La Bourse / Volume VI / Nº VI.01 / Workbook — the executive

Four colleagues walking toward a long table set with dried grasses, the room bright with sun. — Plate VI.01 · Workbook — the executiveThe Table Before the Meeting.A decision rule is furniture. It decides who can be in the room, how long they will be there, and who is not in the room at all — and it does this before anybody has said a word.

WORKBOOK — THE CORPORATE EXECUTIVE

Chapter VI.01 · Who Decides

For the executive with a P&L, a board, and a delegated authority matrix that nobody has priced. Ninety days to replace it with a schedule that was computed, and to learn what the current one is costing in hours and in wrong decisions.

THE PREMISE, STATED COMMERCIALLY

Your organisation already has a decision rule for everything. It is in the delegated authority matrix, the scheme of delegation, the board charter, the signature limits, the reserved-matters schedule in the shareholders' agreement. Every line in those documents is a threshold, and almost none of them was computed. They were inherited, negotiated under time pressure, or copied from the last deal.

That matters commercially for two reasons, and both show up in the accounts.

The first is cycle time. A threshold set too high does not merely slow a decision; the cost rises non-linearly as the threshold approaches the whole body. On a nine-person board where each member accepts a given proposal with probability 0.60, a 6-of-9 rule expects 2.07 rounds and a 9-of-9 rule expects 99.23. That is not a slower process. It is a different process, and the decisions that would have been slowed are instead not brought.

The second is consolidation. Under IFRS 10, control turns on power over an investee's relevant activities. A minority holder with substantive participating rights can prevent you consolidating; protective rights cannot. The drafting of a consent list therefore determines whether a subsidiary or joint venture appears line by line on your balance sheet or as a single equity-accounted investment — which moves reported gearing and covenant headroom. Your governance schedule is a balance-sheet instrument whether or not anybody in the room is treating it as one.

This workbook makes both visible in one quarter, using data your organisation already has.

PART ONE — DISCOVERY

Days 1–30: what your decisions already cost

Exercise 1.1 — The decision register (1 week, one analyst)

Pull twelve months of board and executive committee minutes. For every decision, record: date first raised · date settled · number of meetings it appeared in · who formally approved · loss-if-wrong.

The last column is the work and the last column is the point. Ask the owner of each decision what the downside case was, in money, and record their number with their name beside it. Do not let anyone leave it blank. A blank there is the finding: it means the threshold protecting that decision was chosen without reference to what it protects.

Exercise 1.2 — Cost the register (1 day)

Cost of one round = attendees × hours × fully loaded hourly rate. For a nine-person board at three hours and £120 an hour, that is £3,240 per round — and the loaded rate for a board is usually several times that, so run yours. Multiply by rounds-to-settle from the register.

Total it. In most mid-sized organisations this is the first time anybody has produced an annual figure for the cost of deciding, and it is invariably larger than the finance function expects. Twelve contested decisions on a nine-person board at 6-of-9 is £80,562 a year on the chapter's parameters; at 8-of-9 it is £551,146; at unanimity £3,858,025.

Exercise 1.3 — Find what is already working (half a day)

Appreciative, and it is not a formality. Identify the three decisions in the register that were made fastest and have held up best. Interview the people involved. What was different — the size of the group, the clarity of the downside, the fact that someone had already done the analysis?

Those three are your positive core. Whatever made them work is already in your organisation and did not have to be installed.

PART TWO — THE ARITHMETIC

Days 31–45: compute the thresholds you should be using

Exercise 2.1 — Band the register (half a day)

Sort by loss-if-wrong. Cut into three or four bands. Do not band by type of decision — capex, hiring, pricing — band by stake. A £4,000 capex and a £4,000 marketing commitment belong in the same band and currently sit under different rules for historical reasons.

Exercise 2.2 — Compute k* for each band (half a day)

Run the minimisation with your own n, your observed p, your round cost and the band's loss. On the chapter's parameters the answer moves like this, and only the stake is moving:

   loss to an excluded member      optimal rule
   £5,000                          4 of 9   (44.4 %)
   £10,000                         5 of 9   (55.6 %)
   £30,000                         6 of 9   (66.7 %)
   £100,000                        7 of 9   (77.8 %)
   £300,000                        8 of 9   (88.9 %)

Compare each band's computed threshold with the one in your current matrix. Two findings are near-universal: the top band is under-protected — the largest decisions run on the same majority as the smallest — and the bottom band is over-governed, paying real decision cost for stakes that do not justify it.

