What this section is for
Planning calculates P50, P80 and P95 dates: the dates by which 50%, 80% and 95% of model runs finish. It uses remaining task counts and a team velocity taken from history or entered manually. These are simulation results, not proven probabilities of completing the real project on time. Compare forecasts with actual completion dates.
When to use
- Before discussing a date, to check the task list and the velocity behind the forecast.
- After scope changes, to compare scenarios while holding other inputs constant.
- When reviewing previous forecasts against actual outcomes.
Quick start
- 1Create an initiative and include the remaining tasks. The simulation forecasts task counts, not hours or story points.
- 2In the capacity profile, choose velocity from completed-task history or enter it manually. For historical velocity, select the number of weeks to use. The unit is completed tasks per week.
- 3Run the forecast. Save P50, P80 and P95 with the task list and calculation settings.
- 4Change one input per scenario. The model scales velocity in proportion to team size; a real team may not speed up at the same rate.
Key settings
| Setting | What it does | How to use it | Common mistake |
|---|---|---|---|
| Velocity type | Chooses between velocity derived from completed work and a number you enter. | Use history when it reflects the current team and work. Enter velocity manually when the team or tasks have changed substantially, and record why. | Entering a desired velocity without evidence and treating the resulting date as realistic. |
| Weeks of history | Sets the period used to calculate historical velocity. | You can start with 8–12 weeks. Check that they reflect the current team and its usual tasks. | Setting a lookback that spans a reorganisation, so the simulation averages two different teams. |
| Uncertainty low / high | Sets how much velocity can vary between model runs. | Compare several forecasts with actual results before changing the range. | Narrowing the range only to get an earlier date. This changes the forecast, not the team’s pace. |
| Initiative scope | Defines which items the forecast is counting. | Check the task list before calculating. Run the forecast again after adding or removing tasks. | Treating a task-count forecast as a story-point forecast. Unestimated tasks still count, while their total point estimate remains unknown. |
Usage example
Scenario: You need to discuss a date for an initiative with 20 remaining tasks. This is an illustrative example, not a forecast for your team.
- 1Check the 20 tasks and the history of completed tasks across full weeks, including weeks with no completions.
- 2Run the forecast and save P50, P80 and P95 alongside the history window and capacity settings.
- 3If some work can be deferred, compare a scenario with 17 tasks — 15% less scope — at the same velocity.
- 4When discussing a date, state the scenario and assumptions. Recalculate when scope or conditions change.
What this shows: You compared dates for 20 and 17 tasks at the same velocity. This helps decide what to defer, but does not guarantee a finish date.
FAQ
How should I choose between P50, P80 and P95 dates?
P50, P80 and P95 refer to the share of completed simulation runs. A later date provides more margin within the model assumptions. Choose a planning date based on the cost of delay, record the task scope and compare past forecasts with actual completion dates. No percentile guarantees a real release date.
What does P80 mean?
It is the date by which roughly 80% of model runs finish under the selected inputs and assumptions. It does not establish an 80% probability that the real initiative will finish by that date. That requires checking forecasts against actual outcomes.
Why is there a large gap between P50 and P95?
Completion history, weeks with no completions, remaining scope and the configured velocity spread all affect the gap. Check these inputs. The interval alone does not identify the cause of delays.
Why is one date, or the whole forecast, missing?
Check remaining tasks and team velocity. Manual velocity must be greater than zero. The model calculates up to 520 weeks; if too few runs have finished by then, the corresponding date stays unknown.
Tips
- Compare forecasts for a comparable team and work mix. Splitting a task changes the count even when the underlying work is unchanged.
- Save a new forecast and its assumptions after scope changes. Compare past forecasts with actual dates to assess how useful the model is for your team.
- If manual velocity exceeds historical velocity, record why. Adding story points does not automatically improve a task-count forecast.