Start with what cannot afford to fail.

The useful brief is rarely a feature list. Tell us the system, the constraint, the decision in front of you and what has already been tried. We will respond with the questions we need answered and the first evidence we would collect.