In brief, typical consulting does a very good job of analyzing the situation through a "deep dive problem solving" exercise. But its prescriptions suffer from a linearity bias, in so far as it does not pay adequate attention to the "general equilibrium" effects of stakeholder interaction dynamics (most often, the behavioral changes), which is often considerable in development contexts.
A very good example of this comes from the analysis of energy savings opportunities. A 2009 study by McKinsey & Co showed that the US could save $680 bn over 10 years by improvements to efficiency of its homes, offices, and factories, through strategies like sealing leaky building ducts and upgrading old appliances. But as Brad Plumer writes,
As economists scrutinized those numbers, they realized the picture is more complex. Those engineering studies can’t account for the behavioral changes you might see in response to efficiency improvements... People could, for instance, start adjusting their thermostat if it becomes cheaper to cool the house... One recent study of Mexico, for instance, found that a government program to help people to upgrade their refrigerators with energy-saving models really did curtail electricity use. However, a similar program for air conditioners had the opposite effect — when people got sleeker A/C units, they used them more often, and energy use went up.Similar unanticipated or unpredictable behavioral responses are commonplace with most large social programs. For example, efforts to improve learning outcomes by assessing outcomes of standardized tests, is likely to be gamed by teachers with time. Similarly, efforts to improve performance among public officials through financial incentives has the potential to be subverted in unpredictable ways. A cash transfer program to replace a food distribution system can fail because people may use the cash for other things or the local prices of food grains may fluctuate in an unpredictable manner or something else, the possibilities of which cannot be incorporated in a context analysis based one-off program design. Unlike the private sector - where designing incentive compatible arrangements is much easier and disciplining mechanisms are more effective - the emergent possibilities with public systems are far too many to be fully anticipated. No amount of theoretical and logical reasoning can anticipate all the emergent possibilities, which arise from cognitively constrained or skewed stakeholders.
In simple terms, these impacts are not likely to be assessed with regular problem-solving tools. They require iterative field experiments that can observe outcomes in real-time and try to respond to emergent scenarios using short and tight feedback loops. But they are both expensive and take time. They cannot be part of a "hourly billing" based, "high-intensity" consulting model. Thus the need for a collaborative approach to solving development problems.