The Operations Lens: Designed, Not Held Together

Documented is not the same as designed

Most operations have documentation. Far fewer have a process defined as it is actually performed. The gap between the two is where variance lives: the workaround that became standard, the check that is skipped when the schedule is tight, the informal sequencing one supervisor uses and no one else knows.

Walk the work and the gap is visible within an hour. The written procedure describes eleven steps; the operator performs nine and adds two the document never mentions. Neither the operator nor the document is wrong. The design has simply never been reconciled with reality.

Until it is reconciled, every improvement is applied to a process that does not exist, and every measurement describes something other than the work.

You cannot improve a process you have not first described as it is actually performed.

Rework has to be counted before it can be removed

In an undesigned operation, rework is absorbed rather than recorded. The correction happens inside the same shift, by the same team, and never appears in a number. The enterprise pays for it in capacity, overtime and lead time without ever seeing the line item.

Counting rework is uncomfortable and cheap. A tally sheet at the point of correction, kept for two weeks, produces a more useful picture of an operation than most quarterly reporting packages — because it is measured where the loss occurs rather than where the accounts are closed.

Standard work is a leadership instrument

Standard work is often mistaken for a control on the operator. Its real function is to make abnormality visible: if the standard is followable and the result still varies, the problem is upstream, and leadership now has an honest signal instead of an opinion.

That is why the operations lens weights process and measurement above everything else. Systems, capability and cadence all rest on whether the work itself is defined well enough to be reasoned about.

Exhibit — The two-week rework tally

One sheet at the point of correction. No categories invented in advance; add rows as the corrections occur. Reviewed by the leadership team at the end of week two.

The 'where caused' column is the finding. In most operations, the majority of corrections are caused two or more steps upstream of where they are detected.

How this shows up on the floor

• Output varies materially by shift, crew or supervisor.
• The written procedure and the observed work differ in more than two steps.
• Rework is corrected inside the shift and never recorded.
• Measures describe volume produced rather than the condition of the process.
• Improvement projects are chosen by opinion rather than by where loss is counted.

An operation that is designed can be improved by anyone. An operation that is carried can only be improved by the person carrying it — which is the same as saying it cannot be improved at all.

Operating performance as designed cadence and governed execution rather than heroics.