Cost-based control systems can only see work that carries a cost line, and the work of deciding reliably carries none — so the largest recurring consumer of schedule in most enterprises is invisible to every instrument that governs it.
Ask the finance function what the enterprise spent last quarter on steel, contract labour, freight and electricity, and a figure arrives inside the hour. Ask what it spent waiting on its own decisions, and no figure arrives at all — not a rough one, not a range. The question does not resolve into a query anyone can run.
That is structural, not an accounting oversight. Every enterprise control system in general use rests on one foundation: work is decomposed into units, each unit takes a code, the code carries an estimate, the estimates roll up into a baseline, and everything the board sees is arithmetic on those codes. The code is the unit of visibility. Work that has one is measured, owned and defended. Work that has none does not exist as far as the instrument is concerned, however much of the calendar it consumes.
Deciding has no code. It appears in no breakdown structure, no cost baseline, no earned value calculation and no variance report. And it is ERANORTH's contention that in most large enterprises the interval between the moment a question becomes answerable and the moment somebody answers it is the largest recurring consumer of elapsed time in the delivery portfolio — larger than any technical activity, and far larger than the schedule contingency held against it.
This matters because the missing number changes where blame lands. A control system that cannot see the wait attributes its effects to whatever it can see, which is always the delivery unit. Pressure then falls on the part of the system that was working, and none on the part that was not.
The Strategic Context
The teaching material of the project control discipline is unusually clear about its completeness rule: the structure that decomposes the work must contain one hundred per cent of it, including, in the strongest formulations, the management of the project itself. Omitting management effort is repeatedly named a classic failure and a corruption of the cost baseline.
Then the worked examples arrive. Across a set of exemplar breakdowns in one body of that material, some carry a management branch and some do not, and those that do not are declared complete and audit-ready regardless. Not one carries a branch for the decision and approval cycles between phases. An accompanying cost baseline lists its cost-bearing activities item by item — design, prototyping, drafting, tooling, production, freight — and holds no line for approval, review, certification or waiting.
The same document treats a multi-week shipping leg as a cost-bearing activity, because the vessel generates an invoice, and the wait for a signature as nothing, because a signature generates none. Time enters the cost baseline exactly when somebody bills for it, and never otherwise.
A contradiction that survives inside a discipline's own instructional material is worth more than an anecdote: the rule is known, the failure is named, and the practice breaks it regardless, which makes the breakage structural rather than careless. And the structure that omits the decision cycle is the one that generates the price, the schedule and the board report.
Why the Variance Report Blames the Wrong People
Earned value is the proportion of a work package completed multiplied by that package's budget. A team waiting on a determination completes nothing, so it earns nothing. Planned value keeps accruing, because the calendar does not pause. Schedule variance therefore goes negative, while cost variance stays flat or improves, because idle people on other charge codes consume no project budget.
Read at a governance meeting, that pattern says something specific and entirely wrong: this team is slow but frugal. The correct reading — the team is idle and the decision is late — is unavailable, because the determination appears in neither figure. Two different situations produce one signature, and no field exists to separate them.
Reframing the Issue
Delay is the wrong frame, because it implies a lapse. In a well-run enterprise most decision latency is not a lapse at all: it is the price of assurance, deliberately purchased and never priced.
Every governance design choice is a purchase of confidence paid for in elapsed days. How often the investment committee sits. How many items it takes in a sitting. What threshold escalates a variation to a board. How many signatures a specification change needs, and in what order. Each setting buys scrutiny, and each is paid for in calendar time multiplied by whatever the enterprise forgoes per day.
The same material shows the blind spot precisely. It diagnoses a change control board that clears too many items too quickly as governance theatre, and prescribes a hard cap on items per sitting. Nowhere does it observe that a cap on a periodic sitting is a queue, that the queue has a length, or that the length carries a cost falling on parties who were not in the room when the cap was set.
The question is not how to stop waiting, but what the enterprise buys with the wait, whether that purchase is worth making, and who is authorised to make it.
The Three Places Decision Time Hides
Latency does not vanish. It is absorbed, and it is absorbed in three places, each of which corrupts a different instrument.
Inside the duration of the activity that was blocked
When a work package cannot start until an approval lands, the wait is folded into that package's actual duration rather than recorded separately. The package overruns, the overrun is explained, the explanation is filed. The corruption compounds: the next estimate for that work is anchored on the last actual, and carries an unlabelled latency premium. Over several cycles the estimating base absorbs the average behaviour of the enterprise's own approval queue, and the estimates it produces look conservative to everybody and inexplicable to nobody.
Inside the indirect recovery line
Where waiting extends total duration, it converts into extra periods of overhead recovery. This is the only route by which latency reliably becomes money, and it destroys attribution on the way. An enterprise reviewing the overrun sees an indirect line that grew, concludes its overheads are too high, and attacks its support functions. The queue that caused it never had a code, so it is not in the room.
Inside a queue nobody drew
Decisions do not arrive when they are needed. They arrive when the forum sits. The interval between sittings is therefore a floor under latency, and it appears in no plan and no risk register: half an interval on average, a full one when the paper misses the deadline, two or more when the item is deferred.
Deferral is the sharpest case. The integrated change control workflow in that material has three exits from its decision node — approve, defer, reject. The first two close the item and produce a record. The third parks it in a backlog with no owner, no expiry and no cost attached, leaving the clock running indefinitely on an item that no longer appears on any agenda.
