When Should Software Firms Commercialize New Products ... - MISRC

ucts. First, software products nowadays are increasingly built using a modular architecture
which, in turn, facilitates grouping, separating, or locking certain features. Second, it is
relatively easytoembedinsoftwareprogramstrial-expiration clocksassociated withfeature-
locking mechanisms. Third, software products are digital goods with negligible marginal
reproduction cost, that can be provided in unlimited supply and can be “shipped” via rela-
tively cheap online distribution channels. This way, absolutely all interested customers have
access to the free consumption opportunities embedded in the freemium offering. Fourth,
software products often belong to the category of experience goods. By trying (sampling)
the product or part of it before committing to any purchase, consumers could learn more
about the quality and other attributes (such as speed, functionality, and features) of the
software, capabilities of related modules, compatibility issues, hardware requirements, etc.
It remains largely unknown how well these freemium models fare, compared with con-
ventional and well established business models. As illustrated by the examples above,
freemium models are applied nowadays in both software-as-a-service markets as well as
markets for software products characterized by a one-time purchase and designed to run
autonomously on the consumers’ hardware. In this paper, we focus on the latter. Moti-
vated by inquiries from entrepreneurs, we advance a unified multi-period framework that
accounts for word-of-mouth effects and cross-module synergies and can capture the essence
of several software business models, including freemium and seeding ones. We use this
framework in particular to derive the equilibrium for each model and fully characterize
conditions under which freemium models are superior to conventional charge-for-everything
and seeding models. We further compare freemium models against each other. To the best
of our knowledge, this represents the first extensive analytical benchmarking study that
juxtaposes for-fee, freemium, and seeding models. Our results indicate that, while seeding
models are always dominated, the conventional for-fee model and the freemium models can
each be optimal in separate regions of the feasible space. We also address policy implica-
tions by demonstrating that switching from conventional or seeding models to freemium
models always increases social welfare. Matching welfare results with the profit analysis,
we also explore cases when the firm should be subsidized (i.e., when the interests of the firm
and the society are not well aligned).
The rest of the paper is organized as follows. §2 presents a summary of the relevant
literature. In §3, we introduce the general models. In §4, §5, and §6, we derive the optimal
4