Pricing on-demand services: Alternative ways of combining usage and access fees

Abstract Many firms are adopting an “everything or anything” as a service—“XaaS”—model to sell goods and value-added services. Buyers for such goods are heterogeneous both in the number of units they desire and the valuations for each unit. Suppliers designing such revenue models can offer either a usage-based per-unit fee or an access-based per-period fee, often combining them both as a nonlinear two-part tariff plan that imposes both types of fees on all buyers or let buyers self-select from a menu of per-unit and per-period plans. We develop a theoretical model to analyze the economic implications of alternative designs for mixing per-unit and per-period fees. Our analysis produces the following practical insights on these plans' profitability and market coverage. First, the menu is generally a better way of combining access and usage fees for a firm selling digital goods with zero variable costs. Second, the preferred design switches to a two-part tariff if the firm's production environment resells or runs on top of a back-end infrastructure or data service provider that imposes nonnegligible variable costs. Third, we show that the revenue advantage of these two designs (which employ both fees) over a simple plan that employs only one fee (best of per unit and per period) is most significant when the rate of change in marginal valuation is relatively similar across buyers.