Project Health Check
The picture you're seeing is incomplete.

The question nobody asks 0:00

Your programme reports its own health. Who checks the report? Every programme is described to its sponsor by the people delivering it. One line runs from the work to the sponsor. Everything the sponsor knows has travelled up it.

One line, from the work to the sponsor 0:16

At the bottom is the work itself. Design, test, data, configuration. Above that, the people accountable for the work. Above them, the people accountable for those people.

Four or five hands 0:31

By the time a read reaches the steering committee it has passed through four or five hands. Every one of them is describing their own performance. This is not a claim about honesty. It has been audited.

The Auditor-General checked 0:45

In twenty twenty-two the Australian National Audit Office checked what Commonwealth entities had reported about their own security maturity. Both entities it examined had reported inaccurately. The report named the cause. Optimism bias in self-assessment. Most of this is not dishonesty. Every hand it passes through has a reason to soften it.

Three corrections that make it worse 1:10

So programmes reach for three corrections. Each one is reasonable. Each one has been shown to make things worse. The first correction is scrutiny. Programmes add auditors and tighten the reporting. Where trust is already low, tighter scrutiny produces more misreporting, not less. Instead, the programme starts managing the auditor.

Power distance: seniority makes it worse 1:34

The second is seniority. Programmes put a more powerful sponsor above the work. The wider the gap in power between the person writing the report and the person receiving it, the more that report is biased. A stronger sponsor buys resources. What it does not buy is accuracy.

Escalation, and the deaf effect 1:55

The third is escalation. Programmes trust the process to carry bad news upward. Bad news is most likely to be ignored when it comes from someone with no standing platform to report it. The researchers call it the deaf effect. Three corrections, all applied to the same line. A reporting line cannot verify itself.

Build a second line 2:19

So build a second one. A read of the programme that does not travel up the delivery structure, written by someone who is not on it and is not being assessed by it. Not a compliance audit. A peer review, by a practitioner who has run programmes of this shape before. The evidence is the same evidence. The path is different. And that line is ours.

How often the second line is open 2:41

That leaves one question. How often that line is open. Once, before a decision. Once, on the single question keeping you awake. Or every week, for as long as the programme runs. A risk that has been amber for eleven weeks is a different risk from one that turned amber on Tuesday. In a snapshot, only one of them is visible.

Assurance is not a better report 3:06

Assurance is not a better report. It is a second reporting line. It only works when the people writing it are not the people it describes.

One programme. One reporting line. 3:18

One programme. One reporting line. Written by the people it measures. Rydel Group. Independent PMO advisory.

Both lenses read green 0:00

Governance was green. Delivery was green. The programme still failed. The Five Lenses. A health check looks at a programme through five lenses. Governance. Delivery. Risk and issues. Stakeholder. Benefits. Each one has an owner, a register, and a line on the status report. And each one can be sound on its own while the programme is not.

Programmes fail between lenses 0:29

Programmes do not fail in a lens. They fail in the gap between two of them. Governance was sound. Delivery was tracking. Both reports were true.

Ten pairs, ten places to fail 0:40

The decision that mattered needed both, and no single register owned it. Five lenses make ten pairs. Ten places where a programme can fail and nobody has reported anything wrong.

Seam one: scope stability 0:55

Scope stability is one. It sits between Delivery and Benefits. Delivery reports that the change was absorbed. That is true. Benefits reports that the business case still holds. Also true. Neither reports that the case now rests on a scope nobody re-approved.

Seam two: escalation 1:14

Escalation is another. It sits between Governance and Risk. The register holds the risk. The forum holds the authority to close it. A risk can sit open for nine weeks and never reach the forum that could act. Nothing in either register is wrong. The risk is still open.

Seam three: sponsor presence 1:34

Sponsor presence is a third. It sits between Stakeholder and Governance. Attendance is recorded. Position is not. A sponsor can be at every meeting and behind none of the decisions. This is not only our observation.

What the research found 1:50

In 2020, researchers screened six thousand abstracts on why large projects underperform. Eighty-six papers, eighteen causes, sorted into six themes. The themes are the ones you would expect. Governance. Delivery. Stakeholders. Risk. What the review said it still could not answer was how those themes interrelate.

The register scores risks alone 2:15

A second group went after the register itself. In 2025, researchers at two Australian universities modelled project risk as a network. Because risks are coupled. One of them moving moves the next. A register scores each risk on its own. That is the assumption the network breaks.

What a hundred and twenty buys 2:36

A seam has two sides. You cannot see it from one of them. That is what a hundred and twenty questions are for. Not thoroughness. Triangulation. We ask the same ground twice, from two lenses, where the answers should agree. Where they do not, that is the finding.

A lens, or five 2:56

A lens tells you how one thing is going. Five lenses tell you where they disagree.