Data Visualisation for Project Decisions
Walk past a solar array with one row under a film of dust and you see the problem before anybody tells you there is one. No figures, no comparison, no arithmetic. That is the whole purpose of a good chart: to put a pattern in front of somebody in a form their eye resolves in a second, so the conversation starts at what to do about it.
Most project reporting does not do this. It presents numbers that are accurate, current and incapable of changing anybody's mind, usually because they describe status rather than direction. Section 5 of the PMBOK® Guide Eighth Edition lists visual data representation among the techniques a project uses, and the practical question is narrower than it sounds: which of your data has a pattern in it worth seeing.
Which data deserves a picture
Anything with a direction. A single figure for cost variance tells a board where the project is; four months of the same figure tells them where it is going, and the second is what they can act on. Where the underlying numbers move, a time series is almost always the highest-value chart on the page.
Anything where the spread matters more than the middle. Average turnaround, average duration and average quality all conceal the cases that cause the trouble. A picture of the distribution shows the tail, and the tail is usually the reason somebody is complaining.
Anything being compared across many items. Twenty suppliers, fifteen sites, thirty teams: a ranked picture makes the outliers obvious, and the outliers are where both the problems and the good practice live. A table of twenty rows hides both equally well.
Choosing the picture the decision needs
Start from the decision, not the data. The useful question is what somebody will do differently depending on what they see, and a chart that cannot answer it belongs in an appendix. This single test removes about half the content from most status packs.
For direction over time, a line, with enough history to show whether the recent movement is unusual. Three points are not a trend, and a chart that starts at the moment things got bad will be read as a crisis whether or not one exists.
For comparison, ranked bars, sorted by the value rather than by name. Alphabetical ordering is the commonest way to make a comparison unreadable, and sorting takes one click.
For binary progress, a count. Milestones achieved against milestones due needs no chart type more sophisticated than two numbers, and dressing it up as a percentage invites the partial completion discussion that milestones exist to avoid.
The average that hid two crews
A regional broadband programme reported build rate as an average across its contractor's crews. The figure sat close to target for most of a quarter, the monthly pack showed it as a steady line, and the programme board had no reason to ask a question about it.
The rate per crew, plotted individually, showed something else entirely. Two of the eleven crews were producing roughly double the plan, four were close to it, and five were delivering less than half. The average was being carried by a small number of people, and had those two crews been redeployed, the whole programme would have collapsed against its target within a month.
The cause turned out to be access rather than competence. The underperforming crews were working in an area where wayleave agreements with private landowners were outstanding, and they were spending much of the day waiting or relocating. Nobody had connected the two, because the access issue sat in a different report from the build figures. Reaching for the view that answers the question in front of you, and noticing when a headline number is hiding its own distribution, is practical analysis Omega's PMP® Exam Preparation works through with real reporting.
Where a good-looking chart misleads
Scales and windows do quiet work. An axis that does not start at zero exaggerates small movements, and a chart that begins three months ago can make a long-standing position look like a sudden change. Neither is usually deliberate, and both change what a board believes it is seeing, so the rule worth holding is that the shape of the picture should survive being drawn a different way.
Dashboards drift towards what is easy to measure. Ticket counts, hours logged and tasks closed are all available automatically, so they appear; benefit realisation, user experience and rework rates take effort, so they do not. Over a year, an organisation's attention follows its dashboard, and the things that were hard to measure quietly stop being managed.
For a PMP® candidate, a piece of reporting is read for the decision it supports. A scenario in which a project is producing detailed reports that nobody acts on is describing a communication and analysis problem, and the response worth considering changes what is reported, not how often it is sent.
Take your current status pack and mark every chart with the decision it informs. Anything left unmarked can go, and the space it frees is usually enough for the one view your sponsor has been trying to construct in their head for months.
Interested in going further?
Presenting project information so that a decision becomes obvious to somebody who was not in the detail is a skill that changes how your reporting is received. Omega's PMP® Exam Preparation covers measurement, analysis and communication as connected parts of the role.
The PMBOK® Guide Eighth Edition sets visual representation alongside the measurement techniques it supports, which is the useful context for it.
Ad · Amazon affiliate link.
References
A180: Facilitation Techniques for Difficult Project Conversations
A187: Retrospectives That Lead to Change
A179: Brainstorming That Produces Better Decisions
A178: SWOT Analysis in Project Management
A182: Multicriteria Decision Analysis Explained
PMP and PMBOK are registered marks of the Project Management Institute, Inc.
