Why we publish the numbers that don't flatter us

Every year when the impact figures are compiled, there is a short conversation about the timber pallet number. It is 71%, it has not improved materially in four years, and it is conspicuously worse than everything around it.

And every year we publish it, because the alternative is to publish a set of figures with no mechanism by which anyone could tell whether they were honest.

§The asymmetry of unverifiable claims

Consider our cage reuse figure: 98.4% of intake cages return to service. That is a genuinely good number and it is also completely unverifiable from the outside. You cannot audit it, you cannot infer it, and there is no regulator checking it.

So why should anyone believe it? The only evidence available is the behaviour of the figures around it. A set of numbers where every single one is excellent has been curated. A set where one is plainly bad has probably been counted.

The bad number is not a liability on the good ones. It is the only evidence for them.

§What we publish that we would rather not

The second row is the one that actually stings, because it is a direct consequence of a decision we are otherwise pleased with. Extending into longer lanes increased total diversion tonnage substantially and made our per-unit freight worse. Both things are true and only one of them is a good headline.

§The four rules we try to follow

1. 01Publish the denominator. A percentage without a denominator is a decoration. 92% water recovery means something only alongside the 148-gallon gross and 11.8-gallon net figures.
2. 02Name the boundary. A carbon saving without a stated system boundary cannot be checked or compared, and the boundary is where most of the generosity in impact reporting lives.
3. 03State the weakest assumption. In our model it is the 1.0 substitution ratio for recovered HDPE, and we tell readers to discount the reprocess figure by up to 20% if they want a conservative view.
4. 04Say what is counted and what is modelled. Counted quantities and modelled impacts carry completely different levels of confidence and should never be presented in the same voice.

§What this costs

It is worth being honest that this is not free. Three specific costs:

• Our headline numbers are smaller than they could defensibly be, because we exclude avoided disposal freight and avoided landfill methane — both real, both working against us.
• Published grading criteria mean customers can and do hold us to them, which is the point and also means a mis-grade becomes a documented failure rather than a conversation.
• Saying "ISO-aligned" rather than "ISO certified" costs us on approved-vendor questionnaires, every time, to a competitor who is willing to blur it.

That third one is the most consistently irritating. We run the document control, the records and the internal audit cycle the standard asks for, and we are not certified, and describing ourselves as certified would be a lie that is easy to check and rarely checked.

§What it buys

One thing, which turns out to be enough: customers who have read the methodology trust the quantities. Several of our larger accounts use our factors directly in their own sustainability reporting, which they could not do with a blended headline number from an unnamed boundary.

And a secondary benefit we did not anticipate: publishing the timber figure prompted two customers to switch to plastic bases at purchase, which improved their reuse outcome and ours. A bad number, published, turned into a better outcome. A bad number, hidden, would not have.

§A test you can apply to anyone

Next time you read a supplier's sustainability page, look for the worst number on it. If every figure is excellent, you are reading a selection rather than a measurement.

Then ask three questions: what is the denominator, what is the boundary, and which of these figures is counted rather than modelled. The quality of the answers will tell you considerably more than the size of the numbers.