Governance Metrics That Actually Tell You Something

Ask a steering group how well its governance is working and the answer usually arrives as a report on the machinery. Meetings held against meetings scheduled. Attendance percentages. Papers issued three working days in advance. Actions closed. Returns received from every workstream. All of it is true, all of it is cheap to collect, and very little of it says whether the arrangement is doing the thing it was set up to do. A programme can hold every scheduled board, achieve ninety per cent attendance and still wait seven weeks for a decision it needed inside two.

Governance produces decisions. The papers, the forums, the delegated limits and the escalation routes all exist to get a sound decision made at the right level while it is still worth having. That makes the decisions themselves the measurement surface: how long they take, who ends up taking them, what information they rested on, and whether they survive contact with the following month. Almost all of it can be read from records most projects already keep. The difficulty is deciding what to count, not finding it.

Activity measures dominate because they are countable, blameless and quick. Nobody has to write down that the sponsor sat on a paper for six weeks. They also let a project office show effort without anyone having to examine the behaviour of senior people, which is the honest reason governance measurement is rare. Anything worth measuring here describes how executives use their authority, so it needs a sponsor who has agreed to see the results before the first number is produced.

Measures you can take from the decision log

Decision latency is the elapsed time between a decision being formally requested and an answer being given and recorded. Start the clock when the project asked, not when the item appeared on an agenda, because the wait for the next available forum is part of the delay the project actually experienced. Latency read this way exposes cadence problems that no attendance figure will: a monthly board handling a programme whose decisions arrive weekly is structurally late even when every meeting runs perfectly.

Deferral counts sit alongside it. Record how many times an item was tabled before it was resolved. An item carried forward three times is telling you something specific, and it is usually one of three things: the paper does not give the board what it needs to decide, the person who can decide is not in the room, or the decision belongs at a different level altogether. Counting deferrals separately from latency stops one long delay and three short ones from looking like the same problem.

Durability tests whether decisions hold. Count how many were reopened or reversed inside an agreed window, three months being a reasonable default, and note the reason each time. Some reversal is healthy, because conditions change and a board that never revisits anything has stopped paying attention. A run of reversals within a few weeks usually means decisions were taken without the analysis they needed, or that the forum committed to something it did not have the authority to commit to.

Escalation fit works in both directions. On one side, items that reached the board when a delivery manager could have settled them, consuming senior time on matters inside someone else's remit. On the other, items settled below the threshold that should have come up, which normally surface later as a surprise. Sampling the decision log against the thresholds written into the governance arrangement is the measure most likely to change behaviour, because it puts the thresholds themselves on trial instead of blaming the people working to them.

Information adequacy is the quietest of the five and often the easiest to fix. Track how often a board asked for more information before it would decide, and what kind. If half the deferrals are waiting on the same financial view, that is a reporting fault with a cheap remedy, and treating it as a leadership failing wastes everybody's time.

What the attendance figures are still good for

The activity measures have a real job, as leading indicators of capacity. Quoracy failures and substitute attendees predict latency, because a board that cannot decide will defer, and a deputy with no delegated authority guarantees it. Late papers predict deferral in the same way. Keep collecting them, and read them as early warning of the problems the decision measures will confirm a quarter later.

Single measures also invite gaming. A board told it will be judged on decision speed will decide faster and think less, which is why latency should never be reported without durability beside it. Deferral counts on their own can push a chair towards closing items that genuinely needed another look. Pairs of measures that pull against each other are safer than any single number, and a short written comment against the outliers is worth more than a fourth decimal place.

Comparison between projects needs care. Approving a change inside tolerance is not the same decision as approving a funding uplift, and a programme whose decisions are mostly commercial will always look slower than one whose decisions are mostly technical. Trends within a single governance arrangement are far more informative than league tables across a portfolio.

Six months read from a steering group's own record

A regional distribution business is running a warehouse automation programme with a monthly steering group. The governance section of its dashboard is entirely green: six of six meetings held, average attendance ninety-two per cent, papers issued on time in five months of six, every action closed.

