Crisis Delivery Worked Without the Ceremony: The Case for Decision-Led Programme Governance
People moved quickly because purpose, authority, evidence and consequence were briefly aligned.
Executive Summary
By late April 2020, organisations have delivered operational changes at a pace that their formal transformation programmes had repeatedly declared impossible. Workforces have been dispersed, digital routes have replaced physical ones, supplier arrangements have been altered, capacity has been redirected and services have been redesigned in weeks. Much of this has happened with fewer boards, shorter papers and more direct authority than normal programme practice allows.
The tempting conclusion is that governance was unnecessary. That is too simple. Emergency delivery has worked because governance did not disappear; it changed form. Purpose became singular, decision rights became visible, specialists advised together, evidence came from live operations and delay acquired an explicit cost. Ceremony fell away while decision-making intensified.
The danger now lies in choosing the wrong lesson. One option is to restore the previous approval architecture in the name of control. A second is to preserve emergency command indefinitely and normalise fatigue, shortcuts and concentrated authority. Neither is defensible. The recommended course is a decision-led programme governance model that retains the clarity of crisis delivery while restoring proportionate challenge, sustainability and accountability.
This paper makes the case for that model. It distinguishes control from ceremony, sets out evidence from a representative twelve-week mobilisation, compares the available governance choices and recommends five design principles: one governing outcome; authority located with consequence; concurrent rather than sequential assurance; operational evidence as the primary report; and every exception carrying an owner and expiry.
The aim is not faster governance for its own sake. It is governance that converts uncertainty into timely, accountable decisions.
The Evidence in Front of Us
At the beginning of February, a large service organisation had a three-year programme to move essential customer interactions away from branches and paper. The programme had completed discovery, produced a target operating model and approved a technology direction. It had not released a usable end-to-end service. The board pack showed 114 requirements, 27 dependencies, 41 open risks and a delivery date under review.
By the middle of March, the operating assumption had collapsed. Physical capacity was sharply constrained, demand patterns were changing daily and several thousand employees needed to work away from their normal locations. The existing programme plan could no longer answer the immediate question: how will essential cases enter, move and complete next week?
A temporary decision group was formed with operations, technology, finance, risk, information security, commercial and customer-service authority present. It met for 45 minutes each morning. The group did not approve every design choice. It decided the outcome, limits and exceptions within which the delivery team could act.
The first digital route went live after eleven days. It covered four case types rather than the planned nineteen. A manual reconciliation stood behind one incomplete interface. A daily sample of 200 cases checked identity, routing and completion. By week six, the route handled 64 per cent of new demand. Average intake time fell from nine minutes to four, although the back office still required additional manual effort. By week twelve, fourteen case types were operating and the unreconciled error rate had fallen from 3.8 per cent to 0.7 per cent.
The numbers do not prove that every prior control was waste. They show that the organisation could govern material risk without applying every control to every decision in the same way.
The practical differences were observable:
- The objective was service continuity, not a negotiated bundle of benefits.
- One executive could accept operational trade-offs within a stated financial and risk limit.
- Risk and security advice entered design discussions before decisions, not after documents.
- Scope was sequenced according to consequence.
- Live service measures replaced narrative confidence.
- Temporary compromises were recorded with owners and review dates.
- Decisions that exceeded the delegated perimeter escalated within hours.
This was not governance-free delivery. It was governance stripped to its decision-making purpose.
When ceremony disappears and decisions become clearer, the evidence points to a governance design problem — not to the absence of a need for governance.
How Normal Governance Became an Approval Industry
Programme governance usually grows by accretion. A control is added after an incident. A forum is created to give a function visibility. A template expands so a concern cannot be said to have been omitted. A supplier gate protects a contractual boundary. A financial threshold reflects a previous failure.
Each addition is defensible in isolation. Together they create an approval industry in which information travels more readily than authority.
Three mechanisms cause the drag.
