Hal Macomber provided an interesting post again about the topic of Gantt Charts. He asks the following questions:
- How is the project going?
- When will we be done?
Some other important questions might be:
- How much will it cost when we're done?
- What is the confidence interval for the cost and completion date?
Now Gantt charts don't answer these questions "out of the box," nor is Hal suggesting they do. In fact Gantt charts don't do much for the project manager by themselves - at least in modern times. This is the crux of the problem with Gantt. As a drawing tool they make nice bars showing tasks, with progress moving from left to right, time now, and some indication of how much work is to go.
But the core problem - and this is the core problem with all project management processes - is where does the underlying data come from to draw these pictures?
If the bars are drawn by the PM then they are just that bars drawn by the PM or the tool the PM is using.
And that makes the Gantt chart pretty much worthless as suggested by the critics of Gantt charts
Here's a Better Approach
- Define the work to be performed in some form - preferably work packages with well defined deliverables
- Assess the resources and duration needed for this "package" of work.
- Connect this work in a Network of activities - this Networking approach is the starting point of good planning. The PLAN is a strategy for delivering the work. It is not yet a schedule for performing the work. That comes next. Plan first, then schedule. No Plan, no credible schedule.
- Assign resources to this PLAN. These resources can be leveled with a cleaver tool or assigned and leveled by hand - the preferred approach in my book.
- Hook all the work packages together in their proper sequence. Determining this sequence is not trivial in some cases. But touching the network, looking at the network, assessing the feasibility and credibility of the network is what Planners do for a living. Planning is a verb.
- When the network looks like something you want to try out in practice by executing the project, you can draw the network as a Gantt Chart.
But this time the Gantt Chart has an underlying source of data - a credible source of data - the Network. This is a process skipped by most when they claim Gantt does not provide useful information. They are right - it provides no useful information, because there can be no useful planning information in the absence of the credible network of work.
With the Network We Can Do Other Things
Now that there is a network of work activities, with their durations, dependencies, possibly resource loaded and well behaved relationships (no Must Finish On, Must Start On, Start No Later Than, No Finish to Finish, or Start to Start, ..., just, Finish to Start and As Soon As Possible, except the Auhtorization to Proceed which can have a Must Start On) the network can be used for several purposes:
- A Monte Carlo simulation can be performed to assess the credibility of the network to actually finish with some level of confidence (80% is the number in my domain).
- The probabilistic critical path can be examined to determine which activities in the network are going to be trouble makers with some level of confidence (75% of the time on the critical path is the number we use).
As well the network can be used for Earned Value and now Earned Schedule
- Assign the resources and their cost or assign the cost spreads and you get BCWS time phased across the network
- Define some way of measuring physical progress - Physical Percent Complete - with the activities and at the conclusion of each activity and you can get BCWP.
- Capture the time card information in some way and you get ACWP
With these pieces of information I can tell you call kinds of interesting things
- The Estimated Cost at Completion - EAC
- The estimated time to complete using Earned Schedule
- The performance in units of measure interesting to the customer - business value - that each work package or team is delivering. And this is critical, how much they need to increase their performance to hold schedule and/or cost. This is the To Complete Performance Index (TCPI)
These pieces of information are easy to come by IF you have a well formed network and a credible cost baseline (BCWS) for each activity in the network. Everything can flow from there.
The Gantt is Just a Picture of the Network
Mr. Gantt made is picture before the networking, Monte Carlo, Earned Value, Earned Schedule processes were in place.
Any project manager that uses the Gantt as a Project Management tool - in the absence of an underlying credible activity network with credible BCWS, dependencies and risk adjusted durations - gets just what she asked for - A Poor Picture of the Project that doesn't provide value to anyone
A self fulfilling prophecy - Gantt Charts do not provide useful information
But really folks if you don't have a credible list of work, sequenced in the proper relationships, with credible cost and effort estimates, what do you have - nothing - just a pretty picture of a pipe dream of a project.
Add the Make Ready Concept for Construction
I don't pretend to understand or use Hal's Last Planner approach but it makes sense to me, but from his post
"xx% of the work items are ready when they should be ready. Of those items we are promising to complete yy% of them. This is better (or worse) than it has been over the last four weeks. zz% of the promised work is getting done as promised. Based on that we should be able to make the next two milestones."
Sounds a lot like the discussion we have weekly using Earned Value, with 0%/100% completion activities.
The Gantt is a nice way to draw pictures from the database of well formed baselined performance managed activities. Nothing more, nothing less.
One Final Thought
Planning and scheduling are different:
- The Plan is the stratgey for delivering the business value
- The Schedule is the means of executing the work needed to produce the deliverables that produce the business value