Fourth year of publishing these. All figures are by weight from the truck scale, reconciled monthly, variance stated.
2025 was the year several things we had been working on for a while finally came right, and one thing we thought was solved turned out not to be.
The numbers
| Measure | 2024 | 2025 | Change |
|---|---|---|---|
| Pallets received | 1.50 M | 1.60 M | +6.7% |
| Graded units shipped | 868,000 | 925,000 | +6.6% |
| Units dismantled | 412,000 | 425,000 | +3.2% |
| Wood ground, tons | 9,800 | 10,500 | +7.1% |
| Ferrous recovered, tons | 164 | 175 | +6.7% |
| Landfilled, tons | 108 | 125 | +15.7% |
| Diversion by weight | 99.1% | 99.4% | +0.3 pts |
| Reconciliation variance | 1.4% | 0.9% | −0.5 pts |
What went right
- No spring shortfall. We built roughly 55,000 more units of inventory through December 2024 than the previous year, and Grade A held through April. This was the main lesson from 2024 and it worked.
- Reconciliation variance back under 1%. Tighter outbound weighing discipline, particularly on the mulch streams, which had been the weak point.
- Diversion at 99.4%. Close to what we think the ceiling is for a yard that accepts difficult material.
- Repair share up again, to 28.6% of intake by weight. The dismantling line kept the bench fed consistently for the first full year.
What went wrong
We lost two experienced graders within six weeks of each other in the summer, and the mis-grade rate went from one in 640 to one in 480 for most of Q3 before recovering. That is a real quality regression and several customers noticed it before we did, which is the part that stings.
The honest analysis is that we were running the sort deck with too little redundancy. Two people leaving should not degrade output measurably, and the fact that it did means the training pipeline was thinner than we had told ourselves.
Three trainees started in September and the deck is back to full strength, but it took a quarter and it should not have.
The landfill number, honestly
125 tons, up from 108. As a share of intake it fell — 0.6% against 0.9% — but the absolute number went up and it is worth saying why rather than only quoting the percentage.
We took on two industrial collection customers whose material includes a higher proportion of stained and contaminated units. That material has to be segregated and disposed of properly, and doing so is better than the alternative, which is somebody grinding it into a landscape product.
So the absolute increase is a consequence of accepting harder material, and we would rather that than a better-looking number achieved by being fussier at the gate.
For 2026
Rebuild the grader bench depth so a departure is not a quality event. Hold the variance under 1%. And a piece of work we have been avoiding: our own diesel. Fixed lanes have cut empty running substantially over three years, but the marginal gains are getting smaller and the next step is probably route optimisation software rather than more discipline. That is an unglamorous project and it is the largest remaining term in our own emissions.