What are the metrics that can be analysed or derived from a Gantt Chart?

Question

Firstly, I am a programmer writing a software for analysing gantt charts.

Can someone please explain what types of analysis that project managers often do based on these gantt charts? What types of information do project managers derive from the gantt chart? What things can be automated and what cannot? Why?

Answer 1

I agree with the comments you've gotten so far, the question is a little open-ended. But in the spirit of helping -

I use the Gantt chart to give me a visual representation of things that haven't happened yet. I can look at it and see the flow of the project - when things are scheduled to be done, how tasks/deliverables relate to each other (dependencies), where resources are scheduled at any given time (on not yet scheduled), important dates (milestones).

The Gantt is where all of the pieces of the project are displayed collectively. Outside of this visual aid, I have to look into various docs for information - into contracts to see when a vendor is supposed to deliver something, or when subcontractor is due to start, emails for new dates or resource issues, etc.

It also gives me a place to see the overall project (at least in terms of deliverables. My contract may have the scope and deliverables written down, but that will undoubtably take several pages. The Gantt allows me to list it all in one long list, and then view it as I need to - roll up to milestones only, a specific deliverable, print it out and hang the whole thing on the wall so I can see it all at once.

To be honest, I'm not sure how a s/w package could 'analyze' this in a way that I would find useful, but I'd like to hear how you're going about it.

Answer 2

I second Trevor's description. What I would add is

• Critical Path analysis, but this is already done by most planning software
• Dependencies (inside or outside of the project), which require some time-buffering to mitigate possible slippage

The Gantt is only a representation of the data behind it, based upon start and finish dates (baseline, planned, actual) etc. Data can be used for all kinds of analysis. What I find useful is a "Milestone analysis", being the % that you hit your planned milestones on time (across projects). This is useful for organisations that use some 'standard' methodology for similar projects, with a defined set of milestones.

Answer 3

Interesting question - I would love to see where this question led in the end, if there is one.

I happen to be researching a very similar problem... related to personal scheduling. There are individuals who manage all their tasks and appointments in a single place - their calendar. I happen to be searching for quality indicators that can be used to give a user feedback. (I call them Total Task Schedulers.)

With respect to Gantt charts, I think there are a few metrics which also apply to personal schedules, such as: - an effective completion metric (lag) that tells how effective a chart was in retrospect. In other words, was it followed and to what degree? Did tasks take as long as predicted and were they completed on time - a predictive completion metric (lead) that tells how likely a chart will be followed in reality. - a tightness metric that estimates the amount of slack built in for emergencies - a measure of critical path dependence - how much does the chart rely on the critical path (are other paths also close to being critical?) (lead) - if historical data exits: are task estimates unbiased? (lead) - there can also be measures related to human resources and their ability t to deliver on time (lead)

(In each case I have indicated where a metric is a lead or lag indicator.)

Beyond this list, I suspect an experienced project manager can just look at a Gantt chart and interpret the state of a project with a glance! Perhaps they can also sit in on a project meeting for five minutes and do the same thing. It would be great to quantify those things they are looking for in each case.

Answer 4

• Earned Value Management derives a variety of information from a Gantt chart. (assuming that the Gantt chart has been built to support the analysis).

• Critical path(s) (@Stephan pointed this out and deserves the credit.) Calculation of critical path is vital; automated analysis could concentrate on multiple critical paths.

• Monte Carlo Analysis - this is probably the most valuable thing you could add. Monte Carlo Analysis is not easy to set up, but is the only way to answer questions about the consequences of multiple risks/opportunities.

If you are at a Project Management Office, and you have access to multiple Gantt charts from multiple projects, it might be interesting to analyze the charts for:

• Project closure rate - how many projects close successfully and how many close before the value is delivered?

• Estimation accuracy/variance - how close is estimated duration to actual duration? What variables affect accuracy. You'll need to separate issues of chance (aleatory issues like "the supplier had a trucking strike and was late in delivering materials", from epistemelogical issues like, "The dev team overestimated their familiarity with the BlueDuck programming framework."

• Which types of work packages were late/early and by how much?

• WHich teams were late/early? By how much and why?

If you have access to a set of Gantt charts, you could probably learn a lot by studying change mangement across the set.

• What is the velocity of change? (A colleague once worked on a project that had more than 100 authorized changes in a quarter; this was a clear signal of problems with the project. The project I was managing at the time had changes only prior to financial reviews - which was a sign of a different kind of problem).