Most of the service innovation in e‐government initiatives has been initially focused on making a digital version of an equivalent physical service, followed by creating new services through either adopting new business models or overhauling existing infrastructures. The former approach uses variants of e‐commerce business models to drive service innovation whereas the latter entails changes to the backend for driving system efficiency and integration, which later serve for reorganising and improving service provision. Each approach presents a different way of assembling services, infrastructures, people and business models. The ramifications of each assemblage to service innovation are less understood. This paper seeks to examine the intended and unintended consequences of these two contrasting approaches inductively and deductively. Our findings show that a frontend approach enables quick improvements and enacts a variety of structures, with each structure inscribing a set of new path‐dependent routines in the infrastructures. In the backend approach it took longer to introduce new services. Yet the services were of a higher quality and required less modification in the long run. Quantitative analyses confirmed these findings. We attributed the alignment advantage of backend approach to better interdepartmental collaboration organised around a technology platform for system and service integration, concentrating resources on a coherent rather than a diverse set of business models. This approach realigns practices and routines and technology internally, whereas the explorative, diverse use of business models in the frontend approach for service innovation is less sustainable.