For years we reported our recycling performance the way most of the industry does: units in, units out. It is easy, everyone understands it, and it is the wrong measurement.
In 2023 we moved to weighing every stream. Here is what that changed.
Why counts fail
A count tells you how many pallets arrived and how many left as pallets. It has nothing to say about the ones that were dismantled, and dismantling is where about a quarter of our intake goes.
Worse, counts flatter you. A pallet that arrives and is dismantled into boards that repair three other pallets shows up as one unit lost. In weight terms, most of its mass stayed in service. The count understates what actually happened.
And a count cannot express the grinding stream at all. Mulch does not come in units.
What we started doing instead
Every inbound trailer crosses a truck scale, gross and tare, so the net weight of wood entering the yard is recorded per load rather than estimated.
On the outbound side, every stream is weighed: graded pallets by load, recovered lumber sold on, mulch by the cubic yard converted at a measured bulk density, bedding, boiler fibre, ferrous scrap and disposal.
Monthly, the two totals get reconciled and the variance is recorded rather than smoothed away. That last part turned out to matter more than anything else.
Three things we were wrong about
Reuse was lower than we thought. By count, a lot of units go straight back out. By weight, straight reuse is about 34% of intake — meaningfully less than our internal folklore had it, because heavier units are disproportionately likely to be the damaged ones that need work.
Lumber recovery was higher. 27% of incoming weight leaves as recovered boards rather than as pallets. We had been treating dismantling as a loss and it is a major product stream, and the one that feeds the repair bench.
Moisture is a real line item. Wood loses weight as it dries in the yard. In a dry month the reconciliation variance runs about a percentage point, all of it water leaving the building as vapour. We had never accounted for it and it had been quietly polluting our numbers.
| Stream | By count (old estimate) | By weight (measured) |
|---|---|---|
| Reused as-is | ~48% of units | 34.1% of weight |
| Repaired and reused | ~22% of units | 23.8% of weight |
| Dismantled to lumber | Recorded as loss | 27.2% of weight |
| Ground to fibre | Not measurable | 13.9% of weight |
| Ferrous | Not measured | 0.4% of weight |
| Landfilled | Estimated | 0.6% of weight |
The variance is the interesting number
Inbound and outbound weights never match exactly. Moisture leaves, some material is still in process at month end, and scales have tolerances.
We report that variance rather than allocating it away, and it is the single most useful figure for anyone auditing the claim. A process that reports no variance is a process that is not measuring — it is a process that is balancing the books after the fact.
This is also why we are sceptical of a 100% diversion claim. Not because it is impossible in principle, but because it implies a measurement precision that scales and moisture do not permit.
What it cost
Honestly, not much. The scale was already there. What changed was discipline: weighing outbound streams that previously left on a nod, converting mulch volumes properly rather than estimating, and doing a genuine monthly reconciliation instead of a quarterly guess.
About four hours a month of someone's time, and it produced the only numbers on this website we would be comfortable defending in front of an auditor.