Search

The most commonly cited issues with mobile apps are low user engagement and retention levels. In this paper, we use a randomized control trial to test the efficacy of crowdsourcing on enhancing user engagement and retention in the context of mobile gaming apps. We examine two specific crowdsourcing features: the ability to submit content and the ability to access crowdsourced content. We find that the content submission significantly increases engagement and retention by reducing users’ hazard of ending a session by approximately 11% relative to the baseline and reducing the hazard of abandoning the game app by 14%. In contrast, giving users the ability to access crowdsourced content has no significant effect on engagement but has a positive effect on retention by reducing the hazard of abandoning the game app by 13%. Surprisingly, we find that the interaction effect of these two crowdsourced features is negative on user engagement. Individually, the submission feature manifests itself via empowering users to control their product use experience, whereas the access feature’s positive effect on retention is mediated by diversity and novelty of content. However, the two effects are not complementary. It turns out when these two features are given together, the empowerment enabled by one’s own submission is crowded out by others’ submissions, and this dominates the diversity benefit. Crowdsourcing features have heterogeneous impact on different user segments, with heavy users and users of longer tenure being more affected by the crowdsourcing features. This paper was accepted by Anandhi Bharadwaj, information systems.

With one-third of marriages in the United States beginning online, online dating platforms have become important curators of the modern social fabric. Prior work on online dating has elicited two critical frictions in the heterosexual dating market. Women, governed by age-old social norms of not making the first move, are inhibited in their interactions in that they do not initiate contact with men. On the other side, men send an abundance of messages, the majority of which do not convert to matches. A key distinguishing feature of online dating versus its traditional counterpart is the ability to leave a range of digital signals not replicable in the offline world. These digital signals can impact the nature of online dating platform outcomes. In this paper, we study the impact of a feature that reveals “who likes you” (WLY) on engagement, the number of matches, match efficiency, and match sorting in online dating. This feature reveals the identity of the voters who have rated the focal user with a like. To causally identify the effect of this feature, we conduct a large-scale randomized control trial in collaboration with a major North American dating platform. The treatment causes women to be more proactive, sending 7.4% more messages, which is a highly desirable market improvement given that men send double the number of messages compared with women. Further, we find that the women endowed with this feature increase their matches by 14.4%, whereas men increase their matches by 11.5%. Analyzing the moderating impact of desirability—a key aspect of the WLY feature—provides us with nuanced findings. Depending on the levels of each of the two parties’ desirability, we see evidence of sorting, encouragement, and discouragement. This paper was accepted by Anandhi Bharadwaj, information systems.

The growing popularity of online dating websites is altering one of the most fundamental human activities: finding a date or a marriage partner. Online dating platforms offer new capabilities, such as extensive search, big data–based mate recommendations, and varying levels of anonymity, whose parallels do not exist in the physical world. Yet little is known about the causal effects of these new features. In this study we examine the impact of a particular anonymity feature, which is unique to online environments, on matching outcomes. This feature allows users to browse profiles of other users anonymously, by being able to check out a potential mate’s profile while not leaving any visible online record of the visit. Although this feature may decrease search costs and allow users to search without inhibition, it also eliminates “weak signals” of interest for their potential mates that may play an important role in establishing successful communication. We run a randomized field experiment on a major North American online dating website, where 50,000 of 100,000 randomly selected new users are gifted the ability to anonymously view profiles of other users. Compared with the control group, the users treated with anonymity become disinhibited, in that they view more profiles and are more likely to view same-sex and interracial mates. However, based on our analysis, we demonstrate causally that weak signaling is a key mechanism in achieving higher levels of matching outcomes. Anonymous users, who lose the ability to leave a weak signal, end up having fewer matches compared with their nonanonymous counterparts. This effect of anonymity is particularly strong for women, who tend not to make the first move and instead rely on the counterparty to initiate the communication. Further, the reduction in quantity of matches by anonymous users is not compensated by a corresponding increase in quality of matches. This paper was accepted by Lorin Hitt, information systems.

Demonstrating compelling causal evidence of the existence and strength of peer-to-peer influence has become the holy grail of modern research in online social networks. In these networks, it has been consistently demonstrated that user characteristics and behavior tend to cluster both in space and in time. There are multiple well-known rival mechanisms that compete to be the explanation for this observed clustering. These range from peer influence to homophily to other unobservable external stimuli. These multiple mechanisms lead to similar observational data, yet have vastly different policy implications. In this paper, we present a novel randomized experiment that tests the existence of causal peer influence in the general population—one that did not involve subject recruitment for experimentation—of a particular large-scale online social network. We utilize a unique social feature to exogenously induce adoption of a paid service among a group of randomly selected users, and in the process develop a clean exogenous randomization of treatment and control groups. A variety of nonparametric, semiparametric, and parametric approaches, ranging from resampling-based inference to ego-level random effects to logistic regression to survival models, yield close to identical, statistically and economically significant estimates of peer influence in the general population of a freemium social network. Our estimates show that peer influence causes more than a 60% increase in odds of buying the service due to the influence coming from an adopting friend. In addition, we find that users with a smaller number of friends experience stronger relative increase in the adoption likelihood due to influence from their peers as compared to the users with a larger number of friends. Our nonparametric resampling procedure-based estimates are helpful in situations of networked data that violate independence assumptions. We establish that peer influence is a powerful force in getting users from free to premium levels, a known challenge in freemium communities. This paper was accepted by Sandra Slaughter, information systems.

