End-to-end outcomes · contested ownership
The boundary dispute register
Where a chain waits at a handoff, somebody owns that wait. These 9 entries are the boundaries where two functions each say the other one does. Both positions are recorded in the words the function used, an arbiter is named, and a decision is due on a date. An outcome is assembled, not recorded. No source system in this estate stamps a single reference across a purchase requisition, the order raised against it and the receipt that closes it, so the chains on these pages were stitched together after the fact. Where both documents sit in the same source system the chain is keyed and the join is exact. Where it crosses a system boundary the chain is matched on the party, the market, the region and whether the timing is plausible, and it is labeled inferred with its confidence shown. Where the two sides name no party in common the only evidence left is place and timing, and the chain is labeled cohort and confidence is capped low on purpose. A cohort chain is a hypothesis about which documents belong together. Read it as one.
The register by status
An exhaustive split of all 9 entries into 4 states
Both sides, in their own words
9 contested boundaries
Nothing below is summarized into a neutral house position. Each function’s claim is recorded as that function put it, because the disagreement is the finding.
Who owns the wait between the payables platform and treasury
BD-0001 · Invoice to payment · Finance → AI GBS
Delivery raised against invoice issued
BD-0002 · Order to cash · not anchored to a mapped handoff
The mapped flow and the measured chain disagree about the route
BD-0003 · Hire to productive · HR → IT
Whether an unevidenced deduction is a marketing cost or a revenue loss
BD-0004 · Deduction to settlement · not anchored to a mapped handoff
The certification stage sits outside the close
BD-0005 · Close to report · not anchored to a mapped handoff
A receipt that never posts
BD-0006 · Requisition to receipt · Procurement → Supply Chain
Artwork approval sits between two functions and belongs to neither
BD-0007 · Concept to shelf · not anchored to a mapped handoff
Duplicate entry between the payables platform and the bank gateway
BD-0008 · Invoice to payment · not anchored to a mapped handoff
Whether an agent failure is an incident or a model problem
BD-0009 · Query to resolution · not anchored to a mapped handoff
Why the register exists at all
Read before treating any figure here as a total
A function-level report can only ever attribute a wait to the function that was holding the work. A chain that stops between two functions therefore lands nowhere: neither scorecard degrades, and the outcome absorbs it. The register is the mechanism for forcing that hour onto somebody, and its own numbers say the mechanism is barely used — 2 of 9 entries carry a ruling.
The largest priced claim is Whether an unevidenced deduction is a marketing cost or a revenue loss at $1.3m a week between Marketing and Finance. Costs are derived from the modeled holding rate applied to a measured delay, so the delay is an observation and the dollars are an assumption. The boundaries themselves are priced in full on the handoff economics page, and the chains that cross them are assembled on the outcomes page.
Actions
What is waiting on a person
Every row in this register is a delay that two functions both believe belongs to the other. Until somebody rules, the hours sit on neither scorecard and the outcome carries them anyway.
Rule on a boundary dispute
3 open, 3 scheduled and 1 escalated. Each has both positions written down and a named arbiter, so what is missing is the decision, not the evidence.
Map a dispute onto a named handoff
The argument is not anchored to a boundary in the cross-flow map, so fixing the flow would not obviously fix the dispute.
Put a number on a dispute nobody has priced
The disagreement is real but no dollar figure has been attached to it, so it will always lose a prioritization argument to one that carries a number.
Freeze a handoff while its ownership is contested
Work keeps crossing this boundary at the same rate while the two functions argue about who owns the wait it creates.
Take an escalated dispute to the executive
The named arbiter could not settle it at their level, so it is waiting above them while the delay continues to accrue.
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 record where two functions disagree about who owns a delay, hold both positions in their own words, name an arbiter and a decision date, and price what the disagreement is costing where the delay was measured. It cannot make the ruling, move the accountability or reassign the handoff. The decision is a person’s to make.
Live observability
What the record shows right now
The register by status, an exhaustive partition of all 9 disputes into 4 states: every dispute carries exactly one, and only the decided ones have a ruling recorded against them.
Current distribution
9 disputes
Is policy and strategy coming to fruition
Whether the written intent is holding here
2 of 9 boundary disputes have been decided; the other 7 leave their delay on neither function's scorecard.
Not holding on the record
The design intent is that no delay is ownerless: every hour a chain spends waiting should land on somebody’s scorecard, and where two functions disagree an arbiter settles it on a date. The register shows the mechanism exists and is barely used. 2 disputes are decided, 7 are not, and the largest single claim is $1.3m against Marketing and Finance. The widest touches 300 assembled chains, and the affected sets overlap, so the counts must never be added into a single population. 3 entries record no measured delay and 4 carry no dollar figure, and they are not the same entries: the hours and the cost were recorded independently, so an entry can have one without the other. An entry with neither is not one that costs nothing; it is one nobody has measured. The claims are per-entry, against overlapping populations, so there is deliberately no total anywhere on the page. That is the finding, and it is a finding about the operating model rather than about the people arguing.