How Freemium Gets Customers To Pay A Premium: The Role of Loss Aversion

Motivation. Over the past decade, freemium (a combination of free and premium) has become a popular business model, especially among firms that develop information goods such as computer software, online games, applications, and services. In the freemium model, a consumer receives the basic version of a product for free, but can obtain the premium version, which has added features and more functionality, by paying a certain price (Kumar 2014). This has a huge appeal among the firms in various digital sectors such as Dropbox, Evernote, Asana, and LinkedIn. This model especially dominates the digital gaming market. Rovio’s Angry Birds game became an international phenomenon within a year of release and is now the top paid app of all time, having been downloaded more than three billion times (Robertson 2015). The freemium model is also applicable when the free product is online and the premium product is a physical good (online-offline) (freemium.org 2015). An example of the online-offline variant of freemium is Windowfarms, a firm that sells vertical food-growing gardens designed for homes. Although offering the free product helps expand the product’s consumer-base, another factor is key in selling the product and increasing the firm’s profits: Once consumers use the free product, they emotionally value it more and find losing it painful. Thus, when it is time to pay (either to keep the product or to expand its current service), they often pay to avoid the pain (Psychguides 2015, Seufert 2014, Mayyasi 2013). This anomaly arises because, when purchasing a freemium product/service, consumers exhibit loss-averse preferences (Brustein 2013)—that is, when dealing with the premium product/service, the consumers feel losses more strongly than equally sized gains. For example, when consumers consider paying for Dropbox Pro (as their storage needs have gone beyond the limited space offered by the free version), they may be less likely to think about the price than all the possible ways they can continue using (or extend their use of) basic Dropbox. Despite such evidence from practice, the influence of behavioral biases—particularly of consumers’ loss-aversion—on premium purchase decisions has received little attention in the academic literature. This paper seeks to reduce this gap by answering three main research questions. Research Questions. First, how should a firm optimize its ordering and pricing decisions regarding the premium version of its product, considering the purchase behavior of loss-averse consumers with uncertain demand and valuations? Second, how do the quality levels of the free and premium versions impact the optimal price? Third, how does the optimal price vary with consumers being satisfied, slightly dissatisfied, or entirely dissatisfied with the free version? To answer these questions, we consider a profit-maximizing firm that offers two versions of a single product: a basic version for free and a premium version at a positive price. The market size for the free version is random. Individual consumers are heterogeneous in their tastes and are loss-averse, with reference points that represent their probabilistic beliefs about the products’ values and availability outcome. Given these reference points, they develop purchase plans to maximize their total expected utility, including gain-loss utility, due to their loss-averse behavior. Each consumer first uses the free version, which enables him to learn the value of both the free and premium versions and then decides whether to buy the premium version. In anticipation of consumers’ purchasing behavior, the firm maximizes its total expected profits by setting the best price for the premium version, given its availability and the quality levels of the two versions. Related Literature. There are three streams of literature related to this research. (i) The first is the Behavioral Economics literature that study consumers’ loss aversion, but unlike this paper, they do not consider any of the important operational factors such as uncertain demand, multiple products, and product quality in their models (see, e.g., Kahneman et al. (1990)). (ii) The second stream is the literature in Operations Management that has modeled consumer behavior, although unlike this paper, has mostly