Exercise 2.3 — The delegation break-even (half a day)

This is the number that decides what is reserved and what is not.

   break-even loss  =  decision cost saved / (divergence rate x decisions)

Worked: an assembly of 1,200 deciding directly expects 1.95 rounds at £432,000 a round — £10,114,446 a year for twelve decisions. Delegated to a board of nine at 6-of-9: £80,562. Saving £10,033,884. At a 25 per cent divergence rate the break-even is £3,344,628 per divergent decision. Delegate everything below £3.34m of loss-if-wrong; reserve everything above.

Your divergence rate is estimable from the register: count decisions that were later reversed, materially amended, or that the wider body objected to. It is a crude estimate and it is far better than the zero everyone currently assumes.

Exercise 2.4 — The veto test, item by item (1 day)

For every line on your existing reserved-matters or consent list, run:

   expected cost of the veto  =  P(block) x delay days x value at risk per day

Worked: a 20 per cent chance of a block, a 90-day delay, £4,000 a day at risk gives £72,000, against £150,000 of external cost prevented — a ratio of 2.08 to 1, so the veto is granted. Run every line. Delete the ones that fail. A consent list that has never had a line deleted has never been priced, and it is almost certainly slowing the business for protections nobody values.

PART THREE — DESIGN

Days 46–60: the instrument

Exercise 3.1 — Draft the three classes (2 days, with counsel)

Class A — ordinary operating decisions. Board or executive simple majority. Everything below the lowest loss band.
Class B — decisions above the computed threshold. Board majority plus a named consent: the investor director, a stated percentage of a share class, or the audit committee chair.
Class C — constitutional matters. Shareholder special resolution, which in England and Wales is 75 per cent under section 283 of the Companies Act 2006. You do not have to invent this threshold; the statute supplies it.

Exercise 3.2 — The consolidation memo (1 day, with your auditor)

Before the schedule is signed, walk it through IFRS 10 with the auditor. For each Class B consent right, state whether it is substantive and participating or merely protective, and what that implies for control. Do this at drafting rather than at review: the same commercial protection can usually be written either way, and the difference is visible on the face of the balance sheet.

Exercise 3.3 — The objection grammar (2 hours)

Wherever the schedule uses consent rather than a vote, write the sentence defining an admissible objection — argued, paramount, about whether the proposal will work or will harm the aim. That sentence sets q, and q sets both curves. At q = 0.10 a nine-member body carries in 2.58 rounds against unanimity's 99.23 — faster by 38.4 times — but be honest in the board paper about what that is: the equivalent threshold rule is 6.23 of 9, 69.2 per cent, with external cost of £110,918 a year borne by those whose objection is a preference. Consent is a good instrument. It is not a free one.

PART FOUR — DESTINY AND DELIGHT

Days 61–90: adopt, measure, and hold

Exercise 4.1 — Adopt in one unit (day 61)

One business unit, one quarter. Amend the paper template so every proposal states its class and threshold in the first line. The form enforces the rule; a person does not have to.

Exercise 4.2 — Start the objection log (day 61 onward)

Date · proposal · objection · workability or preference · resolution. Twenty minutes a month. It is the only instrument that detects q widening, which is how a consent rule slides back toward ninety-nine rounds without anyone amending a word.

Exercise 4.3 — The amendment threshold (day 62)

Set the threshold for amending the schedule above every threshold inside it, and set it before anybody wants to amend it. A rule that can be changed by the rule it governs, mid-question, by people who can see which way the vote will fall, is not a rule.

Exercise 4.4 — The day-90 page (day 90)

One page to whoever controls the next governance decision. Four numbers: hours spent deciding before and after · decisions carried · measured q · items deleted from the consent list. Then one sentence on what it implies at the scale of the whole group.

THE FAILURE MODES, NAMED

So you can see them coming

The loss column is filled in by the people who want a particular answer. Every threshold is downstream of loss-if-wrong, and it is the easiest number in the building to inflate. Have it filled in by someone who will not be in the room.

The schedule is adopted and never recomputed. Bands drift in both directions as the business changes. An annual re-run is twenty minutes of arithmetic; an annual renegotiation is a political event and will be won by whoever is most comfortable in that conversation, which is the failure the schedule was built to prevent.

