Cost at the unit · one claim, priced line by line
Deduction to settlement
A retailer takes money off an invoice and asserts a reason. Somebody has to decide whether the reason matches an agreement the group actually signed, and then either accept it or claim it back. Across the week of 10 to 16 August 2026 the sample assembled 170 of these and priced them with 848 rate lines. The outcome pages price a chain at one blended human rate and two flat system rates. These pages price each stage at the rate for the function that ran it, the model class it called and the system it ran in, then add the two components the outcome pages never carried at all: platform license and management overhead. The totals therefore differ, and the difference is a change of method rather than new spending. Both numbers are shown side by side wherever they appear.
What one of these is made of
The five components across 170 chains, divided down to one.
The evidence behind these dollars
Every line on this service declares the tier of the rate that priced it.
Loaded rates, license costs and overhead bases are the customer’s figures, and no customer figure has been supplied. Every rate on these pages is a modeled composite: a plausible number of the right order for a group of this size, not a number taken from a payroll run, a vendor contract or a general ledger. Until the real rates arrive, this wave demonstrates the instrument and proves nothing about the business case. That is the same statement the benefit pages make, and it is not softened here.
Every rate that touched this service
The quantity consumed, the rate applied and the tier that rate sits in. Each row is the whole of that rate across the sample.
The exception path against the clean one
The same rate card applied to both, so the difference is the work rather than the pricing.
An exception here costs 4.73× a clean claim, and loading the rates does not move that ratio — the direct rates the exception register uses give the same multiple to two decimal places, because the loading falls on both sides of the comparison. Only 17 chains sit on the thinner side of that split, below the twenty this product requires, so treat the multiple as a curiosity rather than a measurement.
The dominant code is O2C-PRF, on 7 of the 17 sampled chains that raised anything at all.
A clean chain and an exception chain of the same service are priced with the same rate card, so the multiple between them is a difference in the work, not in the pricing. A service is only marked measured where at least twenty chains of each kind exist in the sampled week; below that the mean is a curiosity rather than a measurement, and the page says so on the row. The weekly excess scales the sampled difference by fifty and assumes an agent removes 60 percent of it, which is an assumption and is stated as one.
The ten most expensive chains in the week
Sorted by fully loaded cost. Every one of them is a real assembled chain you can open.
Inference attributed to this service
Its share of the estate’s weekly inference bill, worked out from the stages it actually ran.