentralized Algorithmic Coordination. In the absence of a manager, participants must rely on algorithms to coordinate tasks. Instead of residing in contracts drawn out by corporate managers, platform rules are encoded in blockchain algorithms that define how participants interact and share data representing value. As long as one has trust in the algorithms, then potentially, one need not have trust in any other individual or organizational collaborators. One can then transact without having to know the real-world identities of their counterparties.

Specifically, blockchain algorithms encode design rules and programs within a shared space and time, thereby regulating routine task performance (e.g., when to perform a specific task with required resources) and rewards distribution (paid out in cryptocurrency). Design rules include how information shared across the network is verified and validated, thus providing a consistent basis for trust in the algorithms.

(2)   Decentralized Social Coordination. Despite the absence of managerial orchestration, social coordination among key contributors remains indispensable to ensure ongoing maintenance of the platform. On decentralized platforms, social coordination follows a similar structure to open-source software development (OSSD)—a common format for the developer community to coordinate their contributions to software upgrades. While in centralized platforms, proprietary development typically involves corporate employees only, decentralized platforms are developed by voluntary contributors who self-select into the project, much like Linux and Wikipedia contributors. In addition to development, decentralized platforms differ in that the verification and validation of routine tasks are performed by network validators (e.g., “miners”, “stakers”), rather than by middle managers. Social coordination ensures that sophisticated checks and balances are in place to account for the respective interests of developers and validators.  

(3)   Decentralized Goal Coordination. While traditionally, organizational goals are set by central authorities (e.g., the CEO), in the absence of managerial orchestration, decentralized platforms sometimes rely on nonprofit foundations to corral everyone’s efforts. Despite having no formal authority, for instance, the Ethereum Foundation can offer a coherent vision of how various classes of platform participants might fit together to achieve long-term goals. Based on its vision, a foundation can then offer knowledge and resources (e.g., developer toolkits and grants) to platform contributors who buy into this imagined future and are willing to help build it.

Coordination Configurations: A Balancing Act

The three mechanisms depicted above do not work in isolation, nor can they be added arbitrarily to the design of decentralized platforms. Without a centralized authority to provide a stable structure, participants must negotiate decentralization by themselves, leading to a fragile equilibrium that ideally should deliver the kind of accountability, predictability, and common understanding necessary for sustainable coordination. Finding the right balance is key to making headway to early-stage growth.

Our findings suggest that, to fuel early-stage platform growth, platform contributors should either empower a small group of core developers or create a foundation to provide a guiding vision. Intriguingly, we did not observe any decentralized platform that grew successfully by decentralizing all three components of coordination at the same time. We also did not find evidence that fully decentralized algorithmic coordination was sufficient to bootstrap a platform. That said, we did find that insufficiently decentralized algorithmic coordination could cause instability and lead to early-stage platform decline.

So, what do decentralized platforms mean for businesses and organization designers?

Fundamentally, blockchain is a coordination technology that requires group adoption. Organizations can only benefit from adopting decentralization technologies and decentralized platform business models if they understand the limits of decentralization. This understanding is particularly important for organization designers, as the adoption of a decentralization technology entails fine-tuning the degree of decentralization across different platform layers to optimize coordination.

As the first iterations of decentralized platforms, cryptocurrencies have grown beyond their initial use cases to help allocate tasks, resources, rewards, and information across a range of industry sectors. As a result, blockchain-based decentralized platforms provide infrastructure for bourgeoning Web3 applications such as DeFi, NFTs, and DAOs. Those DAOs (for “decentralized autonomous organizations”) increasingly populate the Web3 space, but already experience the limits of decentralization first-hand as centralizing forces spontaneously resurface amidst their disengaged token holders, who end up relinquishing control to a few insiders—those with the resources, competence, and time to participate in every aspect of decision-making.

Having diffused beyond cryptocurrencies, decentralized platforms will warrant closer attention from industry, academia, and regulators alike as they embody a new class of competitors in the digital economy.

This blog is based on Ying-Ying and JP’s research published in the Strategic Management Journal, which is included in the Platform Papers references dashboard:

Hsieh, Y. Y., & Vergne, J. P. (2023). The future of the web? The coordination and early‐stage growth of decentralized platforms. Strategic Management Journal, 44(3), 829-857.