The threshold is changed and the effort is not. The Senate raised its cloture threshold from 67 of 100 to a rule needing 60, moving the blocking coalition from 34 to 41 senators — 20.6 per cent larger — and removed the requirement to hold the floor in the same reform. Blocking got harder on paper and nearly free in practice, and it rose. A rule has two parameters and changing only one of them is how an institution reforms itself backwards.

The system claims to have removed the trade-off. Zappos adopted Holacracy on the claim that it lowered decision cost and external cost at once. In April 2015 it offered severance to anyone who would rather leave: 210 people — 14.0 per cent of about 1,500 — took it, and turnover that year was 29 per cent. The decision cost had not been removed; it had been moved onto every individual's week, where no line item reports it. At one hour per person per week across 48 weeks that is 72,000 person-hours, $2,880,000 a year at a loaded $40 an hour. The group that paid was the operational staff, whose work is not procedural. When a governance change reports lower costs on both curves at once, find the group that stopped being counted.

WHERE YOUR OWN NUMBERS ALREADY SUPPORT THE MOVE

Three places in a normal set of management accounts where the case is already made and nobody has read it as governance.

Cycle time on the capital pipeline. Finance already tracks time from proposal to approval. Split that series by approval threshold and the decision-cost curve draws itself out of your own data, without a model. If proposals needing three signatures take more than twice as long as those needing two, you have measured the steep part of the curve in your own building, and the arithmetic in this chapter merely names it.

Reversals and amendments. Every decision later reversed or materially amended is an observation of divergence between the delegated body and the wider one. Count them over twelve months and you have an empirical divergence rate to put in the break-even instead of the 25 per cent assumed here. In most organisations the true figure is lower, which raises the break-even and means you are reserving more matters than the arithmetic supports — a direct, recoverable cost in cycle time.

The consent list in the last financing. Pull the schedule of reserved matters from your most recent shareholders' agreement or facility. Count the lines. Then count how many have ever actually been invoked. The ratio is usually startling, and it is the cheapest possible evidence that consent rights are being granted by template rather than by price. Every uninvoked line is still costing delay risk on every transaction it touches.

None of the three requires new data collection. All three are in systems you are already paying for, and together they replace every assumed parameter in this workbook with a measured one — which is what turns a model into an instrument.

THE NINETY DAYS ON ONE PAGE

Day	Action	Artifact
1–15	Build the decision register from twelve months of minutes	Register with loss-if-wrong
16–30	Cost the register; interview the three best decisions	Annual cost of deciding
31–45	Band by stake; compute `k*`; run the delegation break-even	The priced schedule
46–52	Veto test every line of the consent list	Veto ledger, items deleted
53–60	Draft Classes A, B, C; IFRS 10 memo with the auditor	Draft schedule + memo
61–75	Adopt in one unit; amend the paper template; start the log	Objection log, week one
76–90	Measure and write the page	One page, one person

BOARD PAPER TEMPLATE

Proposal. Replace the delegated authority matrix with a priced schedule of reserved matters in three classes.

Current position. The organisation makes [N] contested decisions a year at a measured decision cost of £[X]. Thresholds are inherited; none is computed against loss-if-wrong.

Recommendation. Adopt Classes A, B and C with thresholds computed from the attached register. Delete [n] consent items failing the veto test. Retain [n] whose prevented external cost exceeds their expected delay cost by more than 2 to 1.

Accounting effect. Class B consent rights have been reviewed with the auditor against IFRS 10. [Statement on control and consolidation.]

The number that decides it. Decision cost saved by delegating each class, divided by divergence rate × decisions × loss-if-wrong. Above one, delegate. Below one, reserve.

Review. Annual recomputation, reported with the measured objection rate.

APPRECIATIVE QUESTIONS FOR YOUR LEADERSHIP TEAM

Which decision of the last two years do we all — including those it went against — still think was decided well? What was different about it?
Where do we already decide fast and safely without consulting the matrix, and what makes that safe?
If every board paper stated its own threshold in the first line, what would our meetings be about instead?
What is the largest loss-if-wrong in this business, and what rule currently governs it?
Which consent right on our schedule are we most confident is worth its delay — and how would we show that to an investor?
If our governance costs and our consolidation position were both on the front page of the annual review, what would we want them to say?