Decisions are hidden inside documents
A programme paper may contain background, status, options, financial data, risks, dependencies and recommendations. Yet the exact decision is often vague. Board members discuss the narrative and ask for more detail; the programme records an action to return next month.
The failure is not insufficient information. It is the absence of a decision contract: who decides, by when, against which criteria and with what consequence if no decision is made.
During crisis delivery, the decision is difficult to hide. “Release the four case types on Monday with a daily reconciliation” can be accepted, rejected or amended. The clarity is created by proximity to consequence.
Assurance is arranged sequentially
A design moves through architecture, security, risk, legal, finance, procurement and operations. Each function receives it at a different stage, often with insufficient context and little authority to resolve conflicts with another function. A two-day judgement becomes a six-week journey through calendars.
Sequential assurance also produces late objections. By the time a specialist sees the design, investment and identity are attached to it. Challenge becomes expensive, defensive and political.
Emergency teams have shown that assurance can remain independent in judgement while becoming concurrent in time. Specialists need not surrender their duties to participate in the same decision window.
Governance measures compliance with itself
Boards often receive milestone status, action closure, budget variance and risk counts. These measures describe the programme apparatus. They do not necessarily reveal whether the service works, whether customers are harmed or whether the operating burden is sustainable.
A programme can be green while 12 per cent of cases require manual repair. Conversely, a crisis response may be amber against the original plan while successfully preserving the essential outcome. Governance fails when adherence to plan becomes stronger evidence than reality.
The Serious Case for Restoring the Old Controls
There is a compelling argument for returning to normal governance as soon as possible.
Emergency arrangements concentrate authority and rely on exceptional effort. They may defer accessibility, data quality, contractual discipline, employee consultation and long-term architecture. Daily senior attention cannot be sustained. Temporary manual controls are vulnerable to error. Decisions made under pressure may privilege visible continuity over less visible harms. Staff fatigue can itself become an operational risk.
The absence of immediate failure is not proof that a control was unnecessary. Some consequences appear later: duplicated records, weak access discipline, unmanageable supplier terms, technical fragility and customers excluded by a rapidly chosen route.
This argument is correct about the risks and wrong about the remedy. Reinstating every previous forum and sequence would confuse familiarity with control. The relevant question is not whether scrutiny should return. It is which scrutiny changes a decision, when it must enter and who remains accountable afterwards.
A control earns its place when it does at least one of four things:
- Prevents an unacceptable decision.
- Detects a harmful outcome early enough to act.
- Makes authority and residual risk visible.
- Produces evidence required to operate, assure or explain the service.
A control that merely repeats information, offers advice after commitment or creates collective attendance without accountable judgement does not become valuable because the emergency has passed.
The Three Governance Choices
Organisations emerging from the initial mobilisation face three broad choices.
| Choice | Strength | Principal weakness | Likely result |
|---|---|---|---|
| Restore the prior model | Familiar roles, full documentation, established assurance | Recreates sequential approval and diluted authority | Control appears stronger while decisions slow |
| Continue emergency command | Fast escalation, concentrated authority, outcome focus | Unsustainable effort, weak renewal, deferred obligations | Speed persists briefly, then risk and fatigue accumulate |
| Adopt decision-led governance | Retains clear authority and operational evidence with proportionate challenge | Requires explicit trade-offs and redesign of forums | Sustainable pace with visible accountability |
Restore the prior model
This choice feels prudent. It is also the easiest administratively because it requires no redesign. Existing committees resume, templates return and temporary delegations expire.
Its weakness is that it treats the crisis experience as an exception rather than evidence. If a decision was safely made in 45 minutes with the necessary expertise present, restoring a monthly chain of approvals requires a positive justification. “That is our process” is not one.
Continue emergency command
This choice appeals to leaders who have seen movement and fear losing it. It retains daily meetings, senior intervention and wide temporary authority.
Its weakness is institutional and human. Command works when the outcome is singular and time is compressed. Normal operations contain multiple legitimate outcomes, and enduring capability cannot depend on executives resolving routine questions or teams absorbing unlimited effort. Emergency command also risks normalising exceptions before their consequences are understood.
