Haute Lumière

Commerce · VI.10 · MMXXVI · daylight

La Bourse  /  Volume VI  /  Nº VI.10  /  Workbook — the executive

WORKBOOK — THE EXECUTIVE

Chapter VI.10 · Auditing a Governance System

For the person with a board, a committee structure, a delegated authority matrix and a governance statement in the annual report. You already pay for governance. This workbook is about finding out what you are getting for it, in the language of the firm, using documents you already produce.

---

WHY THIS IS A FINANCE PROJECT, NOT A COMPANY-SECRETARIAL ONE

Your organisation already runs a governance process and already reports on it. The report is almost certainly a description: committees listed, attendance recorded as a percentage, terms of reference confirmed as reviewed. It is truthful and it is not evidence.

The reason to do better is not conscience. It is three specific costs you are already carrying and cannot currently see:

1. Decisions that were taken and never implemented. Every one consumed executive time, generated a workstream, and produced nothing. You have no figure for this because your governance reporting has no field for it.
2. Decisions taken by exception. Each is a control that did not run. Your internal audit function reports on control failures in processes and is silent on control failures in decision-making, which is the most expensive process in the firm.
3. Delegated authorities that nobody uses. Dormant provisions are where the governance overhead lives: committees that meet to confirm they have nothing to confirm, approvals that have never once been withheld.

All three are computable from documents you have. The whole exercise is extraction and arithmetic, not a new framework, and it costs one analyst three weeks.

---

PART ONE — DISCOVERY

Weeks 1–2: establish what is working, with evidence

Exercise 1.1 — The three best decisions (half a day, with the board chair)

Before any measurement, identify the three decisions of the last three years that the board would defend most confidently. For each, reconstruct from the record: when the demand first appeared in writing, who moved it, how many meetings it took, what changed between first appearance and approval, and what made it possible.

You are looking for the mechanism, and it is nearly always one of four: early written circulation, a named dissenter who was listened to, a decision deliberately deferred once, or a sponsor with unambiguous authority. Whichever it is, that mechanism is your baseline. Every finding later in this workbook is measured against a structure that demonstrably works in your own firm.

Exercise 1.2 — The document inventory (one day)

You need five things, all of which exist:

If minutes record outcomes but never positions, note it now: half the six measures will be unavailable until minuting changes, and that itself is the first recommendation of the audit.

---

PART TWO — THE ARITHMETIC

Weeks 3–5: the six measures on your own numbers

Exercise 2.1 — The decision register (one week, one analyst)

One row per decision across five years, six columns: date of first external trace, date decided, route (standing or exception), mover, outcome, downstream artefact. Expect 300–900 rows in a mid-sized firm.

Budget the analyst properly and say so out loud. This register is the asset. Once it exists, every measure in this chapter is a pivot table, and so is every governance question anybody asks you for the next five years.

Exercise 2.2 — Weighted participation (half a day)

Your firm's tiers are usually: the board, the executive committee, and whoever attends without a vote. Assign decision weight, compute N_eff at three weightings, and publish all three.

The finding to expect, and to handle carefully: in most firms the effective number is close to the size of the executive committee, not the board. That is often correct and occasionally the whole problem. The audit's job is to state it, not to adjudicate it — a board of twelve with an effective number of four may be well-delegated or may be a rubber stamp, and the next two measures are what tell you which.

Exercise 2.3 — Latency, against your own floor (half a day)

Compute days from first external trace to decision, and from decision to first committed spend. Then name the floor: your capital cycle, your budget calendar, your regulatory approval windows.

Report excess over the floor. This will usually exonerate the board and implicate a process nobody had looked at, which is why it is worth doing early: the first finding of a governance audit should be one the board did not expect to be about somebody else.

Exercise 2.4 — Non-implementation (two days)

Match every decision against procurement, work orders and payments. Compute the share with no downstream artefact within twelve months.

This is the number that pays for the project. If a board takes 180 decisions a year and 15 percent have no trace, that is 27 decisions a year of consumed executive attention with nothing behind it. Cost it at your own loaded rate for the preparation time alone — papers, pre-reads, committee hours — and you have a figure the finance director recognises immediately.

Exercise 2.5 — Exception share, with the denominator (half a day)

Two numbers together: the share of decisions taken outside the standing process, and the share of the firm's material decisions the standing process covers at all. Compute the second by taking your fifty largest commitments of the year and asking which ones appear in a minute anywhere.

Exercise 2.6 — Concentration and dormancy (two days)

Count recorded interventions by name for one year: top decile share, singleton count, effective number of voices. Then inventory every provision of the delegated authority matrix that creates a right or a duty, and ask the archive whether it was exercised in twelve months.

Dormancy is where the governance cost sits, and it is the finding boards act on fastest, because retiring an unused approval is a saving with no loser.

---

PART THREE — DESIGN

Weeks 6–7: make it survive contact with the firm

Exercise 3.1 — Compute your precision, then print it (two hours)

n = z²·0.25/d² gives 385 decisions for ±5 points; the design effect for clustering inside meetings, 1 + (m−1)·ICC, is about 2.05 at eight decisions a meeting, so the honest requirement is nearer 788. A board taking eight decisions across twelve meetings produces 96 a year: one year supports ±14 points.

