Haute Lumière

Commerce · I.10 · MMXXVI · daylight

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

WORKBOOK — THE CORPORATE EXECUTIVE

Chapter I.10 · Scaling Without Losing the Thing

For the person with a P&L, a signature limit, a board, and a growth plan already approved. This workbook uses the language of the firm without apology, because the firm's own numbers already contain the answer — they have simply never been arranged as a curve.

---

THE PREMISE, STATED COMMERCIALLY

Your growth plan almost certainly contains a unit count and a date. It almost certainly does not contain the two numbers that decide whether those units will be worth having:

• The fidelity floor — the point at which a unit stops being your product and starts being a liability wearing your name.
• The coordination ceiling — the size above which the shared centre costs the units more than it returns.

Both are computable from figures you hold today. Neither appears in a standard growth model, which is why growth plans are routinely approved at sizes their own arithmetic will not support, and why the failure shows up two or three years later as execution rather than as design.

The commercial statement of this chapter is short. Unit count is not the asset. Units-above-the-floor is the asset, and the gap between those two numbers is carried on your balance sheet as goodwill and discovered by your customers one at a time.

---

PART ONE — DISCOVERY

Days 1–30

Exercise 1.1 — The hop map (one week, with HR and operations)

Build the transmission graph of your own network. For every site, branch, team or franchised unit, record one field: how many teaching hops from the source? The source is whoever actually holds the practice — a founding site, a flagship, a named person, a training centre.

Sources for this data already exist: training records, who opened each site, who the opening manager reported to, secondment history. It takes about a week and almost nobody has ever done it.

Output. A one-page graph. Expect to be surprised twice: by how deep the deepest chain is, and by how many units nobody can place at all. A unit whose provenance cannot be established should be treated as beyond the floor until assessed.

Exercise 1.2 — The four-box sort (one session, with your executive team)

Sort what your firm does into the four boxes, because each scales by a different instrument and the most expensive errors in this chapter are category errors.

Output. Your business, in four columns, with the current instrument written beside each and every mismatch circled. Every circle is a candidate for the next eighteen months.

Exercise 1.3 — The appreciative board conversation (one session)

Change one question in one meeting. Replace "which sites are underperforming?" with:

"Which site furthest from where this started is still unmistakably us — and what did we do right in that case that we did not do deliberately?"

Record what comes back. You are building the evidence that the firm already knows how to transmit, which is a materially easier argument than the claim that it must learn.

---

PART TWO — THE ARITHMETIC

Days 31–45

Exercise 2.1 — Measure φ (two weeks)

Fidelity is measurable and almost nobody measures it. Take the assessment you already run on units — mystery shopping, audit score, clinical review, NPS, first-pass yield, whatever exists — and regress it against hop depth from Exercise 1.1.

You are looking for one number: the average score decrement per hop. Express it as a ratio and that is your φ. Then:

  hops until half the thing is gone     ln(0.5)  / ln(φ)
  hops allowed at a 0.70 floor          ln(0.70) / ln(φ)

At φ = 0.90 that is seven and three. At φ = 0.85 it is four and two. The difference between those two firms is five points of teaching quality and a threefold difference in the network each can hold.

The board-ready sentence: raising φ by five points is usually the cheapest capacity expansion available to the firm, and it does not appear in the capital plan because it has no line item.

Exercise 2.2 — Measure B, k and c (two weeks, with finance)

For the shared centre — group functions, the franchisor entity, the services company, whatever you call it:

Then the two numbers:

  n*  =  k · ln(2·B_max / (k·c))                 the optimum
  n_ceiling : the n at which B(n) − c(n−1)/2 turns negative

On realistic mid-market parameters — £120,000, twelve, £2,800 — those are twenty-four and eighty-seven. Compute your own. Put both in the board pack.

Exercise 2.3 — The royalty diagnostic (if you franchise) (2 hours)

  unit revenue × (royalty + ad fund)              the take
  ÷ unit operating profit before royalty          the share of profit
  break-even quality lift = spend / ((1 − take) × revenue)

If your royalty is 8 percent of gross against a 12 percent margin, you are taking two-thirds of unit operating profit and your operators must clear a ~2.4 percent revenue lift before any quality investment pays them anything. That is the quality trend in your network, and it is contractual rather than cultural.

Three counter-mechanisms, all of which have precedent: co-fund specified quality investment at 50 percent; grant a royalty holiday on the payback period of an approved upgrade; or move the royalty to gross profit. The third is cleanest and the hardest to retrofit.

---

PART THREE — DESIGN

Days 46–60: the instrument

The structure: a member-owned shared services company with a capped levy, a patronage rebate, and a written ceiling clause.

This is the CUSO in American credit unions, the secondary cooperative in Mondragón's structure, the consorzio in Italian law. It is ordinary company law and your counsel will recognise every clause.

