When you set due-dates for goals (for example move to profitability by EOY) you are thinking what would make them ambitious but achievable. The due date can based on current trends, on time it to to achieve similar goals in the past or on any other means. It’s definitely not an exact science. Note that this not effort estimation (at this point you don’t even know which ideas or projects will be chosen)— it’s timeboxing.

Step-projects and tasks you estimate the same way you estimate any work item (same as you do in scrum). Since they’re both rather short the natural tendency to underestimate (the planning fallacy) should not get you into huge time overruns.