If you are a packaging vendor or a procurement lead trying to stand up a closed-loop program with a partner, this is a list of the three things we wish we had known in 2019. We learned each of them the hard way.
Mistake 1: Underestimating cycle time variance
Our first closed-loop program assumed a 14-day average cycle from outbound shipment to empty return. In practice, the cycle averaged 21 days and had a standard deviation of about 6 days. That meant we needed roughly 50% more boxes in circulation than we initially modeled.
Why we got it wrong: the receiving facility could not stage empties on the same dock cadence we assumed. Their inbound dock was busier than our model expected, and empties got pushed to a back-of-warehouse staging area where they waited for a less-busy day for backhaul. The friction was real and we should have measured it before committing to a fleet size.
Lesson: model cycle time with at least a one-sigma buffer before sizing the box fleet. Better yet, run a 90-day pilot with a small fleet to measure the actual variance before committing to full volume.
Mistake 2: Specifying the wrong wall count
We started the program with double-wall Gaylords because the cost per box was lower. After about three months we realized the double-walls were not surviving the round trip well enough — the typical lifespan was three trips before regrade-fail, vs the four-plus we had projected.
We re-specced to triple-wall in month four. The triple-wall version had a higher unit cost but the life-cycle math improved by about 30%. The customer was patient. We were lucky.
Lesson: spec for the second and third trip, not the first one. The marginal cost of a stronger wall is small. The marginal value of an extra trip is large.
Mistake 3: Not budgeting for inspection drift
In the first program, we did not formally calibrate inspectors against the closed-loop fleet. The fleet boxes started showing up at the regrade station with subtly different damage patterns than our general inbound, and our standard rubric started misclassifying them.
We caught the drift about six months in, after a customer-side audit flagged that they had been receiving Grade B-labeled boxes that were really Grade C on their own internal rubric.
Lesson: a closed-loop fleet has its own damage signature. Build a separate calibration drill against the actual fleet, not the general inbound. We do this now as a matter of course.
The fourth mistake we did not make
We were tempted, in mid-2020, to over-promise on portfolio-level reuse rates to win additional customers. We did not. If we had, the customers we onboarded would have been disappointed by their actual numbers, and we would have lost trust with the ones we already had.
Promising specifically and conservatively is the most boring form of customer-relationship discipline. It is also the one that pays compounding interest.