Platform-Paper Updates

It’s been a relatively slow month in terms of new platform papers being published, but there have been plenty of intersting developments in the ‘real world’:

• Last week, the inaugural summit of the European Digital Platform Research Network (EU-DPRN) took place at Bocconi University in Milan. EU-DPRN aims at bringing together European scholars studying platforms from across different disciplines (e.g., strategy, marketing, information systems, economics, etc.). The first summit was a success; it featured a PhD workshop, ten paper presentations, six poster presentations, a keynote and two stimulating panels. Next year, the conference will be hosted by the UCL School of Management in London!

• The competition law community is increasingly interested in so-called ecosystem theories of harm. When large platform companies increase their scope by acquiring a target in a complementary or unrelated market, this can both benefit the acquiring firm in novel ways as well as potentially disadvantage competitors. The question is, do we need new theories of harm to analyze such acquisitions? It’s not an easy question. Recently, two insightful CEPR columns (1 & 2) got published on the topic and earlier this week the British Institute of International and Comparative Law (BIICL) hosted a roundtable debate on the topic.

Platform Papers is curated and maintained by Joost Rietveld.

Inducing more reviews may not help the market and other thoughts on the difficulty of improving reputation systems

This blog is written by Andrey Fradkin, with special thanks to David Holtz.

Reviews are everywhere on digital platforms, yet it is common knowledge that review systems are flawed. Not all users contribute reviews and those who do may have unusual experiences or preferences. In the course of writing a review, the lack of an objective standard for mapping experiences to ratings leads to inconsistencies amongst reviewers. Reviewers may also leave out relevant details to appear amicable or may exaggerate negative experiences as a form of retaliation. Despite these issues being widely recognized and discussed in the media, it seems review systems have barely improved over the past two decades. In this blog post, I will delve into the challenges of measuring reputation system improvements by examining a recent paper by David Holtz and myself.

Before discussing our results, it’s useful to consider the purpose of review systems. Review systems were introduced in the earliest digital marketplaces such as eBay to solve the problem of transacting with anonymous strangers on the internet, who could misrepresent their product or fail to deliver it altogether.[1] Reputation systems incentivized sellers to act honestly while providing buyers with the information needed to avoid potentially low-quality sellers. This created a virtuous cycle where more buyers began to trust eBay, while sellers joined eBay for access to a larger market.

While the success of eBay’s reputation system demonstrates the utility of reputation systems for most marketplaces, it doesn’t provide a blueprint for designing the best system. Platform designers must choose how to solicit reviews and how to display information from reviews. Startup platforms begin with a reputation system that is chosen by early employees based on intuition. As platforms grow, they have the opportunity to use data and experiments to improve these reputation systems. Yet the seemingly unchanging design of reputation systems suggests that this opportunity is not being seized as much as it should be.[2]

One reason for the lack of large changes to reputation systems is the lack of clarity about what signals are indicative of a better reputation system. Data scientists and researchers are often tempted to focus on immediate outcomes that are easier to measure. For example, policies such as reminders, nudges, and incentives can be shown to improve the quantity of reviews.[3] The effects of these policies on reviews are easy to evaluate because reviews come shortly after interventions, and it’s easy to map treatment assignment of potential reviewers to review behavior.

It is more difficult to measure how shifts in reviews translate to improved outcomes for the market. In a forthcoming paper in Marketing Science, David Holtz and I take an initial step in mapping review policy to market outcomes using a large-scale field experiment on Airbnb. The experiment was assigned at an Airbnb listing level for listings which did not yet have reviews. For treatment listings, buyers were offered a $25 coupon post-transaction to review. For the control listings, buyers were not offered any incentive to review after a transaction. The incentives increased review rates and induced more negative reviews (as measured through star ratings and text).

Our primary insight involves examining the impact of these reviews on the quantity and quality of matches for listings. To do so, we connect a listing’s initial treatment (whether a guest to the listing is offered a coupon to review) to its subsequent outcomes. Specifically, we track the listing outcomes for a year following the treatment assignment. This enables us to assess the effects of incentived reviews on the number of transactions and match quality indicators, such as subsequent reviews and complaints. We also provide a measure of match quality based on the subsequent experiences of guests who engaged in transactions with listings from the experiment. Note that this measurement exercise involves merging data up to two years of data after the experiment concluded.