Adopt decision-led governance
This choice treats the crisis as a test of governance design. It preserves the mechanisms that improved decisions while retiring the conditions that depended on emergency.
It is the only option that addresses both speed and control. It is also the hardest because it requires leaders to remove forums, define authority and accept that not every interest can hold a veto.
The Recommended Model
Decision-led governance begins with a simple proposition: programmes exist to produce outcomes through a sequence of consequential choices. Governance should therefore be designed around those choices, not around the production of status.
One governing outcome
Every programme should state the outcome that governs trade-offs for the next decision horizon. This is not a slogan or a list of benefits. It is a measurable consequence, owned by an executive, with a date and population.
For the twelve-week service mobilisation, the governing outcome was: essential cases can be received, routed and completed while physical operations remain constrained. Cost reduction and full process standardisation were relevant but subordinate.
The outcome can change as conditions change. What cannot persist is a programme with several supposedly equal priorities and no rule for conflict.
Authority located with consequence
The accountable decision-maker must have authority over the principal consequence. If the decision affects service continuity, the service owner cannot merely advise a technology board. If it creates material financial exposure, finance must participate, but financial control should not silently become ownership of the service decision.
Delegation should be written in operational terms:
- Decision class
- Financial and risk limit
- Mandatory advisers
- Time limit
- Escalation trigger
- Required evidence
For example, a delivery director may approve a release affecting up to 20,000 cases a week when identity controls pass, error remains below one per cent, no new category of sensitive data is introduced and a rollback route exists. Outside those limits, the accountable executive decides.
Concurrent assurance
Bring the required specialists into the design and decision window. Preserve independence by recording each function’s advice and any unresolved objection. Independence does not require delay; it requires the ability to state a judgement without pressure to disguise it.
A useful pattern has three points:
- Frame: specialists help define the constraints before design begins.
- Challenge: they test evidence before the material decision.
- Verify: they confirm that implementation matches the decision and that outcomes remain within limits.
This replaces serial sign-off with continuous, explicit challenge.
Operational evidence as the primary report
The governance pack should begin with outcome evidence, not a narrative status.
For a live service, the first page might show:
- Demand received and successfully completed
- Elapsed time
- Error and rework
- Unmet or excluded demand
- Customer detriment
- Manual effort
- Control exceptions
- Decisions required in the next seven days
Milestones, budget and risks remain necessary. They follow the operational truth rather than obscuring it.
Exceptions with owners and expiry
Emergency delivery depends on temporary arrangements. Decision-led governance makes temporary status enforceable.
Every exception must contain the rule being varied, reason, affected scope, compensating control, accountable owner, expiry date and exit test. If an exception reaches expiry without resolution, the governing body must renew it explicitly, redesign the rule or stop the affected activity.
A “temporary” manual reconciliation that has run for three months without a renewal decision is not a control. It is an unacknowledged operating model.
From Committees to Decision Architecture
The recommended model does not require abolishing every board. It requires each forum to justify itself through decisions.
A programme should map its recurring decisions and assign them to four levels:
| Level | Typical decision | Cadence | Evidence |
|---|---|---|---|
| Delivery | Routine design and sequencing within delegation | Daily or as required | Working product, control result, operational measure |
| Programme | Cross-team trade-off, material exception, release decision | Weekly or event-driven | Options, consequence, specialist advice |
| Executive | Outcome, major funding, risk acceptance, scope reset | Monthly or event-driven | Outcome evidence, decision record, residual exposure |
| Independent assurance | Effectiveness of governance and controls | Periodic and risk-based | Samples, tests, findings and management response |
Any forum that cannot name its decision class should be challenged. It may be a useful information exchange, but it should not sit on the critical path under the label of governance.
