Training is an easy lever to pull. It feels constructive. It rarely fixes a process that punishes people for doing the right thing slowly - or rewards them for skipping steps.
If throughput is the only praised metric, quality steps will erode no matter how many modules you assign. The floor optimizes what gets measured and rewarded. Training cannot outrun incentives.
Broken handoffs are another training-proof problem. Teaching someone to label perfectly does not help if the next station ignores labels when rush hits. The interface between roles has to be designed, not hoped for.
Overload is the silent killer of training ROI. When every shift runs thin, people revert to habit. New procedures need slack to practice - temporary staffing, simplified menu, or reduced parallel projects. Without slack, training evaporates under pressure.
Good training still matters - for new hires, new equipment, and real skill gaps. It works best when paired with clear standards, visible examples of “done,” and supervisors who reinforce the same lesson on the floor.
If mistakes persist after training, assume the system is teaching a different lesson than your deck. Change the system. Then train once - and mean it.