Operate · item 23
Executive decision queue
This is not the operational inbox. Nothing here is work that an agent could finish or a manager could approve — these are the decisions that need somebody with the authority to spend money, accept risk or change the organization. There are twelve of them. Each carries what it is costing per week to leave it open, and where that number cannot be honestly calculated it is left at zero with the reason written out.
Health
What is waiting on an executive
Decisions open, scheduled and overdue, with the work blocked behind each one.
Decisions waiting
12
12 in the queue in total
Cost of delay
$238.0k/wk
across 6 priced decisions
Past due
3
due date already gone
Deliberately unpriced
6
a weekly number would be invented
Items blocked behind them
114
workflows, agents and claims held up
Decided
0
nothing in this queue has been closed out yet
What the delay costs
Weekly, for the decisions where a defensible number exists
Two source systems have no feed at all$84.0k/wk
Manual reconciliation of the two unconnected estates runs at 214 person-hours a week at a blended 393 per hour. The figure is a modeled labor cost, not a booked loss.
Redeployment of released capacity has no approved plan$63.0k/wk
Cost of holding released capacity in place without a destination, at the average loaded rate for the affected grades. Modeled from the headcount movement already in the platform.
Fifty-eight sub-functions are claiming benefit with no locked baseline$41.0k/wk
Measurement cost of the program spread across the quarters it would take to complete, expressed weekly. Delay does not reduce the cost, it moves it.
Order to Cash inference is running at the cap$28.0k/wk
Queued cash application work at the tower service credit rate, plus the manual clearing effort it generates downstream.
Nine reconciliation breaks have no owner$12.5k/wk
Carrying cost of the disputed value at the internal cost of capital, plus the clearing effort that grows with age. Modeled, not booked.
Seven controls have never been tested$9.5k/wk
Internal audit effort deferred and then repeated, plus the remediation window that shortens each week the test is not run. Modeled from audit day rates.
Decisions carried at zero
Where a weekly cost would be fabrication
German candidate records sitting in Singapore without a sanctioned basis
There is no weekly cash cost. The exposure is regulatory and does not accrue linearly, so a per-week figure would be invented. It is shown as zero deliberately rather than filled with a plausible number.
No agent carries an independent conformity assessment
No cash cost accrues weekly. The cost is the option value of being able to answer a customer or regulator question with an external document rather than an internal one. It is not quantified here because any number would be invented.
Promotion slate agent is running at an adverse-impact ratio of 0.74
Every week the agent runs, another promotion cycle passes through a decision point with a breached ratio. The harm is to people, not to a cost line, so no weekly cash figure is shown.
Platform cost treatment, capitalize or expense
An accounting policy choice with no weekly cash consequence. Delay compresses the time available to restate prior periods if capitalization is chosen.
Access revocation runs unmonitored and already has an appeal against it
No weekly cash cost. The exposure is that an appeal is already recorded against a decision point with no readings to appeal against.
Authority to step an agent beyond A3 is undefined
No weekly cash cost. Step-ups are simply not happening, which is the safe failure mode and the reason this has been allowed to sit.
6 of the 12 open decisions carry no weekly number. A regulatory exposure or an unassessed model does not have a defensible per-week price, and attaching one would make the total look precise rather than make the decision clearer.
EXD-01InvestmentPast due9 days past due
Two source systems have no feed at all
Fund the bank gateway and electronic lab notebook connectors, or accept that every figure downstream of them is modeled indefinitely?
$84.0k/wk
cost of delay
Avery Chen
Chief of Staff to the CEO, on behalf of the CFO
Options on the table
Fund both connectors
One-off build cost and a security review each. Removes the modeling caveat from 2 of 26 feeds.
Fund the bank gateway only
Clears the treasury reconciliation set. The lab notebook stays unconnected and R and D benefit stays unclaimable.
Accept and label
No spend. The product keeps saying these two are not connected, which is honest but permanent.
Recommendation
Fund the bank gateway now and hold the lab notebook until the R and D scope stops moving.
The treasury set carries the largest unreconciled value and the connector has a known integration pattern. The lab notebook has no stable scope to build against yet.
raised 2.0 months ago · 3 items blocked
Blocks the treasury reconciliation set, the research and development lineage facts held at 0.3 confidence, and any benefit claim on the R and D sub-functions.
EXD-12InvestmentPast due21 days past due
Seven controls have never been tested
Fund an independent test round for the untested controls, or leave them recorded as untested?
$9.5k/wk
cost of delay
Grace Abbott
Head of Internal Audit
Options on the table
Fund an independent round
Converts seven untested controls into tested ones, pass or fail. Some will fail and that will be visible.
Owner tests, audit reviews
Cheaper and faster. Keeps the independence weakness that already applies to two controls.
Leave as untested
No spend. The control register keeps showing seven untested, which it does today rather than showing them green.
Recommendation
Fund an independent round and start with the two that are currently tested by their own owner.
A control tested by the person who runs it is the finding an external reviewer raises before they look at anything else.
raised 2.3 months ago · 9 items blocked
Seven of twenty-four controls have never been tested. Two more are tested by their own owner, which is not independent.
