By the time a project is formally declared to be in trouble, the trouble is old. The slippage that triggers the red rating happened months earlier. The decisions that caused it happened earlier still. What the board is reacting to is not the problem arriving. It is the problem finally becoming impossible to hide.

I have been called in to recover programmes for thirty years, and the pattern rarely changes. When you reconstruct the timeline afterwards, the evidence that the project was failing was almost always visible six to nine months before anyone acted on it. Not buried. Visible. Sitting in the reporting, in the room, in the tone of the meetings. The failure was not a failure to have information. It was a failure to read it.

This matters because the cost of intervening rises steeply with delay. A problem caught early is a conversation. The same problem caught late is a recovery. So the most valuable thing a board can learn is not how to run a recovery. It is how to see the need for one before the dashboard does.

The signals are in the human layer

The instinct is to look for the warning signs in the numbers. The numbers are where they show up last. The earliest signals are behavioural, and they are usually the opposite of what people expect.

Reporting that gets smoother is one. A project in genuine difficulty does not usually produce messier reports. It produces cleaner ones, because effort is being spent on managing the message rather than the work. When the narrative in a status report starts to outpace the evidence behind it, that is a signal, not a comfort.

A narrowing of who speaks is another. In a healthy programme, bad news travels upward easily and from many directions. As a programme starts to fail, the range of voices in the review narrows to the few who are managing the relationship with the board. The quiet from everyone else is not agreement. It is the sound of a team that has learned which news is welcome.

Then there are the quiet redefinitions. A milestone date that moves between baseline versions without a decision to move it. Slippage absorbed silently into float. A scope statement that softens from a firm commitment to a conditional one. An acceptance criterion that loses a clause. None of these announce themselves. Each is a small act of making the present look more like the plan than it really is. Individually they are noise. As a pattern, they are the clearest early signal there is.

A project in genuine difficulty does not produce messier reports. It produces cleaner ones.

What to watch, and what it tells you

Watch the gap between confidence and evidence. When questions in a review are answered fluently but not actually answered, when you are given a reassurance where you asked for a fact, the gap between how settled it sounds and how settled it is is the thing to measure.

Watch the direction of the float. Float is consumed quietly and almost never reported as a leading indicator, yet the rate at which it erodes tells you more about trajectory than any milestone. A programme that is using its contingency faster than it is using its time is in trouble whatever the RAG status says.

Watch the people. Capable people leaving a programme they were committed to is rarely about the money. They are often the first to read the situation accurately and the most able to act on it. An unexplained departure of someone good is a data point about the project, not just about the person.

When to act

The hardest part is not seeing the signals. It is acting on a pattern before you have proof. Boards are trained to wait for evidence, and by the time the evidence is incontestable the cheap interventions are gone. The discipline of early recovery is to act on the weight of the pattern, accepting that you may sometimes act on a programme that would have recovered on its own. That is the right trade. The cost of a slightly early intervention is small. The cost of a late one compounds every month you wait.

The signals are nearly always there. The job of the intelligent client is not to acquire more information. It is to be the person in the room who is actually willing to read what the information is already saying.

Almost all project reporting is a record of the past. Percentage complete, milestones hit, spend to date, this period's RAG status. It is a careful description of where the programme has been. It is also, for the purpose of steering, close to useless, because by the time something appears in that kind of report the moment to act on it has usually gone.

The reporting that actually helps a board do its job answers a different question. Not what has happened, but what is about to. That shift, from describing the past to anticipating the future, is the whole difference between reporting that decorates a meeting and reporting that changes a decision.

Why status reporting fails the board

Status reporting has two problems. The first is that it lags. A milestone is reported as met or missed only once it is due, which is the last possible moment to learn anything. The second is that it is easy to manage. A RAG rating is a judgement, and judgements made by the people being judged drift green. The result is the familiar watermelon report: green on the surface, red all the way through, and entirely truthful by its own lights.

Neither problem is solved by a better dashboard. A real-time status report is simply a lagging indicator delivered faster. The issue is not latency. It is that you are measuring the wrong thing.

A real-time status report is a lagging indicator delivered faster. The issue is not latency. It is that you are measuring the wrong thing.

What predictive reporting measures

Predictive reporting is built on leading indicators, the quantities that move before the outcome does. They are less comfortable than a status colour because they are about trajectory rather than position, and trajectory implies a future the team has to own.

The rate of float erosion, not the float remaining. The trend in change volume and the time it takes to close a change, not the change count. The age of open decisions, because decision latency is one of the most reliable predictors of slippage there is. The movement in the forecast cost to complete from period to period, which tells you far more than the variance to date. The trend in requests for information and in quality non-conformances, which lead delay by months. None of these tells you where the programme is. Each tells you where it is going.

Why it is a discipline, not a dashboard

It is tempting to think predictive reporting is a tooling problem, that the right software will surface the right indicators. It will not. The hard part is human. It is choosing which leading indicators matter for this programme, demanding the forecast rather than the actual, and, above all, building a culture in which an early warning is rewarded rather than punished.

That last point is decisive. Predictive reporting only works if bad news delivered early is treated as the system working, not as a failure to be managed. A board that reacts to an early warning by interrogating the messenger will get status reporting back within a quarter, however good its dashboard is. The reporting follows the culture, not the tool.

