La Bourse / Volume IV / Nº IV.07 / Workbook — the Gainshare employee

A woman reading in a rattan chair beside a tall white window, light falling across her and the floor. — Plate IV.07 · Workbook — the Gainshare employeeThe Second Room.Everyone draws a supply chain as a line because a line fits on a page. The thing itself is a population, and it is lit from a room you have not entered.

WORKBOOK — THE LUMINOUS GAINSHARE EMPLOYEE

Chapter IV.07 · Supply Chains as Ecologies

Volume IV — Production and Regeneration

How this mechanism reads from inside a gainshare: what to measure, what to claim, how the ledger treats it, and what to ask for. Every figure is computed in lib/verify/IV_07.py.

WHAT A GAINSHARE ACTUALLY IS

A gainshare pays you a defined share of a measured improvement against an agreed baseline. Three words carry all the weight: measured, agreed, and baseline. If any one of them is missing, what you have is a bonus scheme with better vocabulary.

Supply-chain work is unusually good gainshare material and unusually badly claimed, for one structural reason: most of its value is a cost that did not happen. An audit you cancelled is visible. A stoppage you prevented is invisible, and invisible savings are the ones that get absorbed into general performance at the moment somebody else is being promoted.

This workbook is about making the invisible part countable in advance. Not after the good thing happens — before, because a baseline agreed after the fact is not a baseline, it is a negotiation you will lose.

There are three claimable categories here, and they differ in how hard they are to defend:

Category	Evidence	Difficulty
Cash out — duplicate audits cancelled, expediting avoided	Invoices	Easy. Claim it first.
Risk reduced — a lower `q`, a shorter recovery time	A computed expectation against a signed model	Medium. Needs the model agreed up front.
Reach extended — enforcement depth `n*`	A computed depth against a stated configuration	Hard, and the most valuable.

Claim in that order. The first funds the credibility for the third.

PART ONE — DISCOVERY

Days 1–30: find where the gain actually comes from

1.1 — Find the duplicated audits. Pull the audit register and the vendor list. Find every site your firm inspects that is also inspected by a group scheme your firm or your customers already subscribe to. In the chapter's reference unit, 66 direct suppliers at $6,000 each against $750 through a shared scheme is:

  66 × ($6,000 − $750)  =  $346,500 a year

That is cash, it is on invoices, and it does not depend on anybody believing a model. It is the strongest gainshare claim in this chapter and it is sitting in a filing system right now.

1.2 — Find the unpriced knowledge. Who in the building knows something about tier two that is written nowhere? The buyer who visited the plant in 2019. The engineer who knows which alloy mill actually pours the billet. The planner who knows which two "independent" suppliers ship through the same port.

Write it down and put your name and the date on it. Knowledge that lives in a person is not claimable; knowledge in a dated document is. This is not bureaucracy — it is the difference between being valuable and being able to demonstrate that you were.

1.3 — Find the near misses. Every incident in the last three years where a supply problem nearly stopped a line and did not. Ask what actually saved it. The answer is almost always a relationship, a buffer, or somebody's private knowledge — and each of those is a mechanism your firm currently gets for free and does not maintain.

1.4 — Get the event cost signed. Contribution margin per day, expediting, penalties, customer loss. Take it to finance and have them sign it before you propose anything. Every risk-reduction claim you make will be multiplied by this number, and an unsigned multiplier is a claim somebody can zero in a meeting.

PART TWO — THE ARITHMETIC

Days 31–45: compute what you are owed

2.1 — The cash claim. Straightforward and it should be in the ledger in week six:

  duplicate audits cancelled          $346,500 /yr
  gainshare pool at 20%                $69,300 /yr
  team of six, equal share             $11,550 each /yr

Note what makes this defensible: an invoice that used to exist and now does not, against a supplier list that did not change. Nobody can argue that the savings were caused by the market.

2.2 — The risk claim, done properly. This is the one that requires the model agreed in advance. Suppose your mapping work finds that four "independent" sources for two families all ship through one port, and moving one source reduces the common-mode term from q = 0.015 to q = 0.009:

  ΔP  =  0.006
  event cost (signed)          $28,000,000
  expected annual saving  =  0.006 × 28,000,000  =  $168,000 /yr
  gainshare pool at 20%                             $33,600 /yr

The rule that makes this claimable: the model is agreed before the change, not after. Write one page — the failure model, the two values of q, the signed event cost, the arithmetic — and get it countersigned by finance and operations in month one. Then the payment is a computation rather than an argument.

And be honest in both directions, because it is what makes the whole ledger trustworthy: an expected saving is not a realised one. If your scheme pays on expectation, say so in the document. If it pays on realisation, you are being paid for the absence of an event, which means you must also agree in advance what happens if an event occurs for an unrelated reason.

2.3 — The reach claim. The hardest and the most valuable, because nobody else in your organisation can currently compute it.

  n*  =  1 + ln(V₁/C₁) / ln(b/λ)

  before   C₁ = 66 × 6,000 = $396,000    n* = 1.718 tiers
  after    C₁ = 66 ×   750 =  $49,500    n* = 2.364 tiers
  gain                                    +0.646 tiers

A tier is not a currency, so do not try to claim it as one. Claim it as the capability metric in your scheme — the thing that moves your band, your scope or your title — and claim the cash beside it. In every gainshare negotiation, the cash gets you paid this year and the capability metric gets you paid for the next five.