We find that while the treatment induced more reviews, these reviews failed to improve listing outcomes both in terms of nights booked or in terms of match quality. In fact, we find that measures of match quality actually fell due to the treatment. Our key theory for the fall in match quality, which we support with additional analysis, is that the induced reviews were actually less correlated with quality than non-incentivized reviews. We find these results even though the induced reviews were more negative than non-incentivized reviews, which may naively suggest that they are more accurate.

Our results show that it is insufficient to only look at the quantity and valence of reviews in order to evaluate whether a reputation system design change improves market outcomes. Interestingly, this is our second failure to find benefits from a reputation system change on Airbnb. In a recently published paper, David Holtz, Elena Grewal, and I found that making the reputation system simultaneous reduced reciprocity and increased review rates. Yet, even with large samples, we failed to detect effects of this change on market outcomes.

Our findings, while specific to certain reputation system changes and the context of Airbnb, offer valuable insights into reputation system design more broadly. Had we solely focused on the effects of these changes on reviews, we might have erroneously concluded that these alterations significantly benefited the platform. Other platforms should also adopt this long-term analysis of transaction volume and match quality when experimenting with reputation system design.[4]

Circling back to the motivation of this post, I propose that a lack of innovation in reputation systems is, at least in part, due to the absence of a robust framework for evaluating changes in reputation system design. It is hard to advocate for resources in an organization when data scientists and product managers might struggle to assess the key performance indicators of a product change. It is also hard to advocate for longer-lived experiments and long-run analyses when product development cycles are often much shorter than a year.

What can data scientists and managers do to alleviate these constraints to improving and analyzing reputation systems? An obvious suggestion is to advocate for longer and more ambitious reputation system experiments. Additionally, it can be worthwhile to reserve time to re-analyze old experiments using a longer data horizon. Such re-analysis can yield surprising insights that will have implications for future design decisions. Statistical techniques such as surrogate analysis can also be used to learn about the longer-run effects of reputation system policies.

To summarize, by incorporating a comprehensive perspective that accounts for transaction volume and match quality, platforms can make better-informed decisions and design more effective reputation systems that genuinely benefit users and improve overall market outcomes.

This blog is based on Andrey’s research published in Management Science, which is included in the Platform Papers references dashboard:

Fradkin, A., & Holtz, D. (2023). Do incentives to review help the market? Evidence from a field experiment on Airbnb. Marketing Science.

Does corralling a dominant platform help complementors?

This blog is written by Sruthi Thatchenkery and Riitta Katila.

Digital markets are increasingly dominated by a small number of technology platforms, prompting fears that platform owners may abuse that market power. Of particular concern is the threat that dominant platforms pose to complementor firms that depend on the platform for access to customers. For example, Google has been fined by the EU for privileging its own review and shopping comparison results over competitors on its search engine. Similar concerns have been expressed regarding Amazon promoting its own products over third party sellers in its online retail business or Apple leveraging excessive fees on developers through its App Store.

Traditionally, antitrust regulators focused on preventing firms from achieving a dominant market share in the first place, outlawing actions aimed at monopolizing markets and blocking mergers and acquisitions between firms with already high market share. But outright preventing platforms from growing to dominate their markets may be difficult, since platform markets are characterized by powerful network effects that create significant returns to scale and often yield winner-take-all outcomes. Breaking up platforms that become “too big” therefore may not be viable nor even beneficial to consumers.

Perhaps, then, it is more feasible to restrict a platform’s ability to exploit the market power that comes with massive scale. Regulators around the world are attempting to do so via newly introduced antitrust legislation, such as the Digital Markets Act in the European Union and the American Innovation and Choice Online Act in the United States. In a break from typical antitrust tactics, these new regulations do not seek to stop platforms from accumulating dominant market shares; instead, they enact rules that are meant to prevent large platforms from using that dominance to block fair competition and, in turn, innovation. But will constraining the behavior of an already-dominant platform actually restore competition and spur innovation in platform ecosystems?

In a recent paper, we attempt to shed light on that question by examining how a landmark antitrust intervention against a dominant tech platform impacted complementor firms. Specifically, we analyzed United States v. Microsoft, the first major antitrust case against a digital platform in the US.

The allegations against Microsoft were starkly reminiscent of what firms like Google, Apple, and Amazon stand accused of today: Microsoft purportedly used its dominance in operating systems (i.e. platforms) to block application developers’ (i.e. complementors’) visibility and access to customers, limiting innovation and customer choice. In 2001, the U.S. Department of Justice and Microsoft resolved the case via settlement, in which Microsoft agreed to stop engaging in activities that had been deemed anti-competitive, such as blocking customers from uninstalling its own complementor applications or punishing OEMs that tried to promote third-party complementor products.

