The short version: work management runs in five phases. Work approval, planning, scheduling, resourcing, then execution. The order surprises people, because most operations resource before they schedule, and that single inversion is behind a large share of the jobs that arrive at the face without a part. You cannot resource to a date that does not exist yet. Each phase hands something specific to the next one, and the handovers are where the work actually breaks.
Somebody asks what work management is and the usual answer is a diagram. Five boxes, five arrows, a loop back to the start. Everyone nods, and nothing changes on Monday.
The diagram is accurate and it is also useless on its own, because it shows the sequence and hides the handovers. The handovers are the whole discipline.
What are the five phases of work management?
In the order they occur:
| Phase | The question it answers | What it hands to the next phase |
|---|---|---|
| Work approval | Should we do this work at all? | An approved request with a cost ceiling |
| Planning | How will it be done, and with what? | A work package and a duration |
| Scheduling | Which week, and in what sequence? | A dated, committed schedule |
| Resourcing | Is everything on site for that date? | A job that is ready to execute |
| Work execution | Do the work, record what happened | Confirmations, history, and a closed loop |
In the Anglo American Operating Model these four sit under the Work Management section of the AS&R framework as Resourcing, Plan Work, Schedule Work and Execute Work. Work approval sits in front of them, and it is the gate everything else depends on.
Why does resourcing come after scheduling?
This is the one people argue about, and it is worth the argument.
The intuition says gather the parts first, then schedule the job once you have them. Run it that way and you get a store full of materials reserved against jobs with no date, while the jobs that do have dates wait for parts that were bought for something else.
Scheduling first gives resourcing a target. The scheduler commits the job to a week. That date is what the storeman, the buyer and the contractor coordinator work towards, and it is what makes a late delivery visible as a problem rather than as background noise.
It also creates the escalation that keeps the schedule honest. When a promised delivery date falls after the required date, that is a critical issue and it goes to the scheduler immediately. Not at the next weekly meeting. The schedule can absorb that news on Tuesday and cannot absorb it on Friday afternoon.
What has to be true before work is approved?
Approval is two acts, and most operations only perform the first.
The first line manager sets the user status to APD. The same manager enters an Upper Cost Limit. That limit bounds what may be spent, it gives the planned cost something to be tested against, and it gives the traffic light on the Task Status Report the rule it applies.
Before that, the request itself has to be worth approving. A quality work request names the object, the symptom, the location and something observable: how fast it is leaking, when it started. The test is simple. Can the planner plan this job without phoning the person who raised it?
Two fields carry more weight than their size suggests. The functional location is the spine of the asset history, and a wrong one records the failure against the wrong asset and corrupts the reliability picture permanently. The breakdown indicator drives downtime reporting, so ticking it for work that is urgent while the asset still runs corrupts availability data that somebody will make a capital decision on later.
What does planning actually produce?
A work package. Everything the artisan needs to do the job without going looking for anything: the instructions, the labour, the materials, the tools and the documents.
That definition is a high bar and it is deliberate. If the artisan has to walk to the store, the job is not planned. The same goes for a missing drawing. Wrench time studies keep rediscovering the same thing, which is that the hours lost to work are mostly lost to fetching rather than to fixing.
Planning also produces a duration estimate, and that estimate is only as good as the feedback loop behind it. A task planned for four hours that takes nine must be confirmed at nine, with a reason for the difference. Confirming the planned figure to make the plan look accurate destroys the only data the planner has for estimating that job next time.
How does the loop close?
Four steps, in this order, and the order matters.
- Execute the work.
- Capture the confirmation with actual time.
- Complete the notification with full feedback and malfunction data.
- Technically complete the order, TECO.
TECO comes last because it clears reservations and requisitions. Do it early and you release material the job may still need, then somebody raises a second notification for work that is already half done.
Step three is where the loop either closes or springs a leak. A job completed with “job done” in the feedback field has recorded nothing. The malfunction data is what reliability engineering reads to decide whether a maintenance strategy should change, and reliability engineering sits in Analyse and Improve, which is the section that feeds next year’s plan.
That is what makes this a cycle rather than a pipeline. The feedback an artisan captures on a Thursday afternoon is an input to a strategy decision months away.
Where do these five phases usually break?
At the seams, almost always.
- Approval to planning. The request is thin, so the planner starts the job by conducting an investigation that should have taken the originator ninety seconds.
- Planning to scheduling. The job is planned in name only, with no materials check, so it enters the schedule as a job that cannot run.
- Scheduling to resourcing. The date is set and nobody is accountable for the parts arriving against it.
- Resourcing to execution. The work package is incomplete and the artisan discovers it at the workplace, which is the most expensive place in the operation to discover anything.
- Execution back to approval. Feedback is thin, so the reliability loop starves and the same failure returns next quarter as a fresh surprise.
Each of those is a handover between two people who report to different managers. That is why work management improvement is an operating model problem rather than a training problem, and it is why fixing it one role at a time rarely holds.
DBiLL works on work management inside operations that run SAP PM, and on the capability building that keeps the cycle running after we leave. To talk through where your own handovers break, book a free delivery review.
Related reading: why schedule compliance can read high on a reactive week, and why improvement programmes stop working after six months.
Frequently asked questions
What are the five phases of work management?
Work approval, planning, scheduling, resourcing and work execution, in that order. Approval decides whether the work is justified, planning decides how it will be done, scheduling commits it to a period, resourcing gets everything to site for that date, and execution does the work and records what happened.
Why is resourcing after scheduling and not before?
Because you cannot resource to a date that does not exist yet. Scheduling commits the job to a week, and that commitment is what the store, the buyer and the contractor coordinator work towards. Resourcing first produces materials reserved against undated jobs.
What is the difference between a work request and a work order?
A work request, a notification in SAP PM, is the demand. It says something needs attention. A work order is the authorised instruction to do it, created once the request is approved. Work only appears on the Task Status Report once the notification has become an order.
What is an Upper Cost Limit and who sets it?
The cost ceiling authorised at approval, set by the first line manager at the same time as the APD status. Planned cost is then tested against it, which is what drives the traffic light on the Task Status Report.
What makes a good work request description?
Name the object, the symptom, the location and an observable measure such as the rate of a leak or when it started. The test is whether the planner can plan the job without phoning the originator.
What is a work package in the Operating Model?
The complete set of instructions, labour, materials, tools and documents required to execute a task. The standard is that the artisan can do the job without going looking for anything.
In what order should a job be closed out?
Execute, confirm with actual time, complete the notification with full feedback and malfunction data, then TECO. TECO goes last because it clears reservations and requisitions that the job may still need.
Is work management the same as maintenance planning?
Maintenance planning is one phase inside work management. Work management is the whole cycle from the approval of demand through to the feedback that closes the loop, and planning is the phase that turns approved demand into an executable work package.