2.4 — What not to claim. Three things, named so you do not burn credibility on them:

The whole avoided loss when an event does not occur. You cannot claim $28 million because nothing happened. You can claim the expectation, agreed in advance, at the agreed share.
Improvements that were going to happen anyway. If the group scheme subscription was already signed before you arrived, your claim is the cancellation of the duplicates, not the subscription.
Another team's baseline. If quality already had a plan to cut audits, the claim is joint, and saying so first is worth more than winning it.

PART THREE — DESIGN

Days 46–70: make the uncounted countable

3.1 — Write the one-page measurement agreement. Four sections, countersigned before any work starts.

Baseline. Audit spend by supplier, current source counts, current enforcement depth n* with its inputs, current disclosure rate with its denominator.
The model. The failure expression P(all m fail) = q + (1−q)p^m, the values of p and q you are using, and where they came from.
The multipliers. The signed event cost. The gainshare percentage. The split across the team.
The exclusions. What this does not claim. Name them. A claim document that excludes nothing gets read as a claim document that counted everything twice.

3.2 — Build the instruments that make the ledger self-maintaining. Three, all cheap:

The dated map. Every node carries a verification date; anything over twelve months reads as unknown. This is what stops your own claim rotting.
The exposure column. E beside every family in the risk register, with the formula in the column header so the sort survives a handover.
The four numbers in the monthly pack. n*, tier-two disclosure rate, families with a written source-count justification, and spend covered by a dated map with the remainder stated.

3.3 — Ask for the right thing. When the gainshare conversation opens, the three asks that actually compound:

A share of the audit saving for three years, not one. The saving recurs; a one-year share prices a permanent change as a project.
The capability metric in your scheme. Enforcement depth, named, with its inputs. It is the only metric in the building that nobody can currently game, because nobody else knows how it is computed.
The deputy role formalised. You want a second owner on the map — partly because it is correct engineering and partly because a mechanism with two owners survives your promotion, and a mechanism that survives you is the thing that gets referenced when the next opportunity is allocated.

PART FOUR — DESTINY AND DELIGHT

Days 71–90: make it hold

4.1 — Get the ledger into the standing pack. A gainshare that is reported by exception is a gainshare that is paid by exception. One line in the monthly operations pack: baseline, current, delta, pool, paid.

4.2 — Publish the arithmetic, not the conclusion. Put the inputs beside every number you report. The first colleague who checks one of your figures and finds it exact is the person who will defend your claim in a room you are not in — and that is worth more than any presentation you will give yourself.

4.3 — Hand one piece away deliberately. Give the exposure-ratio column to somebody in quality or engineering and let them own it. You lose a little credit and you gain a mechanism that no longer depends on your enthusiasm, which is the only form of job security that survives a reorganisation.

4.4 — Watch for the three ways this stops paying.

The baseline drifts. Somebody re-tenders and the supplier count changes, and your n moves for reasons that have nothing to do with you. Fix it by reporting n with its inputs every month, so a change in N₁ is visible as a change in N₁.
The saving is reclassified as "market." Fix it with invoices. This is why the cash claim goes first.
The scheme changes its percentage. Fix it by having the three-year term written down at the start. It is the single most common way a good claim quietly halves.

On delight. There is a particular pleasure in this work that has nothing to do with the payment. It is the first time somebody senior asks a question about the upstream and you already have the answer on one page with its inputs printed — and the pleasure is not being right, it is watching the room stop guessing. The second pleasure is better: the meeting where somebody uses your enforcement depth figure in their own argument, for their own purposes, without attributing it. That is the moment a computation becomes infrastructure. It has stopped being yours, which is exactly what you wanted.

KNOW YOUR SCHEME — A CHECKLIST

Ten questions. If you cannot answer one of them, that is this week's work.

What exactly is my baseline, who signed it, and on what date?
Does the scheme pay on expectation or on realisation?
What is the agreed event cost, and who can change it?
What is my percentage, and for how many years?
Is the pool per team, per site or per business unit — and who else is in it?
What happens to my claim if the supplier count changes for an unrelated reason?
What happens if a disruption occurs anyway, in a family I did not touch?
Who verifies the numbers, and are they independent of the people who pay?
What is explicitly excluded, in writing?
What happens to the claim if I change roles inside the company?

Question nine is the one that separates a scheme that will pay from a scheme that will litigate. A scheme with no written exclusions has not been thought about yet, and the time to fix that is before the money exists.

THE CONVERSATION, SCRIPTED

You will get roughly four minutes. Use them in this order.

"Three things. First, we are paying for the same factory inspection twice — once ourselves and once through a scheme our customers already fund. Cancelling the duplicates on sixty-six suppliers saves $346,500 a year and the evidence is invoices."
"Second, the same change extends the reach of our code of conduct from 1.718 tiers to 2.364. I can show you the arithmetic; the inputs are our own supplier count and our own audit cost."
"Third, I want the measurement agreement signed before I start, not after. One page, the baseline and the exclusions, countersigned by finance. If the saving is real it will show up in invoices within two quarters, and if it is not I would rather we found that out on a page than in a year."

The third paragraph is the one that gets the first two believed. An employee who asks to be measured against a signed baseline, and who names the exclusions before anyone else does, is asking for something no scheme has a reason to refuse.

APPRECIATIVE QUESTIONS FOR YOUR TEAM

When has something one of us knew about a supplier saved this company money — and how would we have proved it if we had been asked?
Which of our savings are already visible on an invoice, and which are only visible in a model? What would it take to move one from the second column to the first?
Where has this team been paid for something that nobody had to be argued into? What made that claim easy, and how do we make the next one look like it?
If we could put one number in the monthly pack that nobody else in the building can compute, what would we choose?
What would we want still running here in five years, with our names nowhere on it — and what would we have to hand away this quarter for that to happen?