Consider an illustrative operating review. The same constraint has appeared on the performance report for several weeks. Operations understands the production effect. Engineering has assessed the technical options. Maintenance knows what an intervention would require. Finance can see the cost.

The discussion is informed. The issue is familiar. The meeting ends with another request to refine the analysis.

The missing element may be the decision itself: exactly what must be decided, who has the authority to decide it, which conditions must be satisfied, and when waiting becomes a decision with consequences.

This matters as energy and infrastructure businesses invest in better data, remote operations and AI. Information can travel across an organisation much faster than authority. When the two are poorly connected, a business can become better informed without becoming more effective.

01Visibility and authority solve different problems

A dashboard makes a condition visible. A decision changes what the organisation will do about it.

That distinction is easy to miss because visibility is tangible. A new report can be demonstrated, a model can be tested, and a data pipeline can be monitored. The authority to balance competing priorities is less visible. It sits in delegated limits, technical responsibilities, budget ownership, contractual obligations and management routines.

Those arrangements determine whether evidence produces action or another hand-off.

Better information remains essential. Teams need to know whether a measurement is current, what it represents, where it came from and how much confidence to place in it. But once the evidence is adequate for the decision in question, another reporting layer may add little. The next improvement may be a clearer decision route.

ISO's technical committee made decision-making an explicit focus of its 2024 update to ISO 55001. Its public guidance describes a framework connecting asset-management decisions with value, supported by defined criteria, methods, processes and tools. That is a useful starting point: establish how decisions will be made as well as what information will support them. [1]

02A shared problem can still have divided authority

Take a recurring equipment constraint. “Resolve the reliability problem” sounds like one action. In practice, it may contain several decisions:

These decisions may properly belong to different people. Their evidence requirements and timescales may also differ. A competent technical assessment does not automatically authorise expenditure. Budget approval does not establish that an operating change is technically acceptable.

The problem arises when those distinctions remain implicit. An action owner is nominated, but cannot release the resources. A committee considers the issue, but nobody can say which decision the committee is being asked to make. A decision is deferred, but the conditions for returning to it are not recorded.

Clarity does not mean concentrating every approval in one person. It means making the sequence visible, giving each decision an accountable owner, and preserving the technical, safety and financial controls that apply to it.

In the UK Health and Safety Executive's guidance on organisational change, changes to roles, responsibilities, staffing and outsourcing warrant systematic assessment. Its principles include ensuring that key tasks and responsibilities transfer successfully and that people receive support for changed roles. Moving a decision from a site team to a central function therefore deserves more attention than changing a box on an organisation chart. [2]

03Local targets can make delay look reasonable

The operating team is measured on output. Maintenance is trying to protect a planned work programme. Engineering needs confidence in the proposed remedy. Procurement is managing commercial exposure. Finance is controlling expenditure.

Each position can be reasonable within its own remit. The combined result may still be an unresolved asset problem.

The practical question is how the business will judge the whole decision. Which constraints are non-negotiable? Which consequences should be compared? What can be decided within existing authority, and what requires escalation?

Safety and mandatory technical limits should not be traded away in a financial calculation. Within those boundaries, the decision may involve production loss, service reliability, intervention cost, resource availability, contractual commitments and future exposure. These need to appear together in the decision record.

Otherwise the easiest number to see can dominate. The intervention cost appears in this month's budget. The cost of continuing with the constraint is distributed across several teams and reporting periods. An apparently disciplined spending decision can then preserve a more expensive operating problem.

An explicit decision to wait can be entirely sensible. It needs an owner, a rationale, any authorised interim controls, and a date or condition for reconsideration. “Awaiting more information” is incomplete unless the organisation can identify the information, its decision value and who will obtain it.

04Start with a real decision and follow its path

Between 2011 and 2015, I led collaborative-working process implementation across Shell assets in Africa, Kazakhstan and ONEgas, with remote support to New Orleans. The work centred on ways of working rather than new equipment.

That experience informs the question I would bring to a digital investment today: how will this change the way people reach, authorise and carry out a decision?

A useful review starts with a recent operational issue. Follow it from the first signal through validation, assessment, approval, execution and confirmation of the result. Compare the documented process with what the people involved actually had to do.

Find where elapsed time accumulated. Was evidence unavailable? Was it disputed? Did the work wait for a specialist, a budget decision, a contract change or a meeting? Did an approval depend on a person who had no cover while absent?

