The purpose of this paper is to resolve a question regarding efficiency of a lightning protection system (LPS) for buildings based on the collection volume method (CVM) . The paper has two components. The first, following suggestions of other authors [Abidin and Ibrahim 2004], takes advantage of count data from installed devices, and independent installation-site inspections to develop our statistical analysis. The second component investigates the validity of the underlying theory by introducing a novel methodology of fractional Poisson processes, which are able to reproduce the burstiness of lightning strikes, an essential feature of stochastic time dependence of incidence of lightning strikes. The standard Poisson processes used in the past efforts in this area cannot do that.