Your delivery leader has authority to run a two-year programme. Authority to do what, exactly, within what bounds, until when — and what event ends it?
Most organisations cannot answer that question, and the reason is instructive. Delivery authority is typically granted once, at approval, in a single undifferentiated act: the programme is funded, the leader is appointed, and from that moment the leader is understood to be running it. The grant has a beginning and no defined end short of completion. It has a budget but rarely a movement bound. It lapses, if at all, when somebody loses confidence.
That is an unusual way to delegate. No treasury function would extend a signing authority without a limit and a review date. No board would delegate contracting power without stating what falls outside it. Yet the authority to commit an organisation's scarcest people to a multi-year course of action is routinely handed over as a role rather than as an instrument.
There is a well-worked alternative, and it is twenty years old.
The Strategic Context
The governance architecture described here comes from a comparison of two project management methods written by G.J. Rankins of Goal Professional Services [SOURCE DETAILS REQUIRED — no venue or date is printed on the paper; file metadata indicates 2007 and the filename of a second supplied copy indicates an Australian Institute of Project Management conference, neither of which confirms the citation]. It compares a body of knowledge published in 2004 with a structured method released in 2005. Both have been revised repeatedly since, and one has changed custodianship [FACT CHECK REQUIRED]. Nothing here is offered as a description of any method as it stands today.
What survives the passage of time is the design idea, and Rankins' framing of why the two methods differ is the useful entry point. His account — hedged in his own text as something the differences perhaps reflect — is that the two were answers to different diagnoses. The body of knowledge was compiled at a time when project failure rates were, in his words, much higher than today [FACT CHECK REQUIRED — no supporting data appears in the paper], and it may have been felt that a project manager problem needed addressing: hence its emphasis on equipping the individual with knowledge areas, tools and techniques. The structured method arose later, in a public sector environment that already had trained project managers and was still experiencing significant failures; there, he suggests, it was felt that a governance problem needed addressing.
His conclusion is that the two are complementary rather than competing — the yin and yang of project management, in his phrase, different but not incompatible.
The distinction matters more than the specific methods. An organisation that has equipped its delivery leaders and still experiences failure has a different problem from one that has not equipped them — and the instrument that addresses the second problem is not more training. It is a different structure of authority.
What Leaders Commonly Misread
The first misreading is that authority and accountability are the same delegation. They are not, and the difference is exactly what makes the standard arrangement unstable. Accountability is assigned once and persists. Authority, if it is to function as a control, must be bounded and periodically renewed. Granting the first without structuring the second produces a leader answerable for everything and empowered to commit indefinitely — a mismatch with real consequences, examined in [Related article: Accountability Without Authority: How Organisations Design Delivery Leadership to Fail]. This article is not about that mismatch. It is about the instrument that would bound the grant if anyone used one.
The second misreading is that a budget is a bound. A budget states a total. A bound states how far the holder may move before returning for permission — which is a different quantity, usually much smaller, and applies to schedule and quality as well as cost. An organisation that has set a budget and no movement bound has delegated everything up to the total, and will learn of a problem when the total is threatened rather than when the trajectory changes.
The third misreading is that escalation covers this. Escalation is discretionary and depends on the judgement of the person who would look worst by escalating. A bound is not discretionary: it is a threshold whose breach obliges notification regardless of anyone's assessment of how serious the situation is. The difference between a norm of escalation and a rule of tolerance is the difference between a system that depends on candour and one that does not.
A fourth misreading is that phases and stages are the same thing. They are not, and conflating them removes the instrument entirely. Rankins is precise: a phase relates to project management control by the project manager — a natural division of the work, such as design or construction. A management stage relates to governance control by the governing body. Phases may overlap and their boundaries are at the manager's discretion; management stages may not overlap, and their boundaries are where authority lapses. An organisation whose plan contains phases but no stages has a schedule, not a control structure.
Reframing the Issue
The reframing is to treat delivery authority as a grant with a size, a shape and an expiry, and to specify all three at the point of delegation.
Size is the tolerance: how far the holder may move on schedule, cost, quality, risk and whatever else the organisation chooses to bound. Rankins describes tolerances as giving the manager some freedom to move around schedule or budget while ensuring the governing body remains in control through mandatory notification if a tolerance is breached — the mechanism usually called management by exception.
