Treatment works: the sponsor wants to spend the P80 contingency

Reframe the request before answering it. The gap between P50 and P80 is not spare money; it is the sum that buys confidence in a date that is not negotiable. P50 means a 50% chance of missing a regulatory deadline, which is unacceptable where the consequence is a formal undertaking and a published failure.

What to say to the sponsor

• The confidence level is a risk-appetite decision that the sponsor owns, not the project manager. Ask what confidence the board requires for a hard external commitment. 80-90% is typical, and the answer should be minuted.
• The P80-P50 spread can only be released if the underlying risks are genuinely eliminated or transferred. Nothing in the register has changed, so there is no basis for release.
• Model the cost of lateness (regulatory undertaking, emergency works, reputational and regulatory capital consequences) against the cost of holding contingency. If the cost of being late exceeds the contingency, releasing it is the more expensive option.
• Be honest about the model. A four-month spread between P50 and P80 suggests heavy tails or correlated risks, and the schedule logic may be over-optimised. Present the S-curve, the top risk drivers and the assumptions.
• Offer options rather than a single number: hold at P80; hold at P65 with an accelerated mitigation plan for the top three risks and a stated residual probability; or seek to renegotiate the output date.

Then protect the position. Record the decision, the residual risk, the owner and the accepted probability in the governance papers. If the sponsor still directs the project to plan on P50, confirm the instruction in writing and escalate through assurance, because the project manager will be held to the date regardless.

Treat the offer as an opportunity, not a favour. Log it with an owner, a probability, an impact and a review date, and evaluate it with the same rigour as a threat.

Quantify it

• The planners' six weeks is a single-point estimate. Establish a range and a probability of achieving it, factoring supplier capacity, factory slots, transport and the interface with site works.
• Value the time: if a week of lateness costs £250k in regulatory consequence, six weeks is worth £1.5m against a £900k cost. At 60% probability of delivery the expected benefit is about £0.9m, roughly break-even, so the decision turns on how strongly the schedule risk is correlated with the date.
• Include released site preliminaries and reduced site labour, and the cost of residual site works.

Conditions before committing

• Competitive benchmark or a documented single-source justification for the £900k, and funding drawn from contingency with sponsor approval rather than quietly absorbed.
• Factory acceptance testing with hold points and client witness rights.
• Retained design authority and an interface plan with the site team.
• Transport, lifting and installation method statements, and a warranty that survives the move.
• A decision gate (say, month 4) with a defined fallback to onsite build and a stated exit cost.
• Supplier financial checks and a performance bond proportionate to the £900k.

Governance. The opportunity should appear in board reporting with the downside if it fails, because a six-week saving that fails becomes a six-week delay plus sunk cost.

Risk data decays faster than it is refreshed. Seven of the top twelve risks unreviewed for four months, and three owners gone, means the P80 is built on stale inputs and, more importantly, that nobody is actively driving mitigation. An unowned risk is an unmanaged risk.

What to do

• Reassign every risk to a named individual with the authority and budget to act, not to a role or a team. Confirm acceptance in writing.
• Re-run a workshop with the top twelve owners. Re-score probability and impact against current evidence, and check whether mitigations have actually been implemented rather than merely planned.
• Add an early-warning indicator and a trigger point to each top risk so movement is visible before it becomes an issue.
• Set a review cadence that fits the monthly cycle: full register quarterly, top twelve monthly, event-driven reviews after any change to scope, design or supply chain.
• Ask internal audit or the assurance function to test the register independently. Risk data quality is a governance finding, not an administrative chore.

How to report it. Until the reassessment is complete, present the P80 as indicative with a stated data-quality caveat rather than as a precise number. Show the board the proportion of exposure held by risks reviewed in the last 30 days; that single metric will change behaviour faster than another page of heat map. Then re-run the analysis and reissue the forecast.