The faded floor tape is telling you something specific

Walk into a Brazilian plant that has “already done lean” and the evidence is usually still visible: faded 5S markings, a kanban rack used as shelving, a kaizen board with cards from two years ago. The tools were installed. Something else was missing.

Written for operations leaders running or evaluating manufacturing in Brazil.

The second lean programme is harder than the first. The team has seen this before, watched it fade, and drawn a reasonable conclusion about how much of their attention the next one deserves. Any attempt that ignores that history starts at a disadvantage.

So it is worth being precise about why the first one decayed. In our experience it is rarely cultural resistance, which is the explanation most often offered.

The actual failure modes

It was run as training rather than as a change to how work is supervised

Workshops happen, tools get installed, certificates are issued. Then the facilitator leaves and nothing in the daily routine requires the new behaviour. The old routine is still there, still measured, still what the supervisor asks about. The old behaviour returns because the system asks for it.

Flow tools were applied to an unstable process

Pull systems and levelling assume a base of process stability. Installed on top of an unstable process, they make the instability visible faster — which is technically correct and organisationally fatal, because the kanban gets blamed for problems it merely exposed.

Kanban was sized on textbook assumptions

Loop sizing based on European or Japanese replenishment reliability runs dry in Brazil, where inbound lead times are longer and more variable. The system fails, and the conclusion drawn is that pull does not work here rather than that the sizing was wrong.

Standardisation destroyed knowledge

Operators keep lines running through informal workarounds that encode real understanding of the equipment. Standardising without first understanding why each workaround exists removes capability, and the standard then gets quietly ignored because following it makes the job harder.

Nobody owned the audit

5S without a scheduled audit, a named owner and a consequence for drift is a cleaning campaign. It decays on a predictable timeline, usually within two quarters.

What changes the outcome

1. Diagnose the previous programme first. Why it decayed is the most useful information available, and it is free.
2. Sequence properly. Stability and standard work, then flow, then pull. Skipping to the visible tools is the most common error.
3. Size for Brazilian variability. Buffers larger than the theory prefers, because the theory assumes supply reliability you do not have.
4. Build the standard with the operators who invented the workarounds. They defend what they helped write; they ignore what they were trained on.
5. Make the sustaining mechanism part of the deliverable — layered audits, supervisor standard work, a review cadence with a named owner.
6. Write everything in Portuguese. A standard in English is a standard that will not be followed on the floor.

A note on where lean pays here

Brazilian SME manufacturing tends toward high mix and lower volume. That shifts the priority order. Changeover reduction usually returns more than line balancing, and standard work usually returns more than levelling. Applying the tool sequence from a high-volume export plant is how a technically correct programme produces nothing.

The test

A lean programme is complete when the team runs an improvement cycle without the consultant present and the result holds three months later. Anything short of that is a project, not a change — and projects fade, which the faded tape on the floor already told you.