While popular attention focused on “the browser wars” between Microsoft and Netscape, the allegations against Microsoft extended broadly across both consumer and enterprise software. We specifically examine the effects of the settlement in the enterprise infrastructure software industry. Infrastructure software consists of back-end applications that allow organizations to manage and maintain complex IT systems. The industry is divided into five complementor product markets – application development, application integration, database tools, network and system management, and security – which run on central platforms known as server operating systems. At the time of the intervention, Microsoft owned the dominant platform in the industry (Windows Server) and competed in all five complementor markets. However, Microsoft only achieved significant market share in two complementor markets: database tools and application development.

From a research design perspective, this variation in Microsoft’s complementor market share allows us to distinguish between markets that are likely to be sensitive to the intervention from markets that are not. In the “treated” complementor markets where Microsoft held significant share, the intervention suddenly constrained a powerful competitor, opening up strategic possibilities that may not have seemed viable before. In contrast, in the “control” markets where Microsoft held only minimal share, the intervention would not significantly change the competitive environment. Examining post-intervention differences between complementors in the treated versus control markets can thus yield insight into the impact of the intervention.

What we find is that the Microsoft antitrust intervention was a mixed bag for complementors: innovation went up, but profits went down. On the positive side, complementor innovation (measured as patents or R&D expenditures) increased in treated markets, particularly among those with the lowest market share. So, the intervention did in fact spur innovation among the complementors that would have been most vulnerable to Microsoft’s dominance. However, profitability dropped in treated markets – again, particularly among the complementors with low market shares. High market share complementors (who arguably did not require the same protection from regulators) did not significantly increase innovation nor did they suffer a profit penalty. Thus, while the intervention successfully prompted technical innovation, the firms that innovated did not actually benefit financially.

Why couldn’t complementors translate innovation into profits? Our interviews with industry experts indicate that complementors may have been more reliant on Microsoft than they had realized. Platforms like Microsoft often provide essential assets such as development tools to complementors in order to foster innovation on the platform. After being restrained by antitrust intervention, a dominant platform may not be as willing or able to freely share those assets, forcing complementors to incur the costs of developing replacements themselves. Indeed, several such assets were withdrawn by Microsoft after the settlement. Complementors – especially low-share complementors – were thus able to innovate but lacked the assets to bring those innovations to market in a profitable way.

Complementors and platform owners should therefore be mindful of complementor dependence on platforms. Complementors should not assume that antitrust action against platforms will be solely beneficial to them and should instead seek to develop resources and capabilities that allow them to profitably capitalize on innovation opportunities that might emerge. Platforms, meanwhile, should not encourage excessive dependence among complementors, as the long-term viability of a platform ecosystem depends on a fostering a variety of high-quality complements – a task that requires a healthy population of complementors innovating on the platform.

For regulators, our findings do offer some encouraging evidence that conduct remedies (i.e. interventions that restrict the behavior of dominant firms) can be effective in sparking technical innovation in platform ecosystems. But, innovation without profit is unlikely to foster healthy competition in the long term, particularly if profit penalties are disproportionately suffered by smaller, more vulnerable firms. Antitrust intervention that places shackles on dominant platform is therefore not an automatic cure for competition concerns. Instead, taking targeted steps to reduce complementor dependence on platforms for essential assets (or ensure those assets remain available after intervention) could be a potential win-win for both platforms and regulators: it could reassure regulators wary of abuse of platform market power while also benefitting the platform by invigorating profitable innovation across its ecosystem.

This blog is based on Sruthi and Riitta’s research published in the Strategic Managment Journal, which is included in the Platform Papers references dashboard:

Thatchenkery, S., & Katila, R. (2023). Innovation and profitability following antitrust intervention against a dominant platform: The wild, wild west? Strategic Management Journal, 44(4), 943-976.

Platforms can promote early adoption by shaping market expectations

This blog is written by Kevin Boudreau.