Consider a materials recovery operator reconfiguring a sorting line — hypothetical, but a shape that recurs. Three determinations sit open: the contamination threshold to accept, the offtake specification to write against, and the capital release for an optical sorter. While they sit, the plant runs at its old recovery rate. The operator knows its gate fee, its tonnage and its offtake price, so the cost of a week of waiting is already calculated somewhere in the commercial ledger. It appears in no project report, because the project's cost system tracks the reconfiguration and not the plant.
Decision Framework
The instrument that closes the gap is the decision-latency ledger: a standing register, one row per determination the portfolio is waiting on, priced at a daily rate. Not a RAID log and not an action list: a queue with a meter on it.
| Field | The rule that makes it usable |
|---|---|
| Decision and decider | A question, with one named individual who could sign it today. If none can be named, that is the first finding. |
| Requestable date | The date every input the decider needed already existed, evidenced by the papers. |
| Submission date | The date the paper reached the decider. Requestable to submission is preparation lag, and it belongs to delivery. |
| Determination date | Submission to determination is deliberation lag, and it belongs to governance. Never sum the two into one figure. |
| Daily consumption rate | What the enterprise forgoes per day the item stays open, on one of three bases: capacity held idle, margin forgone at the pre-decision run rate, or exposure that grows with age. |
| Exit code | Determined, rejected or deferred — with a mandatory next-review date on every deferral. |
Four thresholds make it operational rather than decorative. Any decision whose accumulated latency cost has passed the value at stake inside it is a governance defect, not a delivery problem, and that authority should be delegated downward permanently. Any decider whose deliberation lag routinely exceeds one forum cycle is over-subscribed, and the finding belongs in their objectives, not in a delivery exception report. Any deferral without a next-review date closes by default at the next ledger review. And the daily rate is set against the cost of a standing delegation: where rate multiplied by typical lag exceeds what delegated authority costs to assure, the enterprise is buying scrutiny above market.
Ownership decides whether it survives: the ledger belongs with finance and reports into the same forum as the cost baseline. Held by the delivery office, it reads as a complaint and is discounted as one.
Two boundaries matter here. This article concerns work the control system cannot see because it carries no cost code; what sets the size of the units it can see at all, and therefore the earliest date any problem inside one can surface, is not covered here because it belongs to [Related article: The Rule of Thumb Behind Your Detection Time]. And rolling this ledger up across more than one delivery unit raises an arithmetic problem avoided here by holding to one basis — that defect belongs to [Related article: Three Scales, One Word].
From Strategy to Execution
Immediate, within thirty days. Instrument the queue that already exists. Take the fifteen determinations now blocking work anywhere in the portfolio, backfill the four dates, and price each at a daily rate finance will defend. Publish two numbers: open latency days, and accumulated cost. Completeness is not the point; the number's job is to make the category real.
Medium term, one to two quarters. Change the decomposition. Give decision and approval cycles a node in the breakdown structure, a line in the cost baseline, a duration and a named owner, so they become visible to the instruments that govern everything else. Reset delegation thresholds wherever the arithmetic shows the queue costs more than the assurance it buys. Where a forum's cycle time exceeds the useful life of its papers, change the cadence, not the papers.
Long term, beyond a year. Make latency a standing line in enterprise cost reporting, and deliberation lag a performance measure for the decider rather than an exception for the delivery unit. Then rebuild the estimating base: an enterprise whose record separates work from waiting can estimate work honestly for the first time.
Signals to Monitor
Negative schedule variance alongside flat or favourable cost variance, appearing at once across delivery units with nothing technical in common, is the signature of a governance bottleneck rather than a delivery one. Watch also for indirect recovery rising while direct cost holds flat; for a rising deferral rate at any standing board; for a falling share of items determined at first appearance on an agenda; for decisions whose recorded owner is a committee; and for requests routed as clarifications rather than formal changes, which shows the queue bypassed rather than shortened.
The latency this ledger prices is chronic and cumulative rather than catastrophic, and the separate defect by which conventional risk scales cannot represent the rare event capable of ending an enterprise is not treated here, because it is the subject of [Related article: The Floor Beneath Your Risk Scale].
Questions for the Leadership Team
- For the ten determinations our portfolio is waiting on, what date did each become answerable, and what date did the paper reach the person who can answer it?
- What share of the elapsed duration of our three most recently completed initiatives was spent waiting rather than working, and where in the cost baseline does that time appear?
- Which of our standing approval forums has a cycle time longer than the useful life of the papers it reviews?
- What is the daily forgone margin on our largest operating asset while it waits on the capital decision now in front of us, and who calculated it?
- Of the items our change and investment boards deferred in the last twelve months, how many carry a next-review date, and how many remain open?
- If every decision below our authority threshold were delegated one level down, what value would that place at risk, against the latency cost we would stop paying?
Closing Perspective
The pattern is not confined to plant. A commercial insurer holding a coverage position at a technical referral panel pays in a different currency: every day the file ages, mitigation and hire costs accrue and the claimant's tolerance falls. The cost lands on the claims book, while the panel is measured on the quality of its determinations and never on their timing — the arrangement that guarantees the queue will lengthen until something else breaks.
The enterprise is already paying. The choice in front of the executive is not whether to buy decision time, because that purchase is made daily at a rate nobody set. The choice is whether to see the invoice, and then to accept what it shows: that some of the most defensible governance in the organisation is also the most expensive, and that those who set the cadence are almost never those who pay for it.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.