Decision records should be short. A reliable record states:
- The question
- Options genuinely considered
- Evidence and advice
- Decision and accountable person
- Conditions or exceptions
- Implementation owner
- Review trigger
The record is not a substitute for judgement. It makes judgement visible.
Implementation in Ninety Days
The model can be established without a wholesale methodology replacement.
Days one to fifteen: expose the current path
Select three recent material decisions. Trace the elapsed time from identification to action. Record every forum, paper, hand-off, repeated question and period of waiting. Separate time spent producing evidence from time spent waiting for authority.
Create a heat map of delay and consequence. The purpose is not to criticise functions; it is to reveal where the design prevents timely judgement.
Days sixteen to thirty: define the decision perimeter
State the governing outcome for each priority programme. Identify the ten to fifteen recurring material decisions. Assign one accountable owner and specify mandatory advice, limits and escalation.
Close obvious gaps. Where three boards claim part of the same decision, require sponsors to name the final authority.
Days thirty-one to sixty: run the model in parallel
For one programme, convene a weekly decision session with the required advisers present. Use operational evidence and short decision records. Continue any mandatory existing governance until the new model proves reliable.
Measure decision elapsed time, rework, late objections, exception age and outcome performance.
Days sixty-one to ninety: retire and renew
Compare the evidence. Remove duplicate forums and serial approvals where the decision-led route has provided equal or stronger control. Formalise delegations. Set an assurance plan and review dates.
Do not declare success because meeting hours fall. The test is whether decisions are faster, consequences are visible and control failures are found earlier.
Guardrails Against Overreach
A crisis can make speed feel morally superior. It is not. Some decisions should remain slow because they are difficult to reverse, distribute harm unevenly or commit the organisation for years.
Decision-led governance therefore requires explicit brakes:
- Material safety, legal or regulatory concerns can trigger independent escalation.
- A specialist objection cannot be deleted from the record by majority preference.
- Decisions affecting excluded or vulnerable groups require evidence beyond aggregate service volume.
- Temporary authority expires unless renewed.
- Major supplier commitments and irreversible architecture choices receive proportionate long-term scrutiny.
- Staff capacity and fatigue are reported as operational risks.
- Post-implementation verification is mandatory for material releases.
These guardrails prevent the model from becoming an excuse for executive impatience.
“The test of lean governance is not how little process remains; it is how clearly the remaining process changes a consequential decision.”
Measures for the New Model
A governance redesign should be judged through a balanced set of measures.
Decision performance
- Median time from decision request to accountable judgement
- Percentage decided within delegated authority
- Number returned because the question or evidence was unclear
- Late specialist objections
Outcome performance
- Service completion, timeliness, error, customer harm and unmet demand
- Benefit movement attributable to implemented decisions
- Operational effort transferred into manual work
Control performance
- Exceptions past expiry
- Repeated control failures
- Decisions without mandatory advice
- Verification completed after release
Sustainability
- Critical-role concentration
- Overtime and fatigue indicators
- Manual control volume
- Deferred remediation value and age
No single measure is sufficient. Faster decisions with rising harm are failure; stronger documentation with unchanged decision delay is also failure.
Recommendation
Organisations should not return automatically to the governance model they held in February, nor should they preserve emergency command as a permanent virtue. They should use the next ninety days to install decision-led governance across their most consequential programmes.
The recommendation has five commitments:
- Replace multi-purpose programme objectives with one governing outcome for each decision horizon.
- Give each material decision one accountable owner, bounded delegation and explicit escalation.
- Move assurance from sequential gates into the decision window while preserving independent judgement.
- Put operational evidence before programme narrative.
- Give every exception an owner, compensating control and expiry.
The crisis has not proved that governance can be abandoned. It has provided unusually strong evidence that much of what passed for governance was ceremony around unresolved authority.
If the old approval chains return untouched, organisations will have treated extraordinary delivery as a story about heroic people. The more useful conclusion is structural. People moved quickly because purpose, authority, evidence and consequence were briefly aligned.
That alignment should not be reserved for emergencies.