A two-site returnable container pool is the easiest reuse programme there is, which is exactly why it is worth getting right. If you cannot make two nodes work, you will not make eleven work, and most organisations that have failed at a large pool failed the small version first without noticing.
The setup
Site A ships to Site B. Containers travel loaded one way and should travel back empty. That is the entire system. Everything below is about making that second sentence actually happen.
Step 1 — Size the population
The formula we use is deliberately crude because precision here is false comfort:
Population = peak weekly demand × cycle time in weeks × 1.2
Cycle time is the full round trip: load, transit, unload, dwell at B, return transit, grading, dwell at A. Measure it, do not estimate it. In every pool we have helped set up, actual cycle time has exceeded the estimate by 30–60%, almost entirely in dwell.
| Input | Value |
|---|---|
| Peak weekly shipments | 220 containers |
| Measured cycle time | 2.4 weeks |
| Base population | 528 |
| Plus 20% slack | 634 |
| Rounded order | 650 |
The 20% is not padding. An undersized pool generates emergency purchases of new containers, which is precisely how a reuse programme ends up buying virgin board and then being judged a failure.
Step 2 — Name an owner at each node
This is the step that decides whether the programme survives. Outbound is always somebody's job; the return leg almost never is, because it produces nothing and costs time.
One named person at each site, with empties in their objectives. Not a department. Not 'the warehouse'. A person.
Step 3 — Schedule the return, do not request it
A return trip that has to be requested will be skipped whenever the week gets busy, and the week is always busy. Put the empty return on a schedule: same truck, same day, whether or not it is full. An under-filled scheduled return is cheaper than a pool that silently depletes.
Where the lanes line up, the best version is a backhaul: your empties ride on a trailer that was returning anyway. Half the carbon and a fraction of the cost.
Step 4 — Grade on inbound, within 24 hours
Grade the containers as they arrive back, not as they go out. Grading on outbound feels more efficient and is a mistake: damaged containers re-enter circulation, fail under load at the receiving site, and poison confidence in the whole programme. One rack failure costs more goodwill than fifty successful trips earn.
Step 5 — Have a repair path
Either a bench of your own or a partner who has one. Without it, every damaged container is a total loss and your attrition rate roughly doubles.
The four ways it fails
- Nobody owns the empties. By a distance the most common. Symptom: population drifts down 15–30% a year and nobody can say where they went.
- Footprint mismatch. Containers half an inch proud of the pallet take a hit on every rack move. Measure the pallets you actually own, not the ones in the standard.
- No grading step. Damaged containers circulate, fail, and destroy trust.
- Undersized pool. Emergency new purchases, which undermine the business case that justified the programme.
What to measure
One number matters more than all the others: trips per container per year.
- Rising: the pool is healthy.
- Flat while volume grows: you are buying your way out of a process problem.
- Falling: you have a handling problem, a storage problem, or an attrition problem, and the container count will tell you which.
What to expect, honestly
Month one is mostly disruption. You will find containers in places nobody knew about and the first grading report will be worse than expected. Month six: cost per trip visibly falling, emergency purchases near zero, trips per container still flat. Month eighteen: trips per container typically up 30–50% from baseline.
Knowing that curve in advance is what stops programmes being cancelled in month five, which is when they look worst and are about to start working.
Our pieces are written by the people who do the work rather than by a content agency, which is why they contain more numbers and fewer adjectives than is fashionable. If a figure here does not match your experience, we would genuinely like to hear about it — two of our published coefficients have changed that way.