Forecast's Task Overrun Alert is an AI-powered feature to help you keep an eye on tasks that may be at risk of taking longer than expected.
This article covers:
Overview of Task Overrun alerts
Forecast uses an AI model to generate task overrun alerts. This model is used to predict the risk of a task taking more effort than its recorded estimate.
The Task Overrun Alert depends on having accurate time registrations. If time registered does not reflect the true effort spent, your model estimates will be influenced by that.
The model is trained on completed tasks and their time registrations as an accurate measure of how long a task actually took. Forecast uses aggregated tasks and aggregated time registrations to train a foundation model of how long certain tasks usually take. This model is then refined using your company’s data to tailor it to how fast your teams are working.
The model considers the following properties of your tasks when making an estimate:
- Task Title.
- Task Description.
- Task Deadline.
- The role which is assigned to the task.
- Persons who are assigned to the task.
- The person who created the task.
- Task Status (which Status column the task is in).
- Role of the person who worked on the task before.
- Your user-provided estimate.
- How much time is registered on the task so far.
Forecast continuously evaluates the model performance on a subset of your recently completed tasks and only flags the tasks as potentially overrunning when 75% or greater confidence.
Viewing Task Overrun alerts
Task Overrun alerts are displayed on the Scoping and Sprints pages. When Forecast estimates that a task is at risk of taking significantly longer than the entered estimate, a yellow exclamation mark is displayed next to the Estimate field on the scoping pages.
You can click on the exclamation mark to get more details
Using Task Overrun alerts
We recommend discussing the next steps with your team, so you can adjust your estimate based on both the AI’s warning and your team’s expertise. You can update the new estimate at any time, by editing the Estimate field for the task.
Additionally, you can use the filter option to show only those tasks in a project that is at risk of overrunning.
Use this filter to get a quick overview of which tasks you might want to focus your attention on:
Within the Task Overrun Alert pop-up it is possible to discard the warning
Note: After discarding the warning, it will be discarded globally and will thus not affect just your view but also your team members' view.
Training the Task Overrun AI model
A lot of properties go into generating a warning, to help you assess the warning, the system provides you with 3 tasks that based on the system's understanding are similar and were relevant when generating the prediction.
These tasks should give you some reference points, however in some cases, the system might provide tasks that you do not find relevant as references, in this case, you can discard individual tasks, and the system will provide you with another task it thinks are similar.
You can discard irrelevant tasks by clicking on “Not Similar?".
By discarding tasks you do not find relevant, you help the system get better, and provide you with better reference tasks in the future.