- 1
Start from the outcome, not the tasks
Write the finish line in one sentence ("app live in the store", "office moved by March"). Then break it into 3 to 6 phases, and each phase into tasks. A task should be small enough that one person can own it and you would notice if it slipped a few days.
- 2
Estimate durations in working days
Estimate how long each task takes in WORKING days, not calendar days, weekends and holidays will stretch the calendar for you. When unsure, ask the owner of the task, and resist the urge to trim their number to make the plan look nicer. A plan that lies politely is still a lie.
- 3
Wire the dependencies
This is the step that turns a list of bars into a schedule. For each task ask: what must FINISH before this can START? Link those pairs (finish-to-start is 90% of real life; start-to-start and the others cover the rest, and a lag adds breathing room like "2 days after"). Once linked, moving one task cascades correctly through everything after it.
- 4
Assign owners and check the load
Give every task exactly one owner. Then look at load per person per week, if someone is booked at 145% in July, the plan is fiction no matter how nice the chart looks. Rebalance by moving tasks, splitting them, or changing owners before the project starts, not after the burnout.
- 5
Mark milestones and the deadline
Add zero-length milestones for the moments that matter (design approved, beta out, launch). They give the team fixed stars to steer by, and give you an instant answer to "are we on track" at each gate.
- 6
Freeze a baseline
Before work starts, snapshot the plan. From now on you can measure reality against intention: which tasks drifted, by how much, and whether the drift is eating your float or your deadline. Without a baseline you only ever see the current plan, which conveniently always looks intentional.
- 7
Track against the critical path
As work moves, watch two things: progress on each task, and the critical path, the chain that decides your end date. A slip on the critical path is an alarm; a slip elsewhere just burns float. When the deadline is threatened, simulate the fix before applying it: what happens if we compress this task, or move that one?
- 8
Share it in the format the room expects
A plan only works if people see it. Export to Excel for the operations meeting, PowerPoint or PDF for the steering review, MS Project XML for a partner who lives in Project. The chart is for planning; the exports are for alignment.
That is the whole method. The chart itself is the easy part; the discipline is in honest estimates, real dependencies, and checking the critical path every time the plan moves.