Shape is the planning horizon. The method he describes sets one at around three months, on the reasoning that beyond it the uncertainty of the future means the value of detailed planning decreases below the cost of developing the plan. The consequence is stepwise refinement: a high-level plan for the whole, and a detailed plan only for the stage in hand, at whatever granularity that stage requires.
Expiry is the part almost nobody replicates. In the architecture Rankins describes, the manager's authority to execute expires in two ways. It expires normally at a management stage boundary, and must be formally renewed. It expires abnormally whenever a stage or the project breaches its allocated tolerances. In both cases the manager must convince the governing body that the project remains on track, or can be brought back on track, before further work is performed.
Read that again with an executive eye. The default state is no authority. Authority is granted for a bounded period against a bounded plan, and it reverts automatically — not by anyone's decision, but by the passage of a boundary or the crossing of a threshold.
Strategic Analysis
Why automatic reversion is the load-bearing idea
Governance instruments generally fail in one direction: they require someone to act in order to intervene. A gate must be convened, a review must be scheduled, a concern must be raised. Each of these depends on a person choosing to expend political capital, and the person best placed to do so is usually the person with most to lose.
Automatic reversion inverts the default. Nobody has to intervene for authority to lapse; somebody has to act for it to continue. The burden of initiative moves from the party who would be embarrassed to the party who wants to proceed.
This is a familiar and well-tested design elsewhere in the enterprise. A credit facility has a review date. A delegated financial authority has a limit and an expiry. A licence must be renewed. In every case the reversion is automatic and the renewal requires a case. Delivery authority is one of the few significant delegations in a large organisation that is commonly granted without either feature.
What a tolerance breach actually obliges
The obligation on breach is notification, not surrender. Rankins is careful here, and the precision matters: the manager must convince the board that the project remains on track or can be brought back on track before further work is performed. That is a narrower test than viability. It asks about trajectory, not about whether the investment still makes sense.
The wider question — whether this initiative should continue to exist at all — belongs to a different instrument, examined in [Related article: What a Stage Gate Is Actually For]. Keeping the two separate is deliberate and useful. A trajectory test can be run quickly, on evidence the delivery team already holds, without convening the people who would decide the investment question. Most breaches are trajectory problems and should be resolved as such. Occasionally a breach reveals that the premise has failed, at which point the second instrument is invoked — but making every breach a viability review would make breaches expensive to declare, which would make them rare, which would defeat the mechanism.
Who else the boundary concerns
A renewal decision is not taken by one person in isolation. The governing body that grants and renews authority has a composition, and whose interests are formally seated in it determines what the renewal test can actually examine — a question taken up in [Related article: Who Sits on the Board for the Benefits?]. A body constituted only of funders and builders will renew on cost and schedule, because that is what it can see.
The planning horizon, and what it protects
The three-month horizon looks like a scheduling convention and is actually a governance one. Detailed plans beyond the horizon cost real effort and will be rewritten; more importantly, they create the impression of commitment to a course of action nobody has yet decided on. A plan detailed to month twenty is a decision about month twenty taken in month one.
A large-scale grid modernisation is a useful case. The utility knows the endpoint and the total envelope. It does not know, in year one, which substations will need replacing rather than refurbishing, because that depends on condition assessments not yet performed. Planning the whole programme in detail at the outset produces a plan that is wrong in a specific, expensive way: it commits crews and outages to work that assessment will later contradict.
Stepwise refinement handles this correctly, and it also creates the natural stage boundaries at which authority lapses — which is the point. The horizon and the tolerance are the same idea applied to time and to movement. One bounds how far ahead you commit; the other bounds how far you may drift before returning.
Why the instrument meets resistance
Bounded, expiring authority takes something away from people who currently hold it unbounded, and it does so immediately and specifically, while its benefits are diffuse and deferred. That is the standard shape of resistance to any change of this kind, examined in [Related article: Why the Old Order Fights Harder], and it should be expected rather than treated as evidence the instrument is wrong.
The cost of the instrument
It is not free. Formal renewal at every stage boundary consumes governing-body time, and an organisation running thirty programmes cannot convene thirty renewal decisions a quarter without the renewals becoming perfunctory — at which point the instrument is worse than useless, because it produces a record of scrutiny that did not occur.
This is a real constraint and it argues for thresholds: full renewal for initiatives above a materiality line, a lighter regime below it. The alternative — applying the full instrument everywhere — reliably degrades into ceremony.
Decision Framework
Six questions, applicable to any initiative above a material threshold.
