Project Recovery: When to Intervene on a Failing Project

Almost nobody intervenes in a failing project too early. The far commoner error is waiting for proof, and proof usually arrives in the form of a missed go-live. This is a guide to the signals that justify acting before that point, and to the three moments when intervention returns the most.

The one signal that matters more than the others

If you read nothing else, read this. The single most reliable indicator that a programme is in trouble is a growing gap between what the reporting says and what the delivery evidence shows.

Reported status is amber or green. Meanwhile defect counts are rising rather than falling, test environments are not ready, the data migration has never completed a clean dry run, and the integration testing that was scheduled for last month has quietly moved. Any one of those in isolation is normal. All of them alongside a green status is not a delivery problem. It is a reporting problem, and reporting problems are structural.

Programmes where the reporting has detached from reality do not self-correct, because the mechanism that would detect the drift is the thing that has failed.

Warning signs worth acting on

In the reporting

• Status has been amber for several consecutive reporting periods with no change in the underlying issues.
• The recovery plan has itself slipped, or has been reissued more than once.
• Milestones are being met on paper by redefining what the milestone meant.
• Nobody can tell you what proportion of spend to date is earned value, meaning work you would actually keep.
• Risks are being closed rather than mitigated, and the risk register has stopped growing.

In the delivery evidence

• Defect discovery rates are still climbing late in the test cycle.
• Environments or test data are chronically unavailable and this is treated as a nuisance rather than a critical path item.
• Scope is being descoped informally, by things quietly not happening, rather than through a decision.
• The systems integrator and the internal team are in dispute about what was contracted.

In the people

• The people who know most about delivery have stopped volunteering bad news in formal forums, though they will tell you privately.
• Key individuals are leaving, or the programme director has resigned.
• Steering committee attendance is being delegated downwards.

The people signals are the most predictive and the most ignored. When experienced delivery staff go quiet, they have usually concluded that raising issues carries more personal cost than staying silent. That conclusion is itself a governance failure.

The three moments when intervention returns most

Recovery is possible at almost any stage, but the return varies enormously depending on when you act.

Before the second re-plan. The first slip is normal. The second re-plan is the moment to intervene, because it is the point at which the original assumptions have been shown to be wrong twice and are about to be used a third time. Acting here is cheap, and options are still open.

Before an externally committed date. If a go-live has been communicated to a regulator, a lender, the market, or customers, the cost of missing it is no longer just delivery cost. Intervene while the date can still be renegotiated on your terms rather than announced as a failure.

At a leadership change. When the programme director leaves, the accumulated informal knowledge leaves with them, and the incoming leader inherits both the problems and the optimistic plan. This is the cleanest possible moment for an independent assessment, because nobody currently in post owns the previous position.

When not to intervene

Formal recovery is disruptive and it should not be the response to every difficulty.

A programme that is behind but where the cause is understood, the team is candid about it, and governance is making decisions at the pace required does not need recovery. It needs support, or possibly more capacity. Imposing a recovery structure on a programme that is already being honestly managed damages morale and slows delivery.

The test is not whether the programme is behind. It is whether the organisation knows why, and whether the mechanism for finding out still works.

What waiting actually costs

Delay is rarely neutral. Every additional month of a programme running on a plan nobody believes consumes budget that produces nothing you would keep, and it narrows the options available later. Descoping is available early. By the time an external date is imminent, the only remaining options are expensive ones: more people, more contingency, or public failure.

There is also a compounding people cost. Teams that have worked hard against an unachievable plan and watched it slip twice become very difficult to remotivate, whoever takes over.

Frequently asked questions

How do I raise this without accusing the delivery team?

Frame it as an independent view rather than a judgement. A structured programme health check is a normal governance instrument, not a disciplinary one, and running it routinely at defined stages removes the implication that commissioning one means somebody has failed.

What if the sponsor does not accept there is a problem?

Then the assessment needs to produce evidence rather than opinion. Earned value analysis and defect trend data are hard to argue with in a way that status opinions are not. If the sponsor still will not accept the position, the issue has moved above the programme and needs to be raised at board level.

Does intervening mean replacing the delivery team?

Usually not. The most common structural causes are governance, contract, and client-side capability rather than the competence of the people doing the work. Replacing the team is expensive, resets the clock, and is frequently the wrong diagnosis.

How long does an assessment take?

A focused diagnosis should take days rather than months, and should not require stopping delivery. If an assessment is going to take a quarter, it has become a programme in its own right and will produce a document rather than a recovery.

The right moment to intervene is earlier than it feels, and the cost of acting slightly too early is far smaller than the cost of acting slightly too late. Once you have decided to act, the eight steps to project recovery set out the order to follow. If you want an independent read on where a programme genuinely stands, Intology's programme recovery consultancy starts with a fixed-price diagnosis before any wider commitment.