Online platforms have become an integral part of society and are used for a wide range of activities. To continue expanding and innovating in this area, successful launches of new platform ventures are crucial. However, most platform ventures fail to take off due to the chicken-and-egg problem – potential users are hesitant to adopt platforms that do not have a large installed base and network effects. This study empirically tested one way that theorists have long conjectured was a way to induce adoption: by marketing in a way that seeks to shape would-be adopters expectations of future platform success. The logic here is that if (at least some) consumers believe the platform will eventually take off, they will join, thereby catalyzing network effects. A field experiment was conducted to test the theory that the initial growth of a platform depends on the market’s subjective expectations of the size of the future installed base.

Chicken-and-Egg–and the Problem of Multiple Market Equilibria
(And, why is marketing a far more strategic challenge in platform industries?)

Before explaining the role of expectations, it is useful to precisely understand the nature of the chicken-and-egg problem that new platforms must overcome. Better understanding this question clarifies why the job of marketing and promoting a platform involves far more than just the traditional marketer’s job of creating awareness and extolling the benefits of a new offer.

Let’s use a simple example to explain this idea. Imagine a market with only two potential buyers, A and B, who are considering adopting a new platform. The idea extends to cases of larger markets, but let’s just consider two possible buyers to explain the crux of the idea.

For simplicity, consider the costs of adopting the platform is $1. This might include whatever purchase or access charges to get onto the platform, the hassle of learning how to use a new platform, and whatever other financial or non-financial costs.

If there are network effects, the benefits of adopting the platform depend on how many people also adopt it. For simplicity, let us suppose that if everyone adopts the platform, both A and B will enjoy benefits worth $2 (and the net benefit is $2, minus the cost of $1). If no one adopts the platform, there will be zero benefits (and the net benefit of adopting is zero, less the cost of $1). Thus, this simple set up illustrates a simple illustrate of payoffs from adopting a platform where value comes from network benefits (rather than “stand-alone” benefits, which do not depend on whether others join).

We can represent the possible outcomes in a table. The rows represent A’s decision (to adopt or not), and the columns represent B’s decision. The numbers in each cell represent the payoffs (or benefits) that A and B receive, respectively. This game payoff matrix shows that there are two possible stable outcomes, or equilibria in this market. In the “everybody adopts” equilibrium in the top right quadrant, both A and B adopt the platform; in the “nobody adopts” equilibrium in the bottom left quadrant, no-one adopts.

Therefore, the challenge for executives, marketers, and strategists in platform industries is that moving from non-adoption to adoption is more than just “marketing” in the traditional sense of building awareness and convincing individuals of the value of an offer. Here, successfully marketing and promoting adoption of the platform means coordinating market actors—as a group—out of the stable and self-reinforcing “nobody adopts” equilibrium to establish an altogether new equilibrium of “everybody adopts.” But, of course, no one wants to adopt the platform here and find out they are the only one.

While this abstract example is in a market of just two people, the same idea applies to real markets for platforms, with many more potential adopters or even multiple types of adopters, where some minimum critical mass must be coordinated onto a platform to successfully built a network and network value.

Coordinating the Market to Adopt – by Influencing Subjective Expectations

Coordinating a large number of adopters onto a new platform is typically extremely difficult and, in most cases, fails. One theorized strategy for promoting adoption that has been hypothesized by economists as a means of solving the chicken-and-egg adoption problem is to somehow influence consumers’ expectations that others will eventually adopt the platform.

Can this strategy work? How? The multiplicity of possible equilibria creates fundamental uncertainty about what others will do and which outcome will eventually emerge. Accordingly, adopters cannot “look forward and reason back” as in usual rational expectations, and expectations about future adoption are necessarily subjective. Therefore, conventional means of influencing rational expectations, such as economic signaling and pre-commitments, should not work as usual.

So, if expectations can be formed and influenced, how can this possibly work? If it simply isn’t possible for consumers to objectively form a rational expectations that the platform will eventually be widely adopted – can they at least form subjective expectations that a platform will be adopted? Moreover, is it even possible for executives and marketers of a platform to influence expectations themselves, given consumers should see platform executives and managers as biased storytellers about the future of their own platform?

A Field Experiment

It is exceedingly difficult to design research studies to systematically empirically study platforms. This is especially so in attempting to explicitly test how nuanced growth dynamics might be influenced by something unobservable as subjective expectations. The question of empirically testing expectations in platform industries was, in fact, an idea I had even back in graduate school at MIT, when I was thinking about the strategies I saw platform executives pursue. It took me some years to imagine a research design that would be appropriate for testing these questions, and more time to then implement the research in a live platform.

