The blind spot in operational safety: the “during” of execution

In many organisations, safety is well designed.What is not resolved is something else: ensuring that safety operates while the work is actually happening.

Because real risk is not decided in the procedure, nor in induction, nor in an audit. It is decided in execution: when conditions change, interferences appear, the plan is adjusted, pace increases to maintain operational continuity, and decisions are made with incomplete information.

And that is where a familiar pattern appears: the “before” is well controlled and the “after” is investigated, but the “during” remains weak, late, or isolated.

1) The “during” is observed… but it stays contained

On site, many things are indeed detected: a change in conditions, an interference, an unforeseen risk, a control that is inadequate. Sometimes it is even recorded on a Permit to Work or the AST is updated.

The issue is not intent. It is the mechanism.

When those signals remain on paper (or in conversations), they tend to stay contained:

• they do not trigger alerts or escalation in the moment,

• they do not update the status of the task for supervision / HSE,

• they do not enable a structured pause/hold,

• and they do not become formal improvements to the process (critical controls, standard permits, standards).

Result: the organisation “discovers” the same problem again and again. Not due to a lack of rules, but due to a lack of activation and learning.

2) Risk moves faster than the system

Operations today are objectively more complex:

• more requirements and validations,

• more interaction between areas and contractors,

• more condition changes during the task,

• more pressure to maintain operational continuity.

In that context, the typical failure is not “the control didn’t exist”. It is:

• the permit was approved, but a condition changed,

• the isolation was defined, but it was not executed as expected,

• the PPE was specified, but it was not used,

• the risk was known, but it was not identified in time to trigger an action.

That is not solved with more documentation. It is solved with the ability to respond during execution.

3) If it is already difficult with in-house teams, exposure multiplies with contractors

Maintaining consistent standards with your own workforce is already difficult: turnover, shift changes, fatigue, variability and pressure to push on.

With contractors, that variability increases: higher volumes, frequent turnover, changing crews, less familiarity with the site and internal operating criteria. In many operations, contractors are not marginal; they are a structural (and sometimes majority) part of field work.

In response, the typical approach remains documentary clearance:

• compliant documentation,

• entry inductions,

• acknowledgement signatures,

• permits issued.

This is necessary for compliance. But it is insufficient as operational control.

Because a brief induction and a signature do not validate sustained practical competence, nor do they ensure consistent execution when conditions change, interferences arise, or the unexpected happens. In practice, the system becomes strong at “creating evidence” and weak at supporting the “during”.

4) The problem is not measuring: it is that what gets measured isn’t always what matters

Here lies the trap: many organisations believe they are fine because the indicators “look good”.

But dashboards tend to focus on:

• outcomes (accidents, reported incidents),

• and compliance (permits issued, training completed, checklists closed).

That can create a sense of control, while real risk concentrates in:

• near misses that are not recorded,

• condition changes that do not trigger revalidation,

• deviations in critical controls without traceability,

• behavioural observations without an established and consistent practice,

• operational friction (rework, delays) that pushes informal adaptation.

This is not a “lack of culture” problem. It is a problem of instrumenting and measuring the “during”.

Faced with this scenario, what is being done today?

In most cases, the response is still to intensify what already exists:

• more documentation,

• more training,

• more audits,

• more pre- and post-controls.

That improves record-keeping and compliance. But it does not resolve the central issue: execution continues with low assistance and limited real-time visibility.

And when the “during” is not supported, the system ends up depending on two fragile things:

• individual memory and judgement in the critical moment,

• and after-the-fact corrections, when it is already too late.

What a modern solution should provide if the problem sits in the “during”

If we accept that risk is defined during execution, then the question changes. It is no longer “do we have procedures?”. It is:

Do we have a system that operates during the task, with the capacity to respond and to learn?

In practical terms, a modern solution should:

• Support the worker during execution: guide critical tasks step by step when appropriate, without relying on memory.

• Revalidate when conditions change: detect/record changes and trigger revalidations (not merely note them).

• Activate actions: alerts, escalation, structured pauses/holds, reassignment of controls.

• Provide real-time visibility: for supervision and HSE, while the task is live.

• Capture signals and turn them into standards: ensure what is learnt on site becomes formal improvements (critical controls, standard permits, standard AST).

• Scale to contractors: with in-task validation and operational consistency, not only documentary clearance.

This does not replace accountability. It reduces variability and turns safety into an operational system, not a record-keeping system.

The final question

If the system controls the start and documents the close, but does not intervene during execution, how do you ensure that work remains within critical controls when conditions change?

Warning signs: when the “during” may be out of control

• Are risky situations detected, but left on paper or in conversations?

• Are condition changes recorded, but do they fail to trigger revalidation, alerts or decisions in the moment?

• Do critical controls depend on the skill and memory of the person executing, rather than on a consistent mechanism?

• With contractors, does effective control shrink to entry paperwork and induction, with no real in-task validation?

• Do lessons learnt remain local and fail to translate into formal improvements (standards, standard permits, critical controls)?

If you answered “yes” to two or more, the issue is not a lack of rules. It is the absence of a system that supports, activates and captures execution in real time: one that turns signals into actions during the task, and turns what is observed into formal improvement so it doesn’t repeat.

If these warning signs feel familiar, we invite you to visit wiiprot.com