It is almost never the people

When a foreign parent company compares its Brazilian plant against its European or Asian sites, the gap is usually real. The explanation offered locally is usually wrong, and the explanation assumed at headquarters is usually worse.

Written for operations leaders running or evaluating manufacturing in Brazil.

A recurring conversation: headquarters sees output per employee in Brazil well below the group benchmark. The local explanation is that Brazil is different — labour rules, suppliers, infrastructure. The headquarters explanation, rarely said aloud, is that the local team is less capable.

Both are convenient and both are mostly wrong. Having spent time on Brazilian shop floors, the gap we find is real but it is structural, and structure can be changed.

What actually drives the gap

Planning quality, not effort

The single most common finding is that production planning is weaker than the plant’s technical capability. Scheduling lives in one person’s spreadsheet, capacity is assumed rather than calculated, and the plan changes daily under commercial pressure with no rule for who may change it.

The visible symptom is a plant that works hard and delivers late. Operators are busy, machines are running, and the sequence they are running is not the sequence that would deliver the orders.

Changeover, not cycle time

Foreign benchmarks usually come from higher-volume operations. Brazilian domestic demand is more fragmented, so plants run a wider mix at lower volumes. Applying a cycle-time-focused improvement programme to a high-mix plant misses where the time actually goes.

In high-mix operations, setup and changeover frequently consume more available time than anyone has measured. Nobody has measured it because it is considered normal.

Measurement that is not true

We regularly find reported performance well above observed performance — not through dishonesty, but because micro-stops go unrecorded, changeovers get classified as planned downtime, and rated speeds in the system were set optimistically years ago. Improvement built on a false baseline goes nowhere.

A supervisor layer nobody trained

Brazilian supervisors are typically promoted operators with strong technical knowledge and no management training. Asking them to sustain a management system without teaching them how to run one is a design error, not a personnel failure.

What genuinely does not help

• Importing the group standard unchanged. A system designed for high-volume, stable-demand plants applied to a high-mix Brazilian operation optimises the wrong thing.
• Training as the intervention. Training changes what people know. It does not change what the daily routine requires of them, and the routine wins.
• Replacing the plant manager. Occasionally warranted, usually a reset that costs a year and reproduces the same structure.
• Buying capacity. Investing in a fourth machine while three run at half their real capability is the most expensive way to avoid a measurement problem.

What does work

1. Establish a true baseline first. Measure directly, on the floor, with an unambiguous loss taxonomy. Expect the honest number to be worse than the reported one, and expect that to be politically uncomfortable.
2. Fix planning before flow. A stable, respected production plan returns more in most Brazilian mid-size plants than any lean tool.
3. Attack changeover. In high-mix operations this is usually the largest recoverable block of time, and it requires no capital.
4. Build the supervisor role explicitly. Write what a supervisor does hourly, then coach it on the floor until it holds.
5. Install a daily routine that surfaces deviation within the hour, not at month-end.

The uncomfortable part

The honest baseline usually reveals that the plant has been reporting numbers that were not accurate, and somebody has been presenting those numbers to the parent company. Handling that well matters more than the technical work. If establishing the truth becomes a blame exercise, the measurement quietly reverts and you are back where you started within two quarters.