Engineering speed is wasted when the goalpost moves every sprint. The fastest teams we work with spend real time up front naming the problem, the user, and what done means in measurable terms instead of just shipping tickets.
A useful pre-build artifact is a one-page brief covering objective, non-goals, constraints such as time and budget, and a ranked list of assumptions to validate. That document becomes the reference when scope creep appears, and it always appears.
Phasing matters. The first milestone should prove the riskiest assumption instead of polishing the easiest screen. Often that means a thin slice through the stack with real data rather than a beautiful mock with fake content.
When strategy and engineering share the same vocabulary, estimation gets better and surprises get smaller. Everyone knows which decisions are reversible and which ones need executive alignment before you pour concrete.