The programme office analyst reads the decision log instead. Eighteen decisions were requested across the period. Twelve were resolved at the first attempt, six were deferred at least once, and two were deferred three times. Median time from request to answer was twenty-six days, but the six deferred items averaged sixty-one, and every one of them concerned spend above the programme manager's £50,000 delegated limit. Two decisions were reversed within eight weeks, both after the operations director, who sends a deputy in alternate months, saw the substance for the first time.

Three sentences of analysis follow from that. The delegated limit is set too low for the tempo of the programme, so routine commercial choices queue for a monthly forum. The authority the board depends on is not reliably present. And the reversals are not indiscipline, they are the predictable result of the first two conditions. The remedies are equally concrete: raise the delegated limit with a reporting-back requirement, make the operations director's attendance non-delegable for spend items, or add a short fortnightly decision slot with the sponsor. None of it needed a new data collection, and none of it would have emerged from the dashboard.

Making the measures worth the effort

The project office or project manager compiles this; the sponsor owns it. Quarterly is usually the right cadence, because a governance measure needs enough decisions behind it to show a pattern, and a monthly version becomes another paper for a board that is already deferring things. Scale it honestly, too. A project with nine decisions across its whole life does not need a latency distribution, though the question of how long the last three decisions took is still worth asking out loud at a phase review.

Adaptive and hybrid delivery does not remove the measure, it moves it. Where decision rights sit with the team, the equivalent of latency is how long a team stayed blocked waiting for an answer from outside it, and the equivalent of durability is how often a team decision was overturned by a governance body afterwards. A high count of the second is a sign the decision rights were described more generously than they were meant.

The Governance Performance Domain in the PMBOK® Guide Eighth Edition includes metrics and mechanisms for effective project governance at Section 2.1.3. The broad point worth taking from it is that measurement belongs inside the design of the arrangement, settled when decision rights, thresholds and forums are agreed, and not improvised months later when somebody starts wondering whether the board is earning its time. Deciding at the outset how you will know the governance is working is itself a governance decision.

For a PMP® candidate, the connection runs through the Business Environment domain, which accounts for twenty-six per cent of exam items and includes the task of defining and establishing project governance. That task names defining success metrics and outlining escalation paths and thresholds as part of the work, so measurement and escalation design are treated as the same piece of thinking. Because the exam is built on scenarios, the useful preparation habit is recognising the symptoms in a described situation: an item that keeps returning to the same agenda, a decision taken two levels above where the threshold placed it, a board that cannot answer because the paper never contained the numbers. Candidates who work through governance situations in a structured setting, such as Omega's PMP® Exam Preparation, tend to find the measurement question easier, because they have already had to argue about who owned the decision before deciding how to judge it.

On a live project the value is more immediate. Governance is often the slowest process a project depends on and the least examined, and a project manager who raises it usually sounds like someone making excuses. Twenty-six days is not an excuse. It is a number a sponsor can act on, drawn from their own board's record, and it turns a complaint about slow decisions into a proposal about delegated limits and attendance. That shift, from grievance to evidence, is most of what governance measurement is for.

Andre Malowney

Interested in going further?

Governance situations are among the hardest to read in an exam scenario, because the visible problem is nearly always a delay or a disagreement while the real question is where the decision belonged and who should have seen it. Omega's PMP® Exam Preparation works through that reasoning in a structured way, using the same judgement a project manager needs when a steering group starts carrying items forward.

The Governance Performance Domain, including its treatment of the metrics and mechanisms that make governance effective, is set out in full in the PMBOK® Guide Eighth Edition.

Ad · Amazon affiliate link.

References

A090: Project Governance Explained: Who Decides What, and Why
A091: Project Governance vs Organisational Governance: What's the Difference?
A093: Choosing a Project Governance Model
A096: How to Recognise a Governance Question on the PMP Exam
A053: AI Governance for Project Managers

PMP and PMBOK are registered marks of the Project Management Institute, Inc.