Balance-sheet treatment. The unit holds its share as an investment at cost, immaterial by design. The levy is operating expense. The rebate is income in the year received. The services company capitalises platforms and depreciates over useful life — and the one to watch is this: a shared platform's useful life is bounded by the federation's commitment horizon, not the technology's. If members can leave on ninety days and the platform is on a seven-year schedule, disclose it. Raise it with your auditors in month one, not month eleven.

The counterparty. The members, and only the members, until the company has two full years of audited surplus. The first external lender will want security, and the only security this entity has is the members' obligation to keep using it — which is precisely the thing you undertook not to create.

The number that decides it.

              measured benefit per member unit
   -----------------------------------------------------  >  3.0
    levy  +  (coordination hours × loaded hourly cost)

Worked: benefit £86,000, levy £24,000, 180 coordination hours at £75 = £13,500.

   86,000 / 37,500  =  2.29 : 1     does not clear

That centre is not failing; it is the wrong shape. Cluster the network so each unit holds nine relationships rather than eighty-five, coordination falls to 60 hours, and:

   86,000 / 28,500  =  3.02 : 1     clears

Same benefit, same levy, same people. Clustering is the cheapest margin improvement in this workbook and it requires no capital at all.

---

PART FOUR — DESTINY AND DELIGHT

Days 61–90

Exercise 4.1 — Capitalise the tier (one week)

Whatever you cluster under — regional hubs, a services company, a liquidity pool — is now systemically important to everyone who depends on it. Give it its own balance sheet, its own board with unit representation, and an annual stress test that asks one question in writing: if this fails, what happens to our smallest member?

The precedent is not theoretical. American credit unions clustered under corporate credit unions holding their liquidity; in 2009 five of those were placed into conservatorship and the cost was assessed back across the very institutions the tier existed to protect. The second tier solves the coordination problem by concentrating into one point everything the mesh kept apart. That is a trade, and it is a good one, and it must be capitalised as one.

Exercise 4.2 — Change the growth metric (one board cycle)

Replace units opened with units-above-the-floor in the board pack. Two columns, both reported, every month. The divergence between them is the single most useful number in your growth programme and currently nobody owns it.

Attach the retrain budget as a standing line called transmission, sitting where maintenance capex sits, because it is the same idea: an asset that decays and is restored on a schedule rather than replaced on failure.

Exercise 4.3 — Put somebody on a plane (immediately)

Every one of the four families in this chapter transmits partly by a person from the source spending a week doing the work in a distant unit. It is expensive, it does not scale, and it is why the rest scales. Budget it as a fixed number of source-weeks per unit per year and protect it the way you protect a maintenance window.

Exercise 4.4 — Price the four instruments against each other (one session)

Before the next expansion decision, put the four side by side on one page for the same ten units, over five years, at your own cost of capital.

The row that surprises most boards is the third. A federation costs less capital than replication, transmits better than a franchise, and returns a smaller but far more durable margin than either — and it is the only one of the four in which the units have a reason to improve the centre rather than to route around it.

Output. One page, four rows, your own numbers, taken to the board as a comparison rather than as a recommendation. The comparison wins the argument that the recommendation would have lost.

---

THE FAILURE MODES, NAMED

So you can see them coming

The centre becomes the principal. Gradually, by a series of individually reasonable standardisations. First sign: a unit asks permission for something it used to decide. Who sees it first: the longest-serving unit manager, who will mention it once and not again.

Growth outruns fidelity. First sign: audit scores hold at the mean while the tail lengthens. Report the tenth percentile, not the average — an average that holds while a tail forms is the most reassuring number in this chapter and the most dangerous.

The federation admits past its arithmetic. First sign: attendance at the shared forum falls while membership rises. That ratio is your coordination ceiling announcing itself.

The founder is still the mechanism at unit forty. First sign: the training calendar cannot absorb a holiday. Nothing has scaled; one person has been spread thinner.

---

THE NINETY DAYS ON ONE PAGE

---

BOARD PAPER TEMPLATE

Purpose. To set the firm's transmission and coordination parameters, and to adopt a growth metric that reflects them.

Measured. φ = ___ (score decrement per hop, n = ___ units). Floor at 0.70 permits ___ hops. Today ___ of ___ units sit beyond it. B_max = £___ per unit per year; k = ___; c = £___ per relationship per year. Optimum network n\* = ___. Net-negative above n = ___.

Proposed. Adopt units-above-the-floor as the reported growth metric alongside unit count. Establish a transmission line in opex at £___. Cluster the network at n\* into groups of ___. Adopt the ceiling clause.

Decision number. Coverage ratio ___ : 1 against a 3.0 threshold.

Risk. The clustering tier concentrates ___ . Buffer of £___, sized against the loss it could impose on our smallest unit. Annual test, published.

---

APPRECIATIVE QUESTIONS FOR YOUR LEADERSHIP TEAM

1. Which of our most distant units is still unmistakably us, and what did we do right there that we did not do on purpose?
2. What does the centre give a unit that the unit genuinely could not buy alone — and when did we last price it?
3. If we reported units-above-the-floor next to units opened, what would we be proud to show in three years?
4. What would make leaving this group easy, and what are we afraid would happen?