LensReading which lens this session carries.

Cost at the unit · what every number is built from

Rate card and meters

12 rates price every one of the 10,939 cost lines behind the outcome unit costs. Each declares on the record which of three tiers it belongs to, what basis it is apportioned on, who owns it and how much confidence that owner places in 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.

8
of 12 rates have no meter
Rates published
12
Across five cost components, pricing 10,939 lines
Metered by value
5.9%
On 43% of the lines
Apportioned by value
32.0%
A real invoice, divided by a chosen driver
Estimated by value
62.1%
Neither quantity nor price observed
Mean owner confidence
0.60
5 rates sit below 0.50
Metering work open
8
Governing $2.7m a week of estate cost

Every rate, and what it is made of

Sorted as they are applied. The bar is the share of the priced sample that rate carries.

Shared-service loaded hour
People · $44.00 per human hour
Estimated
No meter exists
$6,461 · 11.3% of the sample · 272 lines
Marta Oyelaran
Head of GBS Finance
0.44
owner confidence
Basis: None — charged directly at the stage. A blended delivery-center hourly rate covering salary, employer taxes, benefits and paid absence. Estimated because no payroll extract has been connected; the figure is a market composite, not this company’s cost.
Applies to: Stages run by IT, Customer Service and HR. One piece of work is scheduled against it: Instrument actual handling minutes instead of assuming eleven per touch, owned by Tomas Ferreira.
In-market specialist loaded hour
People · $58.00 per human hour
Estimated
No meter exists
$12,760 · 22.3% of the sample · 409 lines
Marta Oyelaran
Head of GBS Finance
0.44
owner confidence
Basis: None — charged directly at the stage. The single blended rate the outcome ledger used for every stage. Kept here as the in-market specialist rate so the two ledgers can be compared line for line.
Applies to: Stages run by Finance, Procurement, Supply Chain and Sales. One piece of work is scheduled against it: Connect the payroll extract and replace all three labor rates with actual cost, owned by Marta Oyelaran.
Professional loaded hour
People · $96.00 per human hour
Estimated
No meter exists
$5,738 · 10.0% of the sample · 120 lines
Marta Oyelaran
Head of GBS Finance
0.40
owner confidence
Basis: None — charged directly at the stage. Applied where the stage needs qualified judgment rather than processing. Estimated on the same market composite as the other two labor rates and carrying the same weakness.
Applies to: Stages run by R&D and Marketing. One piece of work is scheduled against it: Separate the professional rate by discipline rather than by function, owned by Marta Oyelaran.
Extraction and classification call
Inference · $0.18 per stage
Metered
A meter exists
$169 · 0.3% of the sample · 788 lines
Priya Raghunathan
Platform Engineering Lead
0.86
owner confidence
Basis: None — the token meter reports per call. Short-context extraction and classification. The token meter is real and reports per call, so this line is metered rather than allocated.
Applies to: Receipts, reconciliations, access and change records, close and payroll items. It is already metered, so no metering work is scheduled against it.
Transactional reasoning call
Inference · $0.42 per stage
Metered
A meter exists
$817 · 1.4% of the sample · 1,430 lines
Priya Raghunathan
Platform Engineering Lead
0.86
owner confidence
Basis: None — the token meter reports per call. The rate the outcome ledger charged to every stage regardless of what ran. Retained here for the transactional middle of the estate.
Applies to: Requisitions, orders, invoices, payments and journals. It is already metered, so no metering work is scheduled against it.
Long-context drafting call
Inference · $1.35 per stage
Metered
A meter exists
$1,453 · 2.5% of the sample · 690 lines
Priya Raghunathan
Platform Engineering Lead
0.82
owner confidence
Basis: None — the token meter reports per call. Long-context reasoning and drafting where the model reads a specification or a claim in full. Three times the transactional rate and concentrated in a small number of document types.
Applies to: Exceptions, specifications, artwork, promotions, claims and certifications. It is already metered, so no metering work is scheduled against it.
Core platform seat
License · $2.10 per stage
Apportioned
No meter exists
$4,746 · 8.3% of the sample · 1,126 lines
Declan Whitmore
IT Vendor Manager
0.61
owner confidence
Basis: Annual contract value divided by modeled annual stage volume. The core finance, materials, HR and service-management platforms are bought as enterprise agreements with no per-transaction meter. Dividing by stage volume is a choice, and a different divisor gives a different unit cost.
Applies to: Stages executed in erp-fi, erp-mm, hcm, itsm and cmdb. One piece of work is scheduled against it: Obtain named-user and transaction-band data for the five core platforms, owned by Declan Whitmore.
Specialist application seat
License · $5.40 per stage
Apportioned
No meter exists
$9,175 · 16.1% of the sample · 1,048 lines
Declan Whitmore
IT Vendor Manager
0.55
owner confidence
Basis: Annual contract value divided by modeled annual stage volume. Narrow applications carry a much higher cost per stage because their contracts are sized for a small population. The same divisor weakness applies.
Applies to: Stages executed in the ten specialist applications outside the core five. One piece of work is scheduled against it: Move the ten specialist applications onto consumption billing at renewal, owned by Declan Whitmore.
Orchestration runtime
Infrastructure · $1.10 per stage
Apportioned
No meter exists
$4,355 · 7.6% of the sample · 1,750 lines
Priya Raghunathan
Platform Engineering Lead
0.68
owner confidence
Basis: Monthly compute bill divided by modeled monthly stage volume. Container and queue runtime for the orchestration layer. The bill is real; the split across stages is an allocation because the runtime is not tagged per workflow.
Applies to: Every stage in every chain. One piece of work is scheduled against it: Tag orchestration compute by workflow so runtime stops being allocated, owned by Priya Raghunathan.
Document storage and egress
Infrastructure · $0.24 per stage
Metered
A meter exists
$950 · 1.7% of the sample · 1,750 lines
Priya Raghunathan
Platform Engineering Lead
0.91
owner confidence
Basis: None — the object store bills per object and per byte. The only infrastructure line with a genuine per-unit meter. Small in absolute terms and included because a ledger that only meters the cheap lines should say so.
Applies to: Every stage in every chain. It is already metered, so no metering work is scheduled against it.
Supervision and management
Overhead · $0.30 per human minute
Estimated
No meter exists
$7,679 · 13.4% of the sample · 778 lines
Marta Oyelaran
Head of GBS Finance
0.33
owner confidence
Basis: Share of direct human minutes. Team leads, quality review and the management layer above the people who touch the work. Apportioned on human minutes because that is the driver most defensible without a customer time-writing system.
Applies to: Every human minute recorded on a chain. One piece of work is scheduled against it: Agree a supervision apportionment basis with Group Finance and write it into the record, owned by Anders Vikstrom.
Facilities and corporate services
Overhead · $0.11 per human minute
Estimated
No meter exists
$2,816 · 4.9% of the sample · 778 lines
Marta Oyelaran
Head of GBS Finance
0.28
owner confidence
Basis: Share of direct human minutes. Premises, corporate IT and the shared functions that support delivery. The weakest line in the ledger and the one most sensitive to a change of basis.
Applies to: Every human minute recorded on a chain. One piece of work is scheduled against it: Replace the facilities composite with the property and corporate services recharge, owned by Anders Vikstrom.
Every cost line declares which of three things it is. Metered means a system counted the quantity and a price per unit exists, so the line is a measurement. Apportioned means a real invoice exists but nothing counts the units it bought, so a divisor was chosen and a different divisor gives a different answer. Estimated means neither the quantity nor the price is observed and both were assumed. The three are not interchangeable and the mix is reported as a headline rather than a footnote, because a total made mostly of estimates is a different object from a total made mostly of meters.

