A directed system is illustrated in Gantt charts that record the project work. In directed systems, we set start dates and durations and finish dates for the various project actions and deliverables. We also record the dependencies, or how work items connect to one another. For example: Put socks on first, then put on shoes. To build a schedule that makes any sense will require understanding how the work moves through the company, available talent and resources, and historical data to determine how much time is required for each item.

The dependencies of the tasks and the duration estimates of the tasks allow us to predict, within limits, when the project will be completed. More importantly, they allow us to recalculate our projections of an end date based upon actual performance. For example, as the tasks require more time than planned, we can immediately see the impact on the final estimated delivery date. An interesting, and free, tool for creating Gantt charts can be found here: www.ganttproject.biz.

If you are familiar with kanban, you have experience with pull systems. In pull systems, we do not know how long tasks will take to be accomplished. We make no pretense of knowing things that we do not know or cannot attempt to predict. We do not have a schedule, but we have the work items identified, and will pull things to do from a list, as needed and as talent becomes available. To ensure the workload does not become too heavy, we will limit the amount of work in progress.

Trello (https://trello.com/) is an interesting pull system that is maintained in the virtual world. But, there are benefits to a live system that is maintained in the work area. For example, you can divide up a whiteboard into the steps taken to accomplish the work, starting from a backlog that contains the list of things that must be done to accomplish the scope on individual sticky notes. We will take items from this backlog and move through the phases:

• Backlog.
• Work in progress.
• Testing.
• Finished.

Which one is better? Like all things it depends on numerous factors, a few of which are the specific project, industry and organization. Organizations that have a theme to the work—for example, developing instrument clusters for the automotive industry—may have enough relevant historical information to enable building a schedule. Additionally, the nature of vehicle launches requires some measure of prediction of product availability that is better suited to a directed system. For work that has a high degree of uncertainty, or when team members are highly skilled in their work, we may choose to use the pull system. Context matters.