Packaging returns

Altitude Facilities — pallets and divider boards.

Demonstration data. Every grade split on this page is invented.

Deliveries, quantities and dates follow the shape of Altitude's own records; the Good / Repair / Scrap splits and the names attached to the work are generated, because those have never been recorded. Customer names are placeholders. Nothing here is a finding about Altitude, about a customer or about a person, and none of it may be quoted as one.

Packaging returns

Pallets and divider boards received back from customers, and what they graded to.

What on this page is real?

The first question anyone should ask of a report, and the only one this page can currently answer badly. It is counted from the data, not written in by hand, so it will start telling a better story on its own.

What came back

Deliveries and quantities are Altitude's own. Pallets and divider boards are never added together — they are counted in different units of usefulness, so a combined total would mean nothing.

Pallets Divider boards

Where it ends up

Rates are taken from the totals, never averaged across deliveries — averaging weights a 3-pallet load the same as a 300-pallet one. Loads that have arrived but not yet been sorted are excluded, not counted as good.

A divider board has no repair path. It is good or it is scrap, and the database refuses a repair count on one — so that column is empty by construction rather than by coincidence.

Scrap rate over time

Pallets, by month, with the shaded band showing where the true rate plausibly sits given how much was actually sorted that month. The dashed line is the scrap tolerance, .

tolerance
Scrap rate 95% confidence band tolerance

That figure is our understanding, not Alistair's confirmation. How the tolerance actually works — per delivery, per customer, per month, and what "paid less in proportion" means in rands — is an open question. Until it is answered this line is a marker, and no rand amount is shown anywhere on this page.

Customers, worst scrap first

Sorted by rate, which is the ranking everyone asks for and the one that misleads. Read the bar, not the number: it shows the whole range the sample is consistent with.

CustomerMaterialLoads GradedScrap Where the true rate sits
Show every customer
CustomerMaterialLoads GradedScrap Where the true rate sits

Over tolerance, and demonstrably so

The same customers filtered on the bottom of the range rather than the middle. Everything here is above tolerance even on the pessimistic reading of its sample — which is the claim worth making a phone call about.

CustomerMaterialLoads GradedScrap Where the true rate sits

Grading rates by grader

Pallets only. This panel exists to say where to look. It does not say what is true, and the column that matters most is the one counting customers.

Graded byLoadsCustomers GradedScrap Where the true rate sits

Either side of a handover

The same customers, graded by two different people at different times. Holding the customer mix roughly constant is what makes this comparison worth drawing at all.

Graded byShared customersPeriod GradedScrap Where the true rate sits
Show graders who worked alongside each other

Two people grading the same customers at the same time is a roster, not a handover — nothing was held constant, because nothing changed. Shown for completeness.

Graded byShared customersPeriod GradedScrap Where the true rate sits

How long ware waits before it is sorted

The one measure here about Altitude's own operation rather than a customer's. Median days from arrival to grading, over deliveries.

Median days Slowest 10%

Arrived, not yet sorted

A load checked and found clean is a zero. A load nobody has looked at yet is the absence of a grade. Merging the two destroys the rate, so this backlog is reported separately and excluded from everything above.

MaterialLoadsQuantity OldestNewest