Turnaround Work Sampling Should Not End With a Pie Chart

It is the morning after a major turnaround shift, and the project-controls report is on the screen.

Planned hours, earned hours, and actual hours—often called burned hours—show that execution is behind plan.

The team can see the variance. It still cannot see exactly where productive time was lost at the workface—or which conditions created that loss.

That is not a failure of project controls. They are doing their job: integrating scope, schedule, and cost so leaders can quantify progress and performance against the baseline.

The trouble starts when the variance is treated as the diagnosis.

Project controls show that performance moved away from plan. They do not, by themselves, show which workface conditions produced the loss.

Most turnarounds do not use work sampling as their primary productivity measure. But work sampling can be a powerful diagnostic complement because it makes the distribution of work, support, travel, waiting, and other loss visible.

Used well, it helps leaders move from “we are behind” to “this is where the loss is appearing, and these are the conditions we need to test.”

Work sampling does not replace project controls, and it does not prove root cause by itself. It points the investigation toward the operating conditions behind the variance.

That is the difference between reporting turnaround performance and learning how to improve it.

What project controls can—and cannot—tell us

Planned, earned, and actual or burned hours are essential turnaround controls.

Used properly, they show whether execution is producing the planned progress for the hours consumed. They quantify performance variance, expose trends, and support forecasting and corrective action.¹

But the same unfavorable variance can come from very different operating conditions: permit delays, unavailable material, incomplete isolation, blocked access, late engineering decisions, rework, congestion, poor sequencing, or a crew-level execution issue.

Project controls tell leaders that the plan and the result have separated. They do not automatically show where the lost time accumulated or which mechanism produced it.

That distinction matters. A variance is a signal to investigate—not a root-cause statement.

What work sampling adds—and what it still cannot tell us

Work sampling estimates how observed time is distributed across clearly defined activities using observations taken across a study period. Randomized observation timing and adequate coverage matter because the method is statistical, not continuous.³

It adds a different view from project controls: where and how often observed time appears as direct work, support work, travel, waiting, material handling, rework, or another defined category.

That helps focus attention. It can reveal whether the productivity gap is concentrated in a particular shift, area, craft, phase, or operating condition—when the study design and sample support the comparison.²

But the chart does not reconstruct every event that occurred between observations. It does not automatically tell us whether waiting came from a permit delay, unavailable material, incomplete isolation, late engineering clarification, poor work packaging, blocked access, sequence conflict, or a decision that was not escalated.

Those conditions can produce the same visible category.

When leaders treat the category as the cause, the response becomes predictable: work harder, supervise more, retrain the crew, or increase headcount.

Sometimes one of those actions may help. Often it simply acts on the most visible part of the system.

That creates three practical rules.

1. Define the categories before collecting the data.

   If one observer records “waiting” only when a crew is inactive, while another includes searching, staging, or clarification, the percentages will look precise but mean different things.

2. Make coverage and uncertainty visible.

   A thin sample from one shift may be a useful directional signal, but it should not become a precise verdict on an individual crew or contractor.

3. Use the result to focus the next investigation.

   Work sampling helps identify where attention is needed. It does not remove the need to go to the work and test the conditions around the loss.

In other words, the project-control variance is one level of evidence. The work-sampling pattern is another. Neither is the whole evidence chain.

Waiting is an observation, not a root cause

This is the distinction I want leaders to remember.

Waiting is an observation, not a root cause.

A crew may be waiting because the work was not executable. Or because a constraint that was known yesterday remained unresolved today. Or because the interface between operations, maintenance, inspection, engineering, materials, and the contractor did not produce a clear next move.

The visible loss occurs at the workface. The condition that created it may sit several handoffs away.

That does not mean every delay is systemic. A local breakdown, an isolated equipment issue, or an individual execution deviation may genuinely require a focused local response.

The point is to test the boundary before prescribing the remedy.

Audit the conditions around the loss

Once the pattern is visible, the next step is not a bigger pie chart. It is a targeted evidence review.

I use the word audit here in its practical management-system sense: gather and evaluate evidence against defined expectations. An audit does not prove causality by itself. It helps determine whether required conditions, controls, and interfaces were present, effective, and followed.⁴

For a turnaround delay pattern, the review might test:

Advanced Work Packaging and Installation Work Package release practices reinforce the same principle: work should be deliberately scoped, sequenced, and cleared of constraints before it reaches the crew.⁵

The audit turns a broad category into testable operating conditions.

Classify the problem before prescribing the response

Not every finding deserves the same response.

Classification prevents a common failure: using a local corrective action on a system interaction—or launching a system redesign for a one-time deviation.

Use two clocks for turnaround control

Project controls provide the baseline and performance view. Once a loss pattern needs diagnosis, one important design choice is to separate pattern measurement from same-shift control.

Work sampling operates on a pattern clock. It needs enough observations and coverage to estimate how time is distributed. It becomes stronger as the study accumulates evidence.

Turnaround control operates on a faster clock. Leaders need to know today whether the next work package is ready, whether a constraint is aging, whether the crew has a clear executable task, and whether a decision is stuck.

Trying to make a thin daily sample perform both jobs creates false precision. Ignoring the daily signals until the final study is complete creates a post-event report.

The better design connects both clocks.

Measure the Work. Audit the System. Improve Both.

Bring the evidence into turnaround daily management

The framework becomes useful when it changes the daily conversation.

A good turnaround review should not spend most of its time explaining yesterday’s earned-hour variance or the latest chart. It should use the evidence to decide what changes next.

For each material performance gap, ask:

1. What does project control signal?

   State the planned, earned, and actual or burned-hour variance—and whether the trend threatens cost, schedule, or the critical path.

2. Where is the loss appearing?

   Use the work-sampling pattern and its coverage without overstating precision.

3. What condition are we testing?

   Identify the readiness, flow, interface, control, or execution condition that may explain the pattern.

4. How is the problem classified?

   Local deviation, recurring weakness, system interaction, or emerging risk?

5. What changes this shift?

   Name the owner, due shift, escalation path, and expected evidence.

6. How will we verify?

   Check the condition first, then watch whether both the loss pattern and the project-control trend change with enough evidence.

This is how project control, work observation, and operating-system learning become one management conversation.

Verify changed conditions and changed patterns

An action is not effective because it was assigned, closed, or presented at the next meeting.

Verification needs two levels.

The first check tells us whether the intended condition changed. The second tells us whether the operating result moved in the expected direction.

Neither should be oversold. A better pattern after one action does not automatically prove a single causal mechanism. But together, condition evidence and repeated observation create a much stronger basis for learning than action completion alone.

Measurement should start the management conversation—not end it

A turnaround needs project controls. They tell leaders whether execution is delivering planned progress for the hours consumed.

Work sampling adds a different diagnostic view: how observed time is being used at the workface and where loss is concentrated.

But neither a performance variance nor a pie-chart category is a root cause.

The value comes from connecting the evidence: performance variance → observed loss pattern → tested operating condition → corrective action → verification.

Measure the work. Audit the system. Improve both.

That is how a traditional project-control signal becomes a sharper improvement conversation—and how turnaround productivity moves from reporting the gap to managing the conditions that created it.