Draw them up as tree, so each positive validation results in tackling next riskiest assumption in sequence, and negative validation results in going back a step.