An early company can work on almost anything. The discipline is choosing the one milestone that would most reduce uncertainty about the business.
A good milestone is specific and observable: enough customer conversations to confirm the problem, a first paying pilot, or a product that a defined group uses repeatedly. “Launch the app” and “grow the brand” are activities, not milestones.
Work backwards from it. Which assumption does it test? What has to be true to reach it? What can wait until afterwards?
Then let it shape decisions. If a feature, hire or partnership does not move you towards the milestone, it can probably wait. When you reach it, review what you learned before choosing the next one, and write down why you chose it.