The rapid pace of technological innovation necessitates that information technology (IT) services firms continually invest in replenishing the skills of their key asset base, the human capital. We examine whether human capital investments directed toward employee training are effective in improving employee performance. Our rich employee level panel data set affords us the opportunity to link formal training with performance at the individual employee level. Using a dynamic panel model, we identify a significant positive impact of training on employee performance. A unit increase in training is linked to a 2.14% increase in an employee's performance. Interestingly, we find that in the IT sector, skills atrophy and consequently high-experience employees reap higher returns from training, which highlights the uniquely dynamic nature of IT knowledge and skills. We also find that general training that an employee can utilize outside the focal firm improves employee performance. However, specific training pertinent to the focal firm is not positively linked to performance. On the other hand, although domain and technical training both enhance employee performance individually, the interaction between the two suggests a substitutive relationship. Thus, our findings suggest that the value of training is conditional on a focused curricular approach that emphasizes a structured competency development program. Our findings have both theoretical and practical significance. Most important, they justify increased human capital investments to fuel future growth in this important component of the global economy. This paper was accepted by Lorin Hitt, information systems.

We study settings where a number of sellers simultaneously offer vertically differentiated Vickrey auctions for imperfect substitute goods to unit-demand buyers. Vertical differentiation can arise from differences in item quality, item value certainty, seller reliability, or a combination of these factors. We characterize the form of the bidding equilibria and derive expressions for the corresponding allocative efficiency and expected seller revenue. When bidders are restricted to submit at most one bid, our theory predicts the existence of a unique Bayes-Nash equilibrium that resembles a form of probabilistic “mating-of-likes.” Allowing unit-demand bidders to place an arbitrary number of bids induces complex strategy profiles where bidders place positive bids in all available auctions. Higher bidder types tend to follow more targeted strategies, focusing their “serious” bids on fewer and, generally, higher quality auctions. The nature of the bidding equilibria introduces allocative inefficiencies that arise from the lack of coordination in auction selection among bidders. We test our theoretical propositions in a controlled laboratory experiment while also utilizing a domain specific risk score to help assess how the bidders' risk type affects their bidding behavior. In support of our theory we find evidence of a probabilistic assortative matching between bidder and auction types. We also find that low risk type bidders tend to crowd on the highest auction and will pay a premium for the certainty it offers, whereas high risk type bidders fail to appropriately adjust for risk associated with the lowest auction, leading to overbidding. These lead to an interesting focal anomaly whereby bids are concentrated on the highest and lowest auctions, bypassing intermediate auctions.

We explore the problem of pricing and allocation of unique, one-time digital products in the form of data streams. We look at the short-term problem where the firm has a capacitated shared resource and multiple products or service levels. We formulate the allocatively efficient Generalized Vickrey Auction (GVA) for our setting and point out the computational challenges in determining the individual discriminatory transfer payments. We propose an alternative uniform-price, computationally efficient, revenue-maximizing knapsack formulation called the Multiple Vickrey Auction (MVA). While not incentive compatible, the MVA mechanism achieves bounded posterior regret and can be solved in real time. It has the added benefit of realizing imputed commodity prices for the various services, a feature lacking in the discriminatory GVA approach. For service providers that are concerned about the incentive compatibility but want imputed service prices, we suggest a maximal MVA (mMVA) uniform-pricing scheme that trades off revenue maximization for allocative efficiency. For sake of completeness we discuss the properties of a first-price pay-your-bid scheme. While NP-hard and not incentive compatible, this formulation has the perceived benefit of cognitive simplicity on the parts of sellers and bidders.

Business-to-consumer online auctions form an important element in the portfolio of mercantile processes that facilitate electronic commerce activity. Much of traditional auction theory has focused on analyzing single-item auctions in isolation from the market context in which they take place. We demonstrate the weakness of such approaches in online settings where a majority of auctions are multiunit in nature. Rather than pursuing a classical approach and assuming knowledge of the distribution of consumers' valuations, we emphasize the largely ignored discrete and sequential nature of such auctions. We derive a general expression that characterizes the multiple equilibria that can arise in such auctions and segregate these into desirable and undesirable categories. Our analytical and empirical results, obtained by tracking real-world online auctions, indicate that bid increment is an important factor amongst the control factors that online auctioneers can manipulate and control. We show that consumer bidding strategies in such auctions are not uniform and that the level of bid increment chosen influences them. With a motive of providing concrete strategic directions to online auctioneers, we derive an absolute upper bound for the bid increment. Based on the theoretical upper bound we propose a heuristic decision rule for setting the bid increment. Empirical evidence lends support to the hypothesis that setting a bid increment higher than that suggested by the heuristic decision rule has a negative impact on the auctioneer's revenue.