I think overpromising and delivering a lesser product than expected does more harm to the progress and adoption of 3D Printing than simply saying no.

Always, my first step is set a meeting with client and understand why he think that 3D printing is his first option, then I′m going to explain the complexity of produce this piece with 3D printing and evaluate together, which technology it′s better for his specific application.

4 years ago I would have said "yes" and we'd have spent weeks banging our heads against a wall trying to find a way. Now, we sit down with the client, establish just what they need and work out the correct process to achieve it (which may not be 3D printing)

First understand the 3D printing technique required for the job at hand. Research and understand material compatibility for the said application. Understand the current requirement, is it a one off requirement or a reoccurring one. Evaluate the dimensional accuracy required and if first feasible in the chosen 3D printing technique. Understand the post processing requirements & acceptability after post processing. Only after understanding these, I would reason with the client about his 3D printing requirement.

One of the things you realize is customers often do not have realistic expectations of additive manufacturing. It can print anything in zero lead-time for little cost. Others have provided some excellent suggestions elsewhere in this article. If the task is impossible, you need to let the customer know that you cannot do his job. If it is really bad you don't want to send him to a friend either. One customer I dealt with about 10 years ago came to me with a printing problem that was not the right part for our process. His expectations were way out of line with what we could do. I had to say no. He did find a process that worked. He now owns a large service bureau that is replacing injection molding with additive manufacturing.