Prove · baseline BL-028

Technical Debt

Technical Debt, all regions, as run before the agents were switched on. Owner at the time of measurement: Director, Engineering Effectiveness.

Baseline register

Status

Locked and signed

confidence 0.80

Cost per unit

$42.62

162,539 units in the window

Cycle time

28.8 hrs

pre-agent, measured end to end

Quality

91.5%

first-pass yield

Claimed on this baseline

$13.70m

$2.51m independently verified

Posted to the ledger

2 of 5

cash lines with a journal reference

How it was measured

Invoice reconstruction

Method

Invoice reconstruction

Method note

Vendor invoices, contractor timesheets and purchase orders re-added for the window and tied back to the posted ledger balance.

Window

FY24 Q4, thirteen weeks

Sample

4,876

Confidence in the method

0.80

Sign-off

Who accepted this as the number to beat

Status

Locked and signed

Signed by

Marta Alvarez

Role

Cost Accounting Manager

Signed

18.4 months ago

Caveat carried with the number

Locked and signed. The window is a full quarter and the method leaves an audit trail that can be re-performed.

Where the benefit lands

Account and cost center this baseline reconciles to

General ledger account

6120 · Contract and temporary labor

Agency and contractor spend. The cleanest place to evidence a reduction because it falls out of the invoice run within one period.

Corporate GL · Personnel · owner Ana Duarte, Group Financial Controller

Cost center

CC-3010 · Engineering

Global · owner Daniel Okoye · annual budget $38.11m

Engineeringfunction view EV-BL-028evidence chain

Benefit lines resting on this baseline

5 lines · $13.70m claimed · $9.94m realized · $2.51m verified

Period	Category	Claimed	Realized	Verified	Status	Variance	Journal
FY25 Q3	Labor cost	$1.77m	$1.24m	$1.01m	Partly verified	Timing Benefit landed one period later than claimed. The run rate is intact; the period attribution was wrong.	JV-2025Q3-0060
FY25 Q4	Labor cost	$2.56m	$1.87m	$1.50m	Partly verified	Unexplained Internal audit could not trace part of the realized figure to a driver. The untraced amount stays in the variance column.	JV-2025Q4-0061
FY26 Q1	Labor cost	$3.28m	$2.62m	not verified	Not verified	Volume Seasonal volume ran ahead of baseline, which flatters the realized figure rather than the claim.	not posted
FY26 Q2	Labor cost	$3.91m	$2.81m	not verified	Not verified	Method mismatch The baseline was measured by survey and the realized figure by system extract. The two are not comparable without a bridge, and no bridge has been built.	not posted
FY26 Q3	Labor cost	$2.19m	$1.40m	not verified	Period open	Period open The period is not closed. The realized figure is a run rate, not a result, and nothing in it has been verified.	not posted

What this record is, and what it is not

A benefit realization ledger reconciled to a general ledger needs the customer’s actual baseline figures and account structure. Modeled numbers demonstrate the instrument and prove nothing about the business case.

The measurement window ran FY24 Q4, thirteen weeks over a sample of 4,876. That sample size is on the page because a baseline drawn from twelve cases and one drawn from forty thousand are not the same instrument, and averaging them into a single confidence figure would hide which one you are looking at.

2 of 5 cash lines resting on this baseline carry a journal reference. A benefit line with no journal reference has not touched the general ledger, so it is shown as claimed rather than counted as realized cash.