They can be pretty difficult to change. Normally, when we digitize processes, we only touch a subset of the overall number of interrelated processes. Project constraints, such as time and resources, tend to limit our view insofar as we try to limit the scope of the development task as much as possible. However, as interrelated processes may either receive data from our process, or feed our process with data, we will inherently run the risk of breaking larger, more complex processes when we make smaller changes.
It seems to be a condition we have to deal with when running development projects. Since we cannot afford to take the entire systems/process landscape into consideration due constraints we run the risk of missing important interactions between high level processes; interactions that are pivotal to making our processes function properly in a business context.
The solution? Who knows? My suggestion would documentation. If proper documentation can be obtained, we might be able to pin point high level interactions in the planning phase, thereby mitigating the risk. It’s a pretty simple solution but for some reason it seems to be difficult to carry out in real life. BPM could actually be a really good solution to this problem. However, as one of my coworkers always says: Habet realitatem intercession.