EXD-04PolicyPast due2 days past due
Promotion slate agent is running at an adverse-impact ratio of 0.74
Suspend the agent, restrict it to recommendation only, or accept the reading and continue while the cause is investigated?
not priced
cost of delay
Priya Raman
Chief People Officer
Options on the table
Suspend the agent
Slate preparation reverts to manual for the next cycle. Adds roughly nine days to each slate.
Restrict to recommendation only
A named human composes every slate. The agent output becomes advisory and is logged as advisory.
Accept and investigate
The breach stays live on the register while the cause is worked. Requires a written acceptance from the Chief People Officer.
Recommendation
Restrict to recommendation only until the ratio recovers above threshold for two consecutive periods.
A sustained four-period drift is not noise. Restriction keeps the throughput benefit while removing the agent from the decision itself.
raised 23 days ago · 2 items blocked
Two open threshold-breach alarms against this decision point, one sustained across four consecutive periods.
EXD-09OrganizationOpen17 days to decide
Redeployment of released capacity has no approved plan
Approve the redeployment and reskilling plan for the capacity the program has released, or let it resolve through attrition?
$63.0k/wk
cost of delay
Priya Raman
Chief People Officer
Options on the table
Approve the plan
Commits reskilling spend and a named destination per role. Makes the benefit claim credible because the capacity actually goes somewhere.
Resolve through attrition
No spend and no announcement. The benefit realizes slowly and unevenly, and the ledger will show it as timing variance for several quarters.
Recommendation
Approve the plan.
Benefit claimed on released capacity that has no destination is the claim internal audit rejects first. It is already the largest single source of timing variance in the ledger.
raised 52 days ago · 14 items blocked
Every function reports released capacity and a redeployment intent. None of it is backed by an approved plan with dates.
EXD-03Risk acceptanceOpen22 days to decide
No agent carries an independent conformity assessment
Commission third-party assessment for the nine high-risk agents, or continue on self-assessment and say so publicly?
not priced
cost of delay
Avery Chen
Chief of Staff to the CEO
Options on the table
Assess all nine
Highest cost and the longest calendar. Produces the document a procurement reviewer actually asks for.
Assess the three people-affecting agents first
Covers the screening, promotion-slate and performance agents. Leaves six high-risk agents self-assessed.
Stay self-assessed
No spend. The product must keep stating that zero agents carry an external assessment, which it currently does.
Recommendation
Assess the three people-affecting agents first.
Those three already have open adverse-impact alarms and a live appeal against one of them. Self-assessment on an agent with an open alarm is the weakest position in the estate.
raised 2.9 months ago · 9 items blocked
Nine agents classified high risk. Thirteen agents are self-assessed and three have not been assessed at all.
EXD-06PolicyOpen30 days to decide
Platform cost treatment, capitalize or expense
Capitalize the agent platform build against an internal-use software asset, or expense it as incurred?
not priced
cost of delay
Ana Duarte
Group Financial Controller
Options on the table
Capitalize
Improves the in-year operating result and creates an asset to be tested for impairment. Requires the development phase to be separable, which it partly is not.
Expense as incurred
Simplest and hardest to challenge. The in-year cost stays visible in full.
Recommendation
Expense as incurred.
The development and operating phases are not cleanly separable in the current time records, and a capitalization that cannot be substantiated is a finding waiting to happen.
raised 18 days ago · 1 item blocked
Blocks the close of the 6420 and 6430 account mapping for the platform cost centers.
EXD-08Risk acceptanceOpen11 days to decide
Access revocation runs unmonitored and already has an appeal against it
Extend adverse-impact monitoring to the access revocation decision point, or formally record that it will not be monitored?
not priced
cost of delay
Yusuf Adeyemi
Head of Identity and Access
Options on the table
Extend monitoring
Requires cohort data the identity system does not currently carry. Needs a data protection assessment of its own.
Record a monitoring exclusion
Honest and fast. The register will show one decision point deliberately outside monitoring, with a named accepting officer.
Recommendation
Extend monitoring.
An appeal exists. A decision point with a live appeal and no readings cannot be defended as out of scope.
raised 15 days ago · 1 item blocked
One of eight people-affecting decision points is unmonitored. Appeal APL-0013 is recorded against it.
EXD-11PolicyOpen26 days to decide
Authority to step an agent beyond A3 is undefined
Who signs an autonomy step-up past A3, and what evidence must be in front of them?
not priced
cost of delay
Avery Chen
Chief of Staff to the CEO
Options on the table
Executive committee signs
Highest assurance, slowest cadence. Step-ups become a quarterly event.
Function leader signs with risk countersignature
Faster and still two-person. Requires a defined evidence pack per step-up.
Recommendation
Function leader signs with a risk countersignature and a fixed evidence pack.
The evidence spine already produces everything the pack would need. Routing every step-up to the executive committee would make the authority register decorative.
raised 44 days ago · 6 items blocked
Six agents are at the top of their approved band with a completed evaluation and nowhere to go.
EXD-05PolicyScheduled13 days to decide
Fifty-eight sub-functions are claiming benefit with no locked baseline
Fund the measurement program to close the baseline gap, or stop reporting benefit on the unmeasured sub-functions?