So the board's task is not to commission a new reporting system. It is to start asking a different question. Not 'what is the status', but 'what is the trajectory, and what will bite us next'. Ask that consistently, and reward the honest answer, and the reporting reshapes itself to serve it.

NEC4 Option C, the target cost contract with a pain and gain share, is sold on a simple and attractive idea: align the interests of owner and contractor by giving them a shared stake in the outturn. Beat the target and you split the saving. Overrun it and you share the pain. On paper, both sides now want the same thing. In practice, the alignment is real but it is not symmetrical, and the asymmetry runs against the owner unless the owner does something about it.

This is not an argument against Option C. Used well, in capable hands, it is one of the better instruments available for complex delivery. It is an argument that the alignment it promises is not a property of the contract. It is a property of how hard the owner is able to work the contract, which is a different thing entirely.

Where Option C genuinely works

When it works, it works because it rewards the behaviours you actually want. Early contractor involvement becomes worthwhile rather than a favour. Transparency through the open book gives the owner a real view of cost as it is incurred. A genuine shared interest in finding savings replaces the zero-sum fight over every variation that characterises a lump sum job under stress. On a programme of real uncertainty, where a fixed price would simply be a fixed guess with a large risk premium baked in, Option C lets both parties manage the uncertainty together rather than pricing it and arguing about it.

Where the alignment is one-sided

The trouble starts with the target itself. The whole mechanism pivots on the target cost, and the target is set at the point where the owner usually has the least information and the contractor the most. Set it loose and the gain share pays out for ordinary performance. The owner then congratulates a contractor for beating a target that was never demanding.

Then the target moves. Compensation events adjust it, and over the life of a difficult job the cumulative drift of the target through compensation events can quietly exceed the original pain share exposure. An owner watching only the pain and gain position can be losing the contract through a hundred adjustments to the line it is measured against, and never see it, because each event is individually reasonable.

The owner can be losing the contract through a hundred reasonable adjustments to the line it is measured against, and never see it.

The open book is the next place the symmetry breaks. Open book is only as open as the owner is able to read. Defined cost and disallowed cost are a genuine battleground, and an owner without the capability to challenge what lands in each is not auditing the cost. It is receiving it. And under real stress, when the job is going badly and the pain share is biting, the collaborative posture is the first thing to go. Both sides revert to position, and the contract that was meant to align them becomes the contract they fight through.

What it takes from the owner's side

The reinforcement Option C needs is owner-side capability, applied from before the target is set. A rigorous, independently tested target rather than one accepted under time pressure. Disciplined management of compensation events, with the cumulative movement of the target tracked as a headline number, not a footnote. Cost assurance with the competence to actually interrogate defined and disallowed cost rather than nod it through. And the standing to hold the collaborative line when the job gets hard, which comes from capability, not from goodwill.

Option C transfers real protection to a capable owner and very little to an incapable one. The instrument is sound. The alignment it offers is conditional on the owner being able to claim it. That is the part the brochure does not mention, and it is the part that decides the outcome.

Two words get used as if they were interchangeable, and they are not. Change improves the thing you already have. Transformation replaces it with something different in kind. One keeps the operating model and makes it better. The other alters what the organisation fundamentally is, or does, or how it works. The distinction sounds academic. It is one of the most consequential category errors a board can make, because almost everything that follows, the money, the governance, the sponsorship, the timescale, is sized by which of the two you are actually doing.

The failure mode is a mismatch between the words and the resourcing. A board commissions a programme in the language of transformation, ambitious, organisation-wide, redefining how we operate, and then governs and funds it as change, an improvement project with a sponsor who has a day job and a budget set by analogy to the last upgrade. Or the reverse: a straightforward change is dressed in transformation language, over-governed and over-scoped until it collapses under ceremony it never needed.

The consequences are not abstract

Underfunded ambition is the most common. Transformation work priced as change runs out of money at exactly the point the hard part begins, the part where the old way and the new way have to coexist and the organisation is carrying both. Governance mismatch is next: change governance, light and periodic, cannot steer a transformation, while transformation governance smothers a change. Sponsorship mismatch follows the same line. Transformation requires a sponsor with the authority to alter the organisation, not a steering group with the authority to request alterations.

Transformation work priced as change runs out of money at exactly the point the hard part begins.

And the timescale mismatch is the quiet one. Change has a finish. Transformation has an adoption curve that runs long after the programme closes, and a board that books a completion date where there should be a transition period declares victory while the organisation is still mid-change, and then wonders why the benefits never arrive.

How to tell which you are doing

The test is not the size of the budget or the ambition of the slides. It is a single question: when this is done, is the organisation doing the same things better, or different things entirely? If the operating model survives, it is change, however large. If the operating model is replaced, it is transformation, however modest it looks. Ask it plainly, before you commit, and resist the pull of the more impressive word.

Because the word you choose is not a label. It determines the team you need, the money you must find, the governance you must stand up, and the patience the board must hold. Name it wrong and you have designed the failure in at the start, in the first sentence of the business case, long before anyone breaks ground.