Bring me your project plan – particularly your Gantt chart – and I can tell you if it’s OK
The Gantt chart says everything about a project plan – in fact I don’t even need to know what the names of the tasks are, or who is doing them – just the SHAPE of the Gantt tells me about 90% of the potential problems with a project.
Here is what I look for:
- No critical path – if the Gantt chart is just a collection of bars drawn against dates then it’s missing a key component; we need to know which tasks are defining the minimum duration of the project (the “critical” ones) and which tasks have float – which ones we need to watch carefully and which ones have leeway.
- The plan is too linear – one task after another, with nothing much in parallel. This means the project will take too long, it’s too safe. I bet SOME of those tasks can be done concurrently to save time!
- The plan is too parallel – too much is happening all at once. Yes it’s quick, but it’s too risky. Are you sure that some of those tasks don’t’ depend on each other, and also, are you sure you’ve got enough resource to do everything at once. And remember, if there’s a problem with one of the tasks, you might wish you hadn’t started on some of the others already.
- The plan has a lot of small tasks – by small I mean less than a week – all in a row. Because it’s likely that at least one of them will run late and affect all of the others. This is particularly risk-prone of they are all at the end, or near the end, because you’ve got less time to catch back up by compressing one of the other, larger, tasks.
- One (or two) of the tasks is much bigger than the others. This task needs to be broken down because otherwise it’ll be difficult to track its progress, and you might get a large and nasty surprise when it’s due to be finished and you discover that it’s nowhere near finished.
- Tasks overlapping are allowed but not ideal in a Gantt chart. Concurrent is fine, but partially overlapping is a problem because it’s messy, not clearly defined. At what point during task A can you start B, exactly? OK then, whatever that point is, split A into two parts and plan to start B after the first part of A.
- What is the resource loading like? If there are lots of floating tasks all planned to be done at the same time, when they could be floated forwards or backwards to spread out the resources required, then that’s an improvement to make, and it implies that resourcing / load hasn’t been considered at all.
- If all of the floating tasks are planned to be done at the start of their floating range, i.e. as early as possible – or the reverse, if they are all being left till the last minute – then that makes me wonder if floating of tasks has been properly considered. In a good plan there will be a mixture: some tasks will be done early because their time estimates are hard to predict, while others will be left till later in their range because they are expensive, or their resources clash with something else.
- Sometimes tasks are described as “ongoing” or drawn as one long bar throughout the whole project. For example “communication” or “cost management”. These should be better defined, being broken down into weekly or monthly checks, or phase 1 2 and 3, rather than left vague.
- All tasks should have constraints before and after. The critical ones form one chain with no gaps, comprising the whole length of the project. The others, the non-critical or floating ones, have constraints, usually the end or beginning of critical tasks, which define when they allowed to start (they can’t start before this point) and then they have to finish by – if they run late they will pash the critical path late. So their connections to the critical path are something I would want to see.
- You can’t have loops that go from the finish of a task BACK IN TIME to the beginning of an earlier task, because otherwise the project could take for ever.
- You also can’t have BRANCHES that say “if this happens then go to there, otherwise go to there” because otherwise we don’t know how long the project will take.
Those 12 are the essentials. But also there are two more things I would look for – they are “desirables” rather than pass or fail essentials.
- Ideally each task will have a name assigned to it, for who is going to carry it out, and also a cost assigned to it – at least a total cost, and ideally a spend plan for each week or month as the task progresses. You could also have totals for each week and totals for each task. The hours required could be on one sheet and the costs required could be on another.
- Too many tasks – I don’t like to see more than 30 tasks on a Gantt chart, because I don’t think the human eye / human brain can keep old of the plan if it’s this complicated. I would rather see an overview with 15-30 tasks and then have separate “sub-Gantts” giving more detail. For example the plan to build a house could have a high level plan with this like “Install kitchen, 6 weeks” and then a separate kitchen plan can have all the details. This is much better than one massive Gantt with hundreds of tasks.