A research design was devised and implemented with the goal of investigating how statements issued from a platform about its own expectations of future size of its installed base of users affected the likelihood of individual’s adopting the platform. The experiment was conducted by randomly varying messages to 16,349 individuals identified as potential adopters of a newly launched product development platform focused on the Internet of Things (IoT). The experiment varied the content of the invitation, which included a statement regarding the number of users and companies that the platform expected would join. The invitation also included a statement about the current installed base, to disambiguate effects of statements regarding expectations from the true installed base. A central challenge of this research design was to create random variation in both statements of expected installed base and current installed base across individuals, while avoiding any deception whatsoever. (Details are explained within the paper.) The research was carried out as part of the first 60 business days following the launch.

Findings

The experiment found evidence consistent with adoption patterns being significantly affected by statements made by the platform, and overcoming the initial “chicken-and-egg” problem. Remarkably, the subjective statements regarding expectations of the future installed base had a larger influence on adoption rates than did disclosures of the true current installed base—at least this was true during early adoption. For example, statements of larger numbers of expected users caused more adoption than smaller numbers. Further, stating a small (current or expected) installed base of users led to lower demand than stating nothing at all. Statements of expectations had no effect on adoption once the installed base grew large.

Implications for Promotion, Marketing, and “Evangelism” of New Platform Ventures

The findings have important implications for the promotion, marketing, and “evangelism” of new platform ventures. The results suggest that the promotion of a new platform should focus on the potential size of the future installed base rather than the current size. Small current or expected installed bases of users should not be disclosed, as it can lead to lower demand than stating nothing at all. Instead, the focus should be on generating expectations of a large future installed base. This is particularly relevant for new platforms, which face the chicken-and-egg problem, as the expectations of the market can lead to a self-fulfilling prophecy.

The demonstration that adoption patterns can be significantly affected in a context of fundamental entrepreneurial uncertainty raises important questions. Better understanding of how and why platform entrepreneurs are able to persuade and influence market participants could bring fundamental advances to the ability to market and promote the growth of platforms. In the paper, I discuss alternative plausible explanations of why statements by the platform could have had an influence. The explanation that is most consistent with the facts is that (some) boundedly rational adopters respond to statements by the platform simply at face value, influenced by some plausible characterization or storyline of the future. In such an explanation, the adopters might even believe themselves to be prescient, as their expectations were self-fulfilling. (In this case, the eventual adoption of thousands and thousands of users in continents around the world turned out to be consistent with claimed expectations.)

If this explanation is true, adopters might not require a fully rational and credible basis for accepting statements and we might imagine some considerable role to be played by rhetoric, charisma, moral suasion, and other tools of influence. Certainly the world of platforms and tech have had persuasive storytellers and persuaders, including the likes of Steve Jobs. The last time I checked there were 2,330 “Chief Evangelists” listed on LinkedIn.

This blog is based on Kevin’s research published in Management Science, which is included in the Platform Papers references dashboard:

Boudreau, K. J. (2021). Promoting Platform Takeoff and Self-Fulfilling Expectations: Field Experimental Evidence. Management Science, 67(9), 5953-5967.

Strong platform control can encourage rather than depress complementors’ innovation activity

This blog is written by Yuchen Zhang, Jingjing Li, and Tony Tong

Digital platforms are drawing increasing attention from innovators and entrepreneurs, whether it is about creating a new platform, or providing product offerings and exchanging new ideas on an existing platform. Apple’s App Store, which hosts millions of mobile apps published by developers and used by billions of users, is a great example of a platform that harnesses the power of distributed innovation agency. Indeed, platforms like the App Store are now being seen as a “semi-regulated” marketplace or unique organization that foster innovation and entrepreneurship under the leadership of the platform owner (Apple). Yet, such leadership is surely easier said than done. After all, platform owners (Apple) do not own the product offerings (apps), nor do they have direct control over their business partners—complementors (app developers).

So what can a platform owner do then to coordinate and orchestrate the value creation activities of autonomous complementors (often in thousands or millions) that are critical to the vibrancy and success of a platform?

