Large infrastructure investments are expected to be of sustained value to society for a long time. Such investment projects include, for instance, hospitals, tunnels, sport arenas, power plants, roads, railways, and bridges. They involve a complex organization of contracts and agreements. The client is expected to plan, procure, and determine the critical steps of a project, while the contractor should solve issues raised by the client. Many of these agreements are path-dependent and reflect past routines, experiences, and contacts. As such, many investments tend to return to similar sources instead of replacing routines and collaborations that did not work. This can cause change orders that furthermore reflect consequences such as cost and time overruns. While much is known about these effects in construction projects, this paper sheds light on the drivers of change orders. We build upon a sample of 234 observations responding to a survey on investment planning. The results show that project assumptions are often wrong and inadequate in large investments. Such wrong assumptions are caused by interpersonal and leadership issues, poor planning, or sometimes even intentional profit-seeking. Our results show that clients and contractors have different perceptions and enter contractual obligations differently. The implication is, therefore, that better routines of documentation, more frequent feedback, and more accurate or precise standards may close the gap between planning and what is actually achieved. More precise contractual agreements may also create a better process to procure, manage projects, and allocate resources.