The short answer
The blocker is not measurement, it is comparability. Different contractors report different units, different categories and different periods, so a portfolio view has to be assembled by hand and usually is not.
Normalising it once reveals the obvious: collections scheduled regardless of fill, and recycling rates that reflect what was collected rather than what was diverted.
What makes it hard
- Weight, volume or lift count, depending on the contractor
- Waste categories named differently across suppliers
- Billing periods that do not align
- Estimated weights rather than measured ones
- Sites with several contractors for different streams
Estimated weights are worth checking. Where a contractor bills on assumed weight per lift, the incentive to measure accurately is not strong.
What normalising reveals
| Finding | Typical action |
|---|---|
| Collections on schedule regardless of fill | Move to demand-based collection |
| Contamination charges recurring at one site | Signage, bins, staff briefing |
| Recycling rate lower than assumed | Check segregation at source |
| One site far above comparable ones | Investigate rather than assume |
| Paying for a stream nobody produces | Remove it |
The first row is where most of the money is. Fixed-frequency collections at partially filled containers are common across portfolios and straightforward to change.
Reporting obligations
Many organisations now report waste and recycling as part of broader environmental reporting, and those figures need to be defensible rather than assembled from invoices at the last minute.
Keeping the underlying records and a documented method makes a challenged figure explainable, which is the same discipline any regulated return needs.
Do the simple things first
- Get all contractor data into one place, in one unit.
- Compare cost and volume per site, normalised by occupancy.
- Investigate the outliers rather than averaging them away.
- Switch fixed-frequency collections to demand-based where containers allow.
- Then consider sensors, for the containers where uncertainty remains.
Steps one to four require no hardware and usually deliver most of the available saving.