A suggestion for clients and developers: for non-trivial projects, treat the investigation/discussion/specification phase as a standalone project. Give it a separate budget and timescale, and don’t commit to precise features or accurate costs for the build phase until it’s all planned out. In fact, don’t even commit to who will be doing the building. There’s rarely a downside to this, and it has some clear advantages:

• The overall objective is broken down into more manageable chunks.
• The client and developer both get to see if the working relationship is OK with minimal risk.
• It helps to ensure that the developer doesn’t skimp on consultation and planning (it’s sometimes tempting to go for an obvious, safe solution), and that the client sets aside money for it.
• It avoids premature, ill-informed decisions (often the cause of failed projects).
• There’s more chance of being able to find innovative ways to save time/money, or improve the final result.
• You can’t accurately establish what will be produced for how much and by when until the thinking has been done, so why try? Acknowledging what you don’t yet know helps to reduce stress and confusion.

Unfortunately, as with being open about budgets, many people are oddly reluctant to consider this approach, but I find it works well.