Moving a line from estimated to metered

8 pieces of tracked work, one for each of the 8 rates without a meter. Each names an owner, an effort and a target.

$2.7m
a week of estate cost governed
In progress
3
$839k a week held on an unmetered rate
Blocked
1
$638k a week held on an unmetered rate
Not started
4
$1.2m a week held on an unmetered rate
MT-01Blocked
Connect the payroll extract and replace all three labor rates with actual cost
Estimated to Metered · Marta Oyelaran, Head of GBS Finance · 25 days of effort · target 2026-Q4
$638k
a week, across 409 lines

The single highest-value move on this ladder. Until it lands, every labor line in the product is a market composite and the fully loaded unit cost cannot be defended in front of a controller.

Blocked: Works council consultation in two markets has not concluded, so per-person cost cannot leave the payroll system in any form that supports a rate.

MT-02In progress
Agree a supervision apportionment basis with Group Finance and write it into the record
Estimated to Apportioned · Anders Vikstrom, Group Financial Controller · 12 days of effort · target 2026-Q4
$384k
a week, across 778 lines

Moving this from estimated to apportioned does not make it metered, and the ladder should not pretend otherwise. It makes the basis auditable, which is the honest ceiling for an overhead line.

MT-03Not started
Replace the facilities composite with the property and corporate services recharge
Estimated to Apportioned · Anders Vikstrom, Group Financial Controller · 8 days of effort · target 2027-Q1
$141k
a week, across 778 lines

The recharge already exists in the general ledger. The work is agreeing the driver, not finding the number.