1. What are the bounds? Not the budget. The permitted movement on cost, on schedule, and on whatever dimension of quality or risk matters here. State them as numbers.
2. What obliges notification? Confirm that breaching a bound requires the holder to notify regardless of their own assessment of severity. If notification is discretionary, there is no tolerance — there is a norm.
3. When does the grant expire? Name the boundary. If the only boundary is completion, the grant runs for the life of the initiative and there is no periodic renewal.
4. What must be shown to renew it? The trajectory test — on track, or credibly recoverable — with evidence specified in advance. Deciding this at the boundary rather than before it guarantees an argument about standards at the worst possible moment.
5. How far ahead is the plan detailed, and why that far? If the answer is "to completion", ask what decisions that plan is silently committing the organisation to.
5a. Is this the right initiative to be renewing at all? Renewal governs one investment already chosen. Whether it should have been chosen over the alternatives is a different question, handled by a different function — see [Related article: Is Your Portfolio Function Selecting, or Supervising?].
6. What is the materiality line? Below which the full instrument does not apply. Without this line the instrument becomes ceremony across the whole portfolio.
Two supporting conventions make it work. Bounds are set by the granting body, not proposed by the holder — a tolerance the holder chose is a tolerance calibrated to what the holder expects to need. And breaches are recorded and counted, because the pattern of breaches across a portfolio is a better capability signal than any maturity assessment.
From Strategy to Execution
Immediate. Take the three largest initiatives and establish, for each, whether any movement bound exists and when the current authority expires. In most organisations both answers are absent, and establishing that takes an afternoon. Where an initiative has been running for more than a year without a formal renewal, that fact is worth putting in front of the board on its own.
Medium term. Add three fields to the standard delegation instrument: tolerance, expiry event, and renewal test. This is a change to an artefact rather than to a process, which makes it durable — an artefact survives the departure of the person who introduced it, and a practice generally does not.
Long term. Decide the materiality line and the governing-body capacity behind it. An organisation that wants renewable authority must budget the governing time to renew, and the honest arithmetic often shows that the current governance structure cannot carry the load — which is itself a finding worth having. Who holds the granting authority in the first place, and at what delegation level, is the prior question, treated in [Related article: Sponsorship Is an Office, Not an Endorsement].
Signals to Monitor
- Initiatives running for years without formal renewal. Track time since the last explicit re-authorisation. A programme in its third year on an original approval is operating on a grant nobody has re-examined.
- Escalations that arrive as narratives rather than as breaches. When a delivery leader raises a concern in prose rather than reporting a threshold crossed, the organisation is relying on candour instead of on a rule.
- Zero recorded breaches. A portfolio in which no tolerance is ever breached has tolerances set too wide to bind, or is not recording.
- Detailed plans extending beyond eighteen months. Worth examining not for accuracy but for what they commit. Ask which decisions in year two the plan has already made.
- Renewal decisions taking under ten minutes. The instrument has become ceremony, and a record of scrutiny that did not occur is more dangerous than no record.
- Bounds proposed by the delivery team and accepted unchanged. A reliable indicator that the tolerance was calibrated to expected need rather than to organisational risk appetite.
Questions for the Leadership Team
- For our largest programme, what movement is the delivery leader permitted before they must come back to us — and can anyone state the number?
- When does their current authority expire, and what event ends it?
- If a tolerance were breached tomorrow, would we hear about it because a rule required it or because someone chose to tell us?
- How far ahead are our major initiatives planned in detail, and what decisions has that detail already made on our behalf?
- How many tolerance breaches has this organisation recorded in the last year — and if the answer is none, what does that tell us?
- Do we have the governing capacity to renew authority at every boundary we would need to set, and if not, where should the materiality line sit?
Closing Perspective
Delegation is one of the few genuinely powerful instruments available to an executive team, and delivery is one of the few places it is used carelessly. Elsewhere in the enterprise, authority is granted in bounded, dated, renewable form as a matter of course. In delivery it is handed over whole, at the start, and reclaimed only in a crisis.
The architecture described here is old, and its specific expressions have been superseded. The idea has not been. Authority that reverts by default, bounded by a stated tolerance and renewed against a stated test, changes who has to act for scrutiny to occur — and in governance, that is nearly always the decisive design choice.
An organisation that adopts nothing else from this could adopt one thing: write down, for every material initiative, how far its leader may move before returning. Most cannot answer that question today, which means the answer is currently as far as they like, until someone notices.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.