Lean Six Sigma in logistics

The defining problem in logistics is that the chain is improved in pieces. The warehouse raises its picks per hour, the transport team raises its vehicle fill, the planners raise their forecast accuracy, and the customer still gets the order late and incomplete. Each function optimised its own measure, and each optimisation pushed cost or delay into the next link.

Underneath that sits variability, which is the real enemy in any flow. A carrier whose transit time averages two days is easy to plan around. A carrier averaging two days with a tail at nine is not, and every downstream party protects itself with inventory, which is the chain paying for that variability in cash.

The third is dwell: the hours a load spends waiting to be loaded, waiting for paperwork, waiting for a door. It rarely appears on any report because no single organisation owns it, which is precisely why it persists and why an end to end map finds it immediately.

A logistics project should be scoped end to end even when it is improved in one place, because the measurement has to cross the boundaries even if the fix does not. Define names the order type and the two ends: order placed to goods accepted by the customer, not warehouse gate to warehouse gate.

Measure is faster here than anywhere else, because scanners, telematics and order systems already timestamp almost everything. The work is stitching the sources together into one timeline per order, and the first honest timeline usually shows that most of the elapsed time is dwell rather than movement.

Analyse separates common cause variability from genuine events. Seasonality, weather and a strike are not the same thing as a process that is simply unstable, and treating a Christmas peak as a special cause to be investigated wastes a quarter. Improve then has two families available: remove a wait, or reduce a variation. Control is a chart per lane or per site, reviewed with the carrier rather than about the carrier.

The method assumes you can change the process, and in a chain you often cannot change the part that matters because somebody else owns it. A project that finds the constraint at a third party becomes a commercial negotiation, and pretending otherwise produces a beautiful analysis and no improvement.

The second limit is that local efficiency measures are deeply embedded and individually defensible. Asking a site manager to run at lower productivity so the chain runs better is a reasonable request that their bonus forbids. Unless the measures change together, the improvement is temporary.

The third is volatility beyond the operation. No amount of variation reduction inside a warehouse compensates for demand that swings threefold without notice, and in those conditions the honest advice is to buy flexibility rather than to chase capability.