MT-04In progress
Obtain named-user and transaction-band data for the five core platforms
Apportioned to Metered · Declan Whitmore, IT Vendor Manager · 15 days of effort · target 2026-Q4
$237k
a week, across 1,126 lines

Two of the five contracts contain transaction bands that would support a genuine per-stage charge. The other three are flat enterprise agreements and will stay apportioned whatever is done.

MT-05Not started
Move the ten specialist applications onto consumption billing at renewal
Apportioned to Metered · Declan Whitmore, IT Vendor Manager · 40 days of effort · target 2027-Q2
$459k
a week, across 1,048 lines

This is a commercial negotiation rather than an engineering task, and four of the ten renewals fall outside the current planning horizon.

MT-06In progress
Tag orchestration compute by workflow so runtime stops being allocated
Apportioned to Metered · Priya Raghunathan, Platform Engineering Lead · 10 days of effort · target 2026-Q4
$218k
a week, across 1,750 lines

The tagging is straightforward. The value is not the dollars, which are small, but that it removes an allocation from the middle of the chain.

MT-07Not started
Instrument actual handling minutes instead of assuming eleven per touch
Estimated to Metered · Tomas Ferreira, GBS Operations Director · 30 days of effort · target 2027-Q1
$323k
a week, across 272 lines

The rate is only half the labor line. A flat eleven minutes per touch is doing as much damage to the unit cost as the composite hourly rate, and nothing in the ledger currently exposes that.

MT-08Not started
Separate the professional rate by discipline rather than by function
Estimated to Metered · Marta Oyelaran, Head of GBS Finance · 6 days of effort · target 2027-Q1
$287k
a week, across 120 lines

A low-effort improvement that only becomes worth doing once the payroll extract lands, because splitting a composite more finely does not make it less of a composite.

The same ladder in the existing spend ledger

The cost pages already carry a basis field on every recorded spend line. It tells the same story from the other end.

Metered
37.3%
$143.5m
95 of 270 records, 35% by count
Apportioned
12.2%
$46.9m
105 of 270 records, 39% by count
Estimated
50.5%
$194.0m
70 of 270 records, 26% by count
Counting records understates the problem and counting dollars exposes it: the 26% of lines that are estimated carry 50% of the money, while the 39% that are apportioned carry only 12%. Metered lines are the one place the two counts agree, at 35% of the lines and 37% of the money. There is a second problem underneath that one. In this ledger the basis field maps one to one onto the category field — every human review line is estimated, every platform line is apportioned, every inference and storage line is metered — so the basis is a relabeling of what kind of cost it is rather than an independent judgment about how well it is known.

Actions

What is waiting on a person

Each count below is a rate or a piece of metering work that decides whether a cost line is a measurement or an assumption. Moving one line up the ladder changes every unit cost that uses it.

Operations

What this desk is allowed to start

A surface that only reports is not operable. This is the work this page can set in motion, and the bound it runs into.

Trigger and bound

This page can publish every rate the unit costs are built from, state its tier and its apportionment basis on the record, name the person who owns it and the confidence they place in it, and track the work that would move a rate from estimated to metered. It cannot meter anything: no connector is attached to a payroll system, a vendor contract or a metering endpoint, so every tier change here is a piece of scheduled work rather than a switch.

Live observability

What the record shows right now

The metering backlog by state, an exhaustive partition of all 8 tasks into 3 states: every task carries exactly one. There is one task for each of the 8 rates that has no meter, and none for the 4 that already do.

Current distribution

8 metering tasks

Not started450%
In progress338%
Blocked113%

Is policy and strategy coming to fruition

Whether the written intent is holding here

8 of the 12 rates have no meter behind them, and they carry 94.1 percent of the priced sample.

Not holding on the record

The position this layer exists to test is that a cost line should declare its own evidence, and that the mix should be a headline rather than a method note. The declaration holds: every one of the 10,939 cost lines carries a tier, and every rate names an owner, a basis and a confidence. What the declaration reveals is uncomfortable. 42.6 percent of the lines are metered and they carry 5.9 percent of the money, so the reassuring count and the load-bearing dollars point in opposite directions. 5 rates carry an owner confidence below 0.50 and mean confidence across all 12 is 0.60. The backlog that would fix it is real but small: 8 tasks governing $2.7m a week, of which 3 are moving, 1 is blocked and 4 have not started. The existing spend ledger tells the same story from the other end: 37.3 percent of recorded spend is metered by value against 35.2 percent by record count, and its basis field turns out to be a relabeling of its category field rather than an independent judgment. The intent is holding on the disclosure and not on the evidence.