A PMO dashboard can mix figures that do not speak about the same object. PMO KPIs fall into three families. The first describe each project: milestones, dates, costs, risks. The second describe the portfolio, that is, what the organisation has committed to relative to what it can carry out. The third describe the PMO itself: the freshness of its information, the time it takes for a decision to be made, then implemented.
Each family has its reader and its decision. An indicator is read with its calculation convention, over several periods, and knowing what it leaves out of scope.
PMO KPIs sorted by their reader
Project measures serve the project manager, the sponsor and the steering committee, who decide on the next steps of a project. Except when it keeps a project’s instruments itself or runs it, the PMO does not produce them. It sets their definition, collects them and makes them comparable from one project to another. Portfolio measures serve the management that chooses between projects; the PMO calculates them from data that no single project holds. The measures the PMO keeps on itself serve its principal, who decides on its mandate and its resources.
A dashboard that mixes the three families forces each reader to look for their share. Our entries on the definition of the PMO and on PPM set out the notions of project office and portfolio used here.
The project measures the PMO consolidates
The table lists measures common in a consolidation of projects. The calculation convention chosen for each appears in the dashboard’s legend.
| Indicator | One possible convention | What it does not show |
|---|---|---|
| Milestones met | Milestones reached by their baseline date, relative to milestones planned over the period; variant with a tolerance of a few days | The weight of each milestone, and the meaning of “reached” if no evidence is required |
| Slippage of the end date | End date forecast today, minus baseline date; a delay is positive here, other conventions reverse the sign | The cause of the delay, and the scope removed to hold a date |
| Cost variance at completion | Forecast cost at completion, minus baseline budget; same caveat on the sign | What was delivered for that cost |
| Open risks | Number of risks whose rating exceeds a threshold set in advance | The risks no one has declared |
The first three rows depend on a baseline. If it is reset at each replanning, the variances fall back to zero and the series no longer shows anything. The convention therefore states which one is used for the calculation: the initial baseline, the latest one approved by the committee, or both, displayed side by side.
The aggregate adds its own convention. A “share of projects in green” changes meaning depending on whether it counts the projects or weights them by budget. It also depends on the colour rule, discussed together with the measurement of actual progress in our article on the project review. The number of open risks first tells you about how well the register is kept, the subject of our article on project risk management.
The portfolio is measured in ratios between demand and resources
For the whole portfolio, the PMO calculates, for example, the number of projects in progress and pending, the committed workload relative to capacity, the allocation of the budget between project categories and the age of requests left unanswered.
The number of projects ignores their size, and falls when two projects are merged without the workload having changed. The ratio between workload and capacity, calculated over the whole headcount, can seem comfortable while a scarce profile is booked beyond their available time. It also depends on the granularity chosen (skill, period) and on how ongoing activity is treated, deducted from capacity or not. The budget allocation shows where the money goes and says nothing about what it brings in.
The use of these measures in an arbitration cycle is described on our page on project portfolio management.
The PMO measured by its effects
The measures in this family concern the effect of the office: is the status of projects known, are decisions taken and then implemented, do project managers find a service there? Our page on setting up a PMO uses them as signs of success. Each requires a convention and leaves something out of scope:
- freshness of information, that is the share of projects whose status has been updated since the last cycle: it does not say whether that status is accurate, since an update can copy the previous one;
- decision lead time, counted here from the day a request for arbitration is logged to the day a record of decisions settles it: starting from the appearance of the problem is another convention, harder to date, and this lead time depends on the committee as much as on the PMO;
- implementation of decisions, that is the decisions due and implemented, relative to the decisions due: a vague decision is easily declared implemented;
- use of the PMO, that is the requests for help project managers address to it without being obliged to: the count says they come, and what they get from it is asked in person.
Activity counters (minutes circulated, committees held, templates published) describe output. They help size the team and measure no effect. A last measure looks at cost: the time project managers spend feeding the PMO, known by self-reporting and therefore approximate.
An indicator is read as a series, with a constant definition
An isolated value lends itself to several readings. The same measure, tracked over several periods without changing its definition, rules out some of them. When a definition changes, the date of the change is noted on the series, otherwise a break in calculation passes for a change in the projects.
The threshold that calls for a reaction is written down before the reading, together with the expected reaction. Set after the fact, it can be adjusted to what everyone wishes to conclude.
Some measures are read in pairs. Milestones met are only reassuring if the scope has not been reduced to achieve them, and a cost on target only if what was planned has been delivered.
Finally, measuring can change what is measured. When a project manager’s appraisal depends on an indicator, the declarations can adapt to it: milestones broken down more finely, baselines reset, risks kept quiet. The PMO therefore has an interest in using its indicators to open questions, and in keeping the evaluation of people separate.
Where the figures stop
Some subjects remain beyond the reach of a PMO dashboard. The value obtained after delivery is observed later, with the business owner who committed to it. Whether the choice of projects was sound is a matter for management’s judgement.
Several of these indicators also rely on declarations. As long as declarations and facts agree, consolidation is enough. When the consolidated figures and the account of the people close to the project contradict each other from cycle to cycle, one more indicator settles nothing. The next step is then an independent examination, conducted on documents and through interviews: our page on project audit describes its formats.
To sort through an existing dashboard, the PMO notes next to each indicator the last decision it informed and the name of the person who took it. An indicator whose line remains empty for several consecutive cycles is a candidate for withdrawal.