Stefan Wagner is an associate professor of strategy and innovation and director of PhD studies at ESMT Berlin. His research interests cover the intersection of firm strategy, technological innovation, industrial organization, and law.
Price discrimination – charging different customers different prices for the same goods – is common in offline markets. This is not too surprising as charging different customers different prices for the same product based on their willingness to pay allows companies to increase profits.
However, despite thebenefits in offline markets, price discrimination has not been implemented on a large scale for digital products. This is surprising as price discrimination could be implemented at comparably lower costs in the digital realm where consumer data that is prerequisite for individualized pricing is routinely collected. The big data collected could be used to determine personalised prices for each consumer based on their characteristics or search histories, for example.
To analyse how much profit digital companies forego by not charging individualised prices for different users, weanalysed different pricing strategies in the online video gaming industry. The full paper is available here and is joint work with Louis-Daniel Pape, Christian Helmers, Alessandro Iaria and Julian Runge. Specifically, we study pricing strategies employed in a highly popular mobile game similar to Candy Crush Saga. Gaming accounts for more than three quarters of the total app revenue in major app stores, including Apple’s App Store and Google’s Play Store, but while game developers in this market collect around 90% of their revenues through paying customers on freemium models, alternative pricing strategies in this industry have not yet received much attention in research.
The game used in our study consisted of levels that can be solved in a short amount of time with the completion of each level rewarded with a certain number of stars. The first 39 levels were free before reaching a pay-gate to unlock level 40, with pay-gates then present every 20 levels thereafter. Pay-gates can be unlocked with a sufficient number of stars. If players reach a pay-gate with insufficient stars they can unlock the gate by replaying levels to increase their stars, inviting a friend to play, or paying for a ‘key’ with 70 virtual coins (equivalent of $1).
For this study, we obtained detailed data tracking how around 300,000 players interact with the video game over a 15-day period. The data not only contains player characteristics but, more importantly, in-game purchase decisions of the players when it came to unlocking pay-gates. Exploiting this data, we were able to identify a structural model of player decision making factoring in player characteristics, progress in the game, and different pricing mechanisms employed by the gaming company. Such a decision model allows simulating consumer choices under different pricing regimes and thus allows to link alternative pricing strategies to firm profits.
At the time of data collection, the game developer charged the same (uniform) price to all players and was not engaging in any form of price discrimination. Simulating different pricing strategies, we found that the price initially set by the game developer was suboptimally low. The company could have charged a significantly higher price without losing too many customers. As a result, setting a uniform price at a higher and optimal level can have increased profits by 340%. We suspect that the gaming company overestimated the price responsiveness of players and thus shied away from charging higher prices. Interestingly, and contrasting conventional wisdom, the results also showed that first-degree price discrimination, i.e., charging individualized prices, would have only very limited additional effects on profitability. Compared to an optimal uniform price for all players, discriminatory pricing strategies with prices optimised for each player separately would only lead to slight in profits compared to the optimal uniform price (+3%).
In summary, our results indicate that the game developer can substantially increase profit by using basic information readily available on player characteristics and in-game behaviour. However, a simple uniform pricing strategy may already guarantee most of the profit implied by elaborate forms of price discrimination, which might explain why price discrimination has been used sparsely in online markets.
Keep in mind that attempts of price discrimination in online markets havealso been met with animosity in the past; Amazon’s early attempt to price discriminate buyers of DVDs based on purchase history was met with resistance and negative publicity. The minimal gains made through price discrimination when compared with uniform pricing might not be considered worth the potential backlash, explaining the lack of price discrimination online.