Sixteen years of pallets: what we got wrong

In 2013 we invested in an automated sorting line on the reasonable assumption that consistency would beat human judgement. Vision systems, dimensional gates, the lot.

It sorted dimensions beautifully and it could not find a hairline stringer crack. A grader's hand running along a notch finds fatigue that no camera sees, because the failure is internal before it is visible. We shipped pallets that measured perfectly and failed in service, and it took most of a year to work out why.

We still use machinery for dimensional screening and stacking. The grade call is human and I no longer expect that to change. Details of how the sort actually runs are in inside the sort line.

For a stretch around 2015 and 2016 we measured ourselves on volume recovered. It is an intuitive metric for a recycling business and it is almost perfectly wrong.

Volume rewards taking every load, including loads that are 70% unusable, on long lanes, from sites that cannot load. We took them, our tonnage rose, and our margin went to nothing. We were paying to haul other people's landfill.

The metric that replaced it was recoverable value per mile. Less impressive on a slide, and the business became solvent again within two quarters.

For years the repair bench was treated as an overhead — something you had to do to make the pallets sellable. We did not price it separately, did not measure it separately, and did not sell it as a service.

It is the single highest-return activity in the building. Around three dollars of labour and materials returning an asset worth eleven to sixteen, at roughly ninety seconds a pallet. Once we started offering it as a service to customers with their own fleets, it became one of our best businesses, and our customers' cost per trip fell by a third.

The arithmetic that we should have run in 2011 is written out in the arithmetic of a repaired pallet.

For a long time we sold pallets and treated the truck as a necessary evil. Customers negotiated the unit price hard and we negotiated back, while 30 to 50% of the landed cost sat in a line nobody discussed.

The change was structural: itemise freight on every quote, ask every customer whether they generate empties, and route deliveries and collections together. Backhaul pairing takes 15 to 30% off a lane and it costs nothing except knowing that the pairing exists.

We now ask about outbound empties on the first call, every time. The number of customers who were paying a third party to haul away pallets from the same building we were delivering to still surprises me.

Early on we said the things everyone says. Green. Eco-friendly. Diverting from landfill. All broadly true and all unfalsifiable, which is precisely what made them worthless.

It became useful when we started publishing numbers that could be checked, including the ones that do not flatter us — the 8% residual, the fact that most of our diversion is reuse rather than recycling, the uncertainty range on the landfill methane factor. Customers building ESG disclosures need figures that survive an auditor, and a defensible 92% is worth more than an indefensible 99%.

The full position is on the sustainability page, and the reasoning behind the numbers is in carbon accounting without lying to yourself.

1. Publish your grade criteria on day one. It costs yield and buys everything else.
2. Measure recoverable value, not tonnage.
3. The repair bench is the business. Staff it properly.
4. Ask every single customer about their empties.
5. Never claim a number you cannot show the working for.
6. Grade twice, with two people. It is the cheapest quality control there is.

Sixteen years, 4.1 million pallets, and six things that would have saved most of the difficult parts. If you are running a yard, take them for free — we paid for them.