Sometimes the project manager places too high an expectation on the amount of foresight and vision that clients have. Some project managers figure that the client is requesting the work, so they must have a very good idea of what needs to be done. They go to the client looking for complete answers and find instead partial, vague, and sometimes conflicting information.
In fact, this situation occurs fairly frequently and it does not mean that the client is stupid. In many cases, especially for large projects, the client has a vision of what the results will be, but cannot yet articulate this vision into concrete objectives and deliverables. They may also not know enough to help the project manager define scope, risks, project organization, etc.
It would be a bad idea to actually start a project with a high degree of uncertainty about what it is actually trying to do. Instead, once you understand what you know and what you don't know, you need to work with your client to determine how best to fill in the remainder of the puzzle. There are a number of reasons why you might be having trouble getting the project properly defined. Once you understand the cause, you can put in place a remedy to move the planning process to a successful conclusion.
In some cases, the project details are probably available, but you're having a hard time getting the major stakeholders to focus because of other commitments. In this case, you will need to escalate the problem to the sponsor. The sponsor needs to help you by requiring the major stakeholders to allocate the time needed.
In some cases, the major stakeholders may be giving you the time you need, but the information you're receiving is inconsistent or conflicting. There are two ways to move forward. The first is to see if the sponsor will make the final decisions using the stakeholder feedback as input. The second is to set up a facilitated session with all of the major stakeholders. If you can get the key people in a room together, they should be able to come to a consensus on project objectives, scope, risks, etc.
In some cases, it's not a matter of focus or consensus, but simply a lack of knowledge. For instance, the sponsor may have a vision for what needs to be done, but the entire initiative may be new. You may find that neither the sponsor nor the major stakeholders understand enough about the details to provide all the guidance that you need. In this case, a good technique would be to only define an initial project to gather the detailed requirements. At the end of this initial project, the solution should be defined in enough detail that the remainder of the work can be defined and planned as a second, follow-on project.
One of the main reasons for project failures is that the project was never properly defined and planned. Don't fall into the trap of pushing toward project execution with incomplete or conflicting information. This is recipe for failure. In fact, projects that are hard to define will benefit most from the extra time invested in the project definition process.