How to Audit a Hauler’s Diversion Claims

Somewhere in most enterprise sustainability reports is a diversion rate that originated as a number in a hauler’s quarterly summary. It was not fabricated. It was produced by a defensible-looking internal process at the hauler, using conventions that differ from the ones your reporting framework uses, applied to a facility average rather than to your material, and rounded.

That is not fraud and treating it as fraud gets you nowhere with your vendor. It is a measurement provenance problem, and it is auditable. This is the procedure — what to ask for, in what order, and what each answer tells you.

Why are hauler-reported diversion rates systematically high?

Four mechanisms, all of them ordinary, and all of them pushing the same direction.

The compounding of those four is why a portfolio can report a diversion rate that no individual site would recognise from walking its own dock.

What evidence should I demand for each claim?

The most useful published checklist for this is not in a carbon accounting standard. It is buried in LEED v4.1, in the criteria a third-party certifier must satisfy before USGBC will accept a commingled facility’s recycling rate. Whether or not you are pursuing certification, those criteria describe what a credible facility rate looks like — and by implication what an uncredible one is missing.

The LEED v4.1 Building Design and Construction guide requires that a certifying organisation audit facility-submitted data, and specifies that the audit include "the evaluation of recyclable sales records, verification of facility sales into commodity markets, an assessment of downstream materials and how these materials are managed after they leave the site, monitoring off-site movement of materials, and a review of the facilities’ customers weight tags information." Site visits must "verify use and accuracy of scales including calibration frequency." On the rates themselves: "measurements must be based on weight (not volume), using scales," rates "must be available on a website and viewable by the general public," and the reported rate must include "overall facility diversion rates with and without ADC/Beneficial Reuse."

Every one of those clauses converts directly into a question you can put to a vendor.

How do I check whether the numbers are internally consistent?

Before you escalate anything with a vendor, run four arithmetic tests on the data you already hold. They cost an afternoon and they find most of the problems.

1. Recompute the rate yourself from the underlying tonnage. If the vendor’s reported percentage does not reproduce from their own numerator and denominator, something is being adjusted between the data and the summary. Ask what.
2. Check whether the rate responds to anything. A site whose diversion rate is identical to two decimal places across four quarters, through a seasonal demand cycle and a container reconfiguration, is reporting an assumption rather than a measurement.
3. Compare across sites with the same operating model. Real diversion rates vary between similar sites because contamination, staffing and local end markets vary. A portfolio where thirty sites report the same rate is reporting one facility average thirty times.
4. Test the tonnage against physical capacity. Reported weight per pull divided by container volume gives an implied density. If a 30-yard container of mixed office recyclables implies a density that a compactor could not achieve, the tonnage is inferred, not weighed.

Where can I verify a hauler’s numbers against an independent source?

In some jurisdictions, against a regulator. California is the most developed example. CalRecycle’s Recycling and Disposal Reporting System collects quarterly reports from recycling and composting facilities, disposal facilities, transfer and processing facilities, contract haulers and food waste self-haulers, and brokers and transporters, covering the types and quantities of material they sell, transfer or dispose — and it publishes public reports drawn from that data. If a facility’s state-reported throughput and your hauler’s account of what happened to your material tell different stories, that is a conversation worth having.

California also imposes recordkeeping obligations that work in your favour as a customer. Under SB 1383, CalRecycle’s organics programme requires jurisdictions to provide organic waste collection to all residents and businesses and to recycle that organic waste, and covered entities "must maintain records used for compliance inspections required by local and state agencies." The final regulations text is the operative document. If your vendor cannot produce destination records for your organics, they may have a compliance problem independent of your reporting problem.

Outside jurisdictions with mandatory facility reporting, the independent source has to be you. Which is the uncomfortable conclusion of this whole exercise.

What does a genuinely independent verification look like?

Weighing at the point of generation, before the material enters anyone else’s custody.

This is the only measurement in the chain that is not produced by the party being paid. A weighed, timestamped, stream-tagged record created at the dock gives you a denominator the hauler did not supply, which turns their diversion claim from an assertion into a hypothesis you can test. Reconciled monthly against invoiced tonnage, the two series either track or they diverge, and a persistent divergence in one direction is the finding.

The GHG Protocol’s Corporate Value Chain (Scope 3) Standard names this directly in its data quality indicators, defining reliability as "the degree to which the sources, data collection methods and verification procedures used to obtain the data are dependable." Verification procedures are part of the definition of quality, not an optional overlay on it. And its technical guidance for Category 5 sets a similar bar for the supplier-specific method, whose activity data is the "allocated scope 1 and scope 2 emissions of the waste-treatment company (allocated to the waste collected from the reporting company)" — a number very few haulers can currently produce, which is itself worth knowing before you build a methodology around it.

On the emission factor side, keep one caveat in view. EPA is explicit that its Waste Reduction Model "is a comparative tool rather than a comprehensive measurement tool" and "should not be used for developing GHG inventories, which need to establish a baseline and measure reductions over time on an annual basis for an entity," per the model’s own documentation. WARM factors are widely used and well understood by assurance providers, and they are also not designed for the job many inventories give them. Disclose which factor set you used and why. We go deeper on this in our EPA WARM guide.

What is a plausible diversion rate anyway?

There is no benchmark that transfers cleanly to a single site, and anyone offering you one is selling something. But national data is useful as a sanity check on the order of magnitude of a claim.

Those figures come from EPA’s national overview of materials, waste and recycling facts and figures, which also notes that MSW as EPA defines it excludes "construction and demolition (C&D) debris, municipal wastewater sludge, and other non-hazardous industrial wastes." The boundary matters: a manufacturing site with a heavy, clean, single-commodity stream can legitimately exceed the national rate by a wide margin, and a C&D project is measured against a different universe entirely. What the figure is good for is calibrating scepticism. A general-purpose office or retail site reporting a diversion rate in the high eighties, from commingled containers, with no scale anywhere in the process, is making a claim that deserves the records behind it.

What should we actually do with the findings?

Restate carefully rather than dramatically. If you have been reporting a rate that included ADC as diversion, the correction is a methodology note and a restated baseline, and doing it voluntarily in a year when nobody has asked is considerably cheaper than doing it in an assurance engagement.

Then fix the contract, because the audit only stays clean if the data stays clean. The clauses that matter are unglamorous: tonnage reported by weight per load, receiving facility named per load, the operation performed at that facility identified, ADC reported separately from other beneficial reuse, residual rates disclosed for commingled loads, and raw records available on request rather than summaries. Vendors who can supply this generally will if it is written down at renewal. Vendors who cannot are telling you something useful about which of your sites need independent measurement.

For completeness on how we fit: Dyrt built Impact Engine to hold this reconciliation — ingesting hauler invoices, facility scale data, RFID readings and field measurements, normalizing billing periods, units, duplicates and facility IDs, and maintaining an auditable chain of custody from bin to receiving facility. Where a site has no independent measurement at all, a DWIT kiosk and industrial scale supplies it: weighed, timestamped, stream-tagged load records against a per-site taxonomy, described in detail in Inside DWIT. The same normalized records feed Spend Intelligence, which is why an audit that starts as a sustainability exercise frequently ends up as a procurement one.

The underlying point is not that haulers are dishonest. It is that a diversion rate is a measurement, measurements have provenance, and a provenance that runs entirely through the party being paid is not a provenance an assurance provider will accept. Establish one independent point in the chain and the rest of the conversation gets much easier.