This question was the focus of a research paper we recently published in Strategic Management Journal. We invoke the idea of “platform governance”, or the “set of rules and policies that platform owners adopt in order to coordinate and deploy co-specialized assets and capabilities of complementors.” While platform governance manifests itself in different mechanisms, in this paper, we argue and show that platform owners can practice what we call “access control” to shape complementors’ innovative activities. Rather than looking at complementors’ economic exchange activities on a platform (e.g., listing and selling of mobile apps), we examine their exchange of knowledge (about developing apps) with each other. The “free” sharing of problem-solving techniques, code writing tips, and product development experiences, is critical to software development and has even been considered by some as underpinning the recent open source revolution.

We look at how the lack of access control in the form of “jailbreaking” shapes app developers’ knowledge sharing activities. Specifically, Apple’s iOS platform is well-known for adopting a strict gatekeeping policy that controls for what (apps) or who (app develops) have access to the platform; this policy denies platform access to imitating apps and developers in order to encourage app innovation. The jailbreak of the iOS—which was more common in the early years of iOS development—is a hacking that exploits loopholes to remove Apple’s built-in restrictions, allowing users to install apps not officially approved by Apple’s App Store. Jailbreaking would therefore introduce pirated apps that imitate or directly copy the features and functionalities of the existing apps approved by the App Store and trigger so-called business stealing and user attrition, putting a competitive pressure on those iOS developers who develop “legit” apps and seek to profit from app sales.

As illustrated in the figure below, we leverage the unexpected timing of the jailbreak of iOS 7 in 2013 as a natural experiment, to compare the Q&A activities of iOS app developers (treated group) and otherwise comparable Android app developers unaffected by the jailbreak (control group), on StackOverflow.com, an active online forum of software developers. The jailbreak significantly weakens (for a short period) Apple’s strict access control, such that imitating app developers (and apps) can now gain access to the App Store and users of iOS devices. We find that the jailbreak reduces the amount and quality of the knowledge shared by iOS app developers—their posting of Q&As decreases by 7%, and the quality of their posts by 15%. Such effects can be seen as causal given the research design.

The lapse in access control therefore dampens developers’ incentives to share knowledge (and innovate). In other words, strong access control, at least in the case of Apple’s iOS, will appear beneficial to the platform ecosystem and platform owner, because it enables more vibrant knowledge sharing among app developers, which should enhance their innovation and product development effort.

In practice, a platform owner can resort to other tools or design features to implement access control. These include charging a membership fee or implementing an involved process to weed out low-quality complementors (e.g., eHarmony charging registered users a non-trivial fee), designing certain screening processes to determine which complementors will be allowed platform access (e.g., Uber requiring drivers and their cars to pass certain criteria), and selectively restricting or opening up the use of boundary resources (e.g., APIs, SDKs, code libraries) to particular complementors. However, it’s important to note that none of these is meant to imply that stronger control is always “better.” In fact, the degree to which platform owners should grant access to complementors will depend on many factors, including the specific business model or ways in which platform owners profit from innovation. For instance, Apple makes a lot of profits from the commissions charged on app sales through its App Store, making it important that it implement strong access control to encourage the creation of high-quality apps. By contrast, for Google’s Android, most profits are coming from advertisement and related services; lesser access control does not affect it as much.

What other governance mechanisms, in addition to access control, may be at the platform owner’s disposal to coordinate and orchestrate complementors’ value creating activities? In a review article published in the Journal of Management, coauthored with Liang Chen, Shaoqin Tang, and Nianchen Han, we examine this question by reviewing existing knowledge on digital platforms written by scholars in management, MIS, marketing, operations, economics, and other related fields. This is done based on our view that although the leadership role of platform owners is well recognized, a coherent approach to understanding how they govern their relationship with complementors to create and appropriate value is still lacking. We use two core dimensions of organizational governance—incentive and control—to synthesize platform governance research into eight mechanisms, and we map them to a multitude of seemly idiosyncratic platform design tools and features, summarized in the figure below.

To conclude, platform governance matters a great deal. Our study shows that one specific form of platform governance, access control, matters to complementors’ innovation activity by encouraging cooperative behaviors in knowledge sharing. Before we herald openness as the holy grail of innovation, it is time to also celebrate the important role of platform control in motivating complementary innovation.

This blog is based on Yuchen, Jingjing, and Tony’s research published in the Strategic Management Journal, which is included in the Platform Papers references dashboard:

Zhang, Y., Li, J., & Tong, T. W. (2022). Platform governance matters: How platform gatekeeping affects knowledge sharing among complementors. Strategic Management Journal, 43(3), 599-626.