Different causes require different responses. Poor evidence may need a data correction. An overloaded specialist may need capacity. A recurring approval delay may need clearer delegation. A constraint created by a contract or operating model may need a larger change. Calling all of them “digital maturity” makes it harder to choose the right intervention.

05Write the decision so somebody can use it

A short decision record should allow an authorised person to understand the choice without reconstructing the entire history. The following is a practical format, not a replacement for formal engineering, safety or investment approvals.

Field What it must make clear
Decision required The specific choice, separated from the broader problem.
Evidence and uncertainty What is known, the source and date, what remains uncertain, and whether that uncertainty could change the choice.
Options and consequences Feasible options, including a controlled deferral where permitted, assessed against the same criteria.
Authority and assurance Who decides, within what limits, and which specialist reviews or approvals are required.
Timing and escalation When the decision is needed, why that timing matters, and where it goes if unresolved.
Execution and outcome Who will carry it out, what resources are committed, and how the result will be checked.

For the illustrative equipment constraint, replace “Engineering to investigate” with a specific question: “Do we authorise inspection during the next approved access window, and who releases the resources?”

The relevant technical authority confirms the inspection requirements and applicable controls. The manager with the necessary delegated authority decides whether to commit the resources. The work coordinator then owns execution. The record names those people, the decision date, and the route for resolving disagreement. Any subsequent modification requires its own decision and approvals. One named action owner no longer has to bridge several undeclared decisions alone.

The aim is a usable record, not additional paperwork. Put it where the work is already managed. If the same facts must be re-entered in a slide deck, spreadsheet, work order and approval form, the administrative burden can obscure the decision again.

06Measure waiting, then verify the outcome

An open-action count does not explain why a material decision remains unresolved, or whether closing the action improved the asset's performance.

A useful review separates time spent obtaining necessary evidence from time spent waiting for authority or resources. It also distinguishes a decision made from an action completed and an outcome achieved.

Start with a small sample of consequential decisions and examine:

These measures need context. A quick, poorly supported decision is not an improvement. Neither is a low backlog produced by closing records before the underlying problem is resolved. Compare decisions of similar consequence and complexity, and examine the quality of the reasoning alongside elapsed time.

HSE's offshore health guidance makes a relevant management point: monitoring information should reach people with the authority to take remedial action, and review should assess whether the arrangements are effective. My broader application is to connect operational information with authority and verified action. [3]

07AI increases the need for clear decision rights

AI can help assemble evidence, identify patterns, compare options and prepare a recommendation. It may reduce part of the analytical workload. It cannot establish an organisation's legitimate authority simply by producing a confident answer.

Before embedding a model in an operating workflow, define what it may recommend or execute, what must be reviewed, who can challenge or override it, and how that intervention is recorded. The control needs to match the consequence of the decision and the organisation's approved arrangements.

NIST's AI Risk Management Framework 1.0 places responsibility with people and organisations. Its governance provisions address clear roles and communication, executive responsibility for AI risk decisions, and differentiated responsibilities for human oversight. The framework is voluntary guidance; its relevance here is the discipline of making accountability explicit. [4]

Putting “human in the loop” on a diagram does not establish that the person has the competence, time, information or authority to exercise meaningful oversight. Those conditions need to be designed and tested.

08Test the decision route before the next investment

For an asset owner, the next step can be modest. Select one recurring, material decision. Trace how it is made today with the people who do the work. Identify the missing evidence, unclear authority and competing objectives. Agree the change, test it within the existing control framework, and check what happens to the outcome.

For an investor, the same exercise is a useful diligence question. When an asset underperforms, can management demonstrate how an operating issue becomes an authorised intervention, with resources and a measured result? An investment case depends on that capacity to respond as well as on the original technical assumptions.

The value of operational intelligence is realised when evidence changes what the organisation does. A decision route that people can use gives better data somewhere to go.

Sources

  1. ISO/TC 251, public guidance on ISO 55001:2024, particularly clause 4.5. This article draws on the committee's public summary, not a reproduction of the full standard.
  2. Health and Safety Executive, Organisational change.
  3. Health and Safety Executive, Managing offshore health risks, measuring performance and learning from experience.
  4. NIST, AI Risk Management Framework 1.0, Core, GOVERN 2 and 3.2.

The scenarios are illustrative. The decision-record format and wider commercial interpretation are the author's analysis.