Two consequences for your design. Use five years of archive, not one. And run it across committees rather than across time where you can — the clustering is in the meeting, so four committees for two years beats one board for eight.

Put the resulting precision on page one. A governance report that states its own margin will be treated as a measurement; one that does not will be treated as an opinion, and correctly.

Exercise 3.2 — Choose the holdout and get it excluded in writing (one hour)

Name one measure that will be computed and published every period and that may never appear in anybody's objectives or any incentive plan. The count of distinct names appearing in minutes is a good default.

Get the exclusion into the terms of reference of the remuneration committee. That sentence is worth more than the rest of the project, and it takes one meeting to obtain while nobody yet has a stake in the number.

Exercise 3.3 — The three gameable measures (half a day, with internal audit)

Work through each with the people who will be measured, in the open:

Doing this openly is not naivety. It is the strongest available evidence that the audit is a measurement rather than a performance review, and it is the difference between a register people maintain and one they route around.

---

PART FOUR — OPERATIONALIZE

Weeks 8–12: attach it to money

Exercise 4.1 — The internal case (half a day)

One page for the audit committee, in this order: the three best decisions and their mechanism; the six measures with weights and precision; the non-implementation cost in pounds; the dormant provisions with their meeting overhead; and one recommendation per finding, each of which changes a rule rather than a person.

Exercise 4.2 — The governance covenant (four weeks, with treasury)

Take the measures to your lender. This is a sustainability-linked loan with the performance targets replaced by governance measures — standard mechanics under the Loan Market Association's principles, with an independent verifier and a margin ratchet.

The arithmetic that decides it:

        annual margin saving
   ---------------------------------  >  1.0
    audit cost + verification cost

At 25 basis points on £5,000,000 the saving is £12,500 against audit and verification of about £11,500 — a cover of 1.09×. Breakeven is 23.0 basis points, or a facility of £4.6 million at a 25-point step. Below either, drop the ratchet and keep the audit: the six measures earn their keep at £11,500 whichever way the negotiation goes.

Three terms to insist on, because they are what make the covenant honest: the verifier is engaged and paid by the lender; the targets are three separate measures and never a composite; and the holdout is named in the agreement as excluded from the ratchet.

Exercise 4.3 — Put it in the pack (one hour, and it is the whole project)

Anything reviewed monthly persists; anything reviewed by exception does not. Get the six measures and the precision line onto the standing board pack. This is worth more than any presentation you will give about it, and the chapter's Destiny movement is blunt about why: the failure mode is not disagreement, it is quiet discontinuation.

---

SELF-ASSESSMENT

The test. Give the two pages to a non-executive director who joined last month. If they can tell you which rule to change and why, it is an instrument. If they ask whether the score is good, you built a composite without meaning to, and the fix is to remove the total.

---

WHERE THE FIRM'S OWN NUMBERS ALREADY SUPPORT THIS

Three places the business case is already made and nobody has written it down.

Your internal audit plan already costs more than this. A single process audit in a mid-sized firm runs well above £11,500. The governance audit covers the process that authorises every other process, and it is cheaper because the evidence is already written.

Your board pack preparation is a large, unmeasured cost. Papers, pre-reads, legal review, secretariat time. Multiply it by a non-implementation rate and you have the annual cost of decisions that went nowhere — in most firms a six-figure number, and the first one nobody will dispute.

Your lender is already pricing governance, informally and badly. Credit committees form a view on management quality from meetings and instinct. Handing them a verified five-year series replaces a judgement with evidence, and a verified series is precisely what a margin ratchet is designed to reward. You are not asking for a concession. You are removing an uncertainty they are currently charging you for.

---

THE THREE CONVERSATIONS THIS PROJECT REQUIRES

Sequence matters more than content. Held in this order, each one makes the next one easier.

With the company secretary, first and privately. This person owns the archive, has almost certainly noticed most of what your audit will find, and has had no mechanism for saying so. Go with the document inventory, not with conclusions. Ask what the minutes do and do not record, where the gaps are, and which committee's papers are the cleanest. You will save a week and acquire an ally who can tell you which findings are already known and which will land as news — and knowing the difference is most of the skill in reporting this.

With internal audit, second. Frame the register as an extension of their evidence base rather than a competing assurance activity, because it is one. They already sample controls; you are handing them the one process they have never sampled and the sampling arithmetic to do it defensibly. The gameability table belongs in this conversation and nowhere else first: internal audit thinks in terms of how a control is circumvented, and will improve your counters.

With the chair, last and with two pages. Open on the three best decisions and their mechanism. Then the six measures with their precision. Then the non-implementation figure in pounds, which is the number that turns interest into a decision. Ask for one thing only — the measures in the standing pack — and leave the covenant for a second meeting, because a chair who has agreed to monthly reporting has already agreed to everything that follows from it.

What not to do, in any of the three. Do not lead with a comparison to another company. Do not present a maturity model. Do not bring a score. Every one of those invites the conversation to become about position rather than about a rule, and a governance audit that produces a ranking produces exactly one response, which is a request for a better ranking.

And the standing note on tone. This audit will find that a board of able people, meeting diligently, takes some decisions that go nowhere and holds some authorities it has never used. That is the normal condition of every governing body ever measured, including the ones that are working well, and saying so plainly at the outset is what allows the findings to be read as information rather than as an accusation.