Research projects often start with a grower telling a researcher that their biocontrol program ‘did not work’. Ideally, the researcher will then ‘reverse engineer’ the IPM program and the crop production system, in order to find the part (or parts) that cause the problem. The challenge is that every IPM program is unique and has to be adapted to the crop, location, production practices and many other factors. Also, most of the time, individual research projects only look at a small part of a production system, making it hard to predict the performance of the proposed solution in the real world. Finally, it is often difficult to do scientifically sound on-farm validation trials, due to the lack of control treatments or variable natural pest infestations. This presentation will discuss different factors that influence IPM programs, including biocontrol agent quality, release methods, compatibility, host plant, production practices, lack of commercially available natural enemies and external interference. This will be illustrated by past and current projects on aphid control in greenhouse crops.