Expected change is derived from benchmarks, as already stated. In order to be comparable, operations would need to:

· be of same or similar size – this can be relaxed by using relative rather than absolute figures; some impact of scale remains because it also impacts the relative improvement potential through scale effects per se,

· share a similar network cost structure – this is satisfied by excluding extremely untypical data points from the benchmarking database,

· have about the same degree of maturity – this is reflected in our model and database by qualifying the operation under investigation as above-average, average, or below-average, compared to the benchmark average; this means that for every combination of delivery option, network domain, and process activity, our database has to provide three different benchmark figures depending on the operation’s current degree of efficiency,

· present the same baseline and future scenarios – given the uniqueness of each operation, it would be impossible to find a sufficiently large sample of benchmark projects with same baseline and same target scenario as the operation under investigation; also, the effort of maintaining such a fragmented database would be enormous; therefore, we apply a concept of normalization and use existing benchmarks departing from a fully neutral baseline (i.e. pure local delivery) to any modified scenario to subsequently adjust other benchmarks departing from a baseline corresponding to that modified scenario to an even different target case – thus, only a limited number of relationships need to be computed, not all possible combinations,

· be synchronized in terms of what year a measure is being taken – this is reflected in our model and database by providing a ramp-up function of the cost reduction benchmarks, which is different for each delivery scenario; broadly-scoped measures promise greater future savings, but less immediate effects due to the higher complexity of such solutions, which requires more time to implement (e.g. central vs. regional delivery, multi-vendor managed services vs. outsourcing),