$41.0k/wk
cost of delay
Ana Duarte
Group Financial Controller
Options on the table
Fund the measurement program
Closes the gap over three quarters. Requires finance business partner time that is currently committed to close.
Stop reporting benefit on unmeasured scope
The headline benefit number falls immediately and visibly. It becomes defensible.
Measure the largest twenty only
Covers most of the value at a fraction of the effort. Leaves a long tail permanently unmeasured.
Recommendation
Measure the largest twenty and stop reporting benefit on the remainder.
The tail is individually small and collectively unverifiable. Reporting it unverified is worse than not reporting it.
raised 34 days ago · 58 items blocked
Fifty-eight of 154 sub-functions have no locked baseline. Their claimed benefit is excluded from the ledger entirely rather than shown unverified.
EXD-10InvestmentScheduled3 days to decide
Order to Cash inference is running at the cap
Raise the Order to Cash inference cap, hold it and accept the queue, or move the workload to a cheaper model?
$28.0k/wk
cost of delay
Sofia Lindqvist
Finance Business Partner, Operations
Options on the table
Raise the cap
Immediate relief. Increases the 6430 run rate and the tower unit cost.
Hold and queue
No cost increase. Cash application slows and days sales outstanding drifts.
Move to a cheaper model
Lower unit cost, and a measured accuracy drop that has to be re-baselined before the benefit can be claimed.
Recommendation
Raise the cap for one quarter while the cheaper model is evaluated against the locked baseline.
Switching model without re-baselining would break the comparison the whole ledger rests on.
raised 8 days ago · 1 item blocked
One hard cap at the alert threshold. Cash application queues when the cap is reached.
EXD-07OrganizationScheduled4 days to decide
Nine reconciliation breaks have no owner
Assign the unowned breaks to a named person, or route them to a standing clearing desk?
$12.5k/wk
cost of delay
Ana Duarte
Group Financial Controller
Options on the table
Assign to named owners
Clearest accountability. Adds work to people who already own breaks.
Stand up a clearing desk
Single queue with a service level. Costs a small standing team and takes a quarter to be effective.
Recommendation
Assign now, stand up the desk for the next quarter.
An unowned break at 129 days is an audit finding today. The desk is the durable answer but it does not clear the current backlog.
raised 29 days ago · 9 items blocked
Nine open breaks with no owner, carrying 1.09 million in disputed value. Twenty open breaks are past ninety days, the oldest at 129.
EXD-02Risk acceptanceScheduled6 days to decide
German candidate records sitting in Singapore without a sanctioned basis
Repatriate the placement, obtain a transfer basis, or record a formal risk acceptance with an owner and an expiry?
not priced
cost of delay
Nadia Kovac
Data Protection Officer
Options on the table
Repatriate to Frankfurt
Removes the exposure. Adds latency to candidate screening in the Asia Pacific window and needs a migration window.
Obtain a transfer basis
Keeps the placement. Requires a transfer impact assessment and supplementary measures before it can be relied on.
Record a time-boxed acceptance
Fastest. Needs a named accepting officer and an expiry date, and it will appear on the register as accepted rather than resolved.
Recommendation
Repatriate. The latency cost is measurable and small; the alternative leaves an unsanctioned placement on the register.
The screening agent is already classified high risk. Carrying an unsanctioned placement underneath a high-risk agent is the combination an inspector opens with.
raised 41 days ago · 1 item blocked
One unsanctioned residency placement covering 15,001 records. It is the only unsanctioned placement in the estate.
Actions
What is waiting on a person
Every item here is a decision only an executive can take, and every week it waits has a price on 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 desk can raise a decision, attach the evidence and the options to it, price the cost of delay and route it to a forum. It cannot take the decision, set policy, or reallocate budget — that is the point of the page.
Live observability
What the record shows right now
Where the decisions sit.
Current distribution
12 decisions
Scheduled433%
Open542%
Overdue325%
Is policy and strategy coming to fruition
Whether the written intent is holding here
4 of 12 decisions have a forum date. 3 are overdue, holding 114 work items.
Not holding on the record
The operating principle is that agents escalate cleanly and executives decide quickly. The first half works. The second is where the estate loses time: 3 decisions are overdue and 5 are unscheduled, together carrying a recorded cost of delay of $156,500 per week and blocking 114 items of work. The cost of delay figures are the ones authored on each decision by the person who raised it; they are estimates, not measured losses.
What this page is, and what it is not
The queue, the ownership, the cost-of-delay basis and the option framing are real structures. The twelve decisions themselves are the genuine open questions this estate raises — every one of them traces to something visible elsewhere in the application: an unconnected connector, an unowned reconciliation break, a fairness ratio below threshold, an untested control, a budget over its cap.
$238.0k per week is the sum across 6 decisions where a weekly cost can be derived from something measurable. It is not the cost of the whole queue. 6 decisions are carried at zero because inventing a number for them would make the total look more precise than the evidence allows.
3 decisions are already past their due date. Leaving them visible and late is the point — a decision queue that quietly reopens its own deadlines is a to-do list.