The Reckoning: What It Cost, and What Changes

Boards are arriving again. On Tuesday 26 May the line went back to full rate, and the stores in Rockford now hold more CB-40s than at any point since February. The other kind of board meets on Thursday 4 June, and Eleanor Vance has ninety minutes to tell six non-executives what twelve weeks cost and what she proposes to do about it.

She knows how this goes wrong, and it is not dishonesty but relief. The factory is running, and the easiest paper in the world to write is the one saying the team responded magnificently and the numbers largely recovered. Both are true. Neither is the account she owes.

What lands on Eleanor’s desk

From Claire Beaumont comes the closed-out cost of the response and the reconciled revenue position; from Ingrid Sørensen, a risk assessment and four costed controls; under both, Siân Pritchard’s log of every promise made during the fortnight of calls. Nothing operational is hers to reopen: Tomasz’s allocation ran, Grace’s dates held, Marcus’s ranking decided who waited. Eleanor is being asked the only two questions left at the end of a crisis: what did this cost, and what is now permanently different?

What a chief executive owes a board after a shock

A board cannot run a factory or choose between customers. What it decides is what the company keeps paying for once the emergency stops feeling like one — and it decides that almost entirely on how the emergency was described to it. Describe $700,000 as damage and they will cut it; describe it precisely and they can see which half was damage. So the paper runs two thirds review, one third what it means.

The software behind the board pack

The ERP answers what the disruption cost, and it answers it precisely, which is why the board will look there first. The CRM answers the more important question — what happened to the customers — and answers it far less precisely, which is exactly why Eleanor puts it on the same page rather than in an appendix.

The risk system carries the four controls with four owners and four dates, and the board pack does what board packs do: reduce twelve weeks of work by forty people to one page somebody will actually read. It is worth noticing that this is the same tool Ravi used on day one to escalate honestly, and the same one Claire used to make four incomparable options comparable. In both cases in this collection, the decisive artefact at the end is a single page — and what makes it honest is the eleven desks of work standing behind it.

Part one: the account

The units, and where they went

Module one’s arithmetic said 440 ovens would not be built. About 370 actually failed to reach a customer inside the twelve-week window.

Eleanor is careful about that gap, because the flattering reading is that clever management saved seventy ovens. Most of it was never real: Yusuf’s restatement took demand from 960 to 826, removing units that lived in a forecast rather than a customer’s plans. Grace’s resequencing then held roughly 76 ovens a week through the phasing instead of 67, Siân’s team found thirty tail orders where the customer volunteered a later date, and the broker’s forty boards passed test.

The money divides in two. Roughly $3.6m of revenue slipped into the following quarter; about $1.1m is permanently lost. Against an exposure of $5.3m the shape is better than the fear, and the $1.1m is the half that matters, because it is the only part Calder cannot earn back by working harder in June.

The quarter closed in week nine, mid build-ahead, with the factory busy and invoicing almost nothing. The quarter was missed. Expedite’s two weeks then compressed the recovery, overtime cleared the backlog through June and July, and the half was met.

$700,000, and the most important line on the page

The response cost about $700,000: $85,000 of air freight, $44,000 of broker boards, $115,000 of overtime, $38,000 of liquidated damages to Harlow, $180,000 of Kestrel tooling and $250,000 for the redesign.

As one number it is a disaster with six causes. Split properly it is two different things that happen to have been spent in the same quarter. About $282,000 was firefighting — freight, brokers, overtime, penalties — money that bought time inside the event and leaves no residue now it is over. About $430,000 bought permanent capability: a second board supplier Calder will still have in five years, and a control board that no longer depends on one microcontroller from one manufacturer.

Eleanor puts that split in bold near the front and spends longer on it than on anything else. Next spring there will be a cost review, and someone entirely reasonable will see $700,000 of crisis costs of which $430,000 still recurs — Kestrel’s unit price, qualification work, buffer stock tying up cash. Recurring spend attached to an event that is over is the easiest cut in any business, and precisely the cut that recreates the exposure.

The customer ledger

The money is the smaller half of this, because a shortage is finally settled in customers rather than dollars.

Bellwether made its dry dock — sixty units, complete and early, into a window that could not move by a day. Harlow slipped three weeks and paid $37,800, at $12,600 a week against a cap of $126,000; the programme continues, and its director accepted the slip because Marcus had written down why Harlow was the shock absorber before anyone rang to ask.

Nordfoods was phased into three tranches of eighty, took every delivery, and has still moved a quarter of next year’s volume to its second supplier. This is the largest single loss in the crisis. At roughly $1.4m of revenue a quarter, that volume is worth about $1.4m of revenue and $360,000 of gross profit every year, against $38,000 paid to Harlow once. It came from the account with no penalty clause and the lowest margin per board in the building, $1,560 against the Meridian’s $5,320 — precisely the loss Marcus’s ranking was built to catch. And it did catch it, which is why a quarter moved rather than all of it. Protection limited the damage; it did not prevent it. Eleanor says both sentences.

St Chad’s was two weeks late, triggered its supplier review and survived it, so the contract place holds. Peninsula re-based to thirty units and took a 3% discount — the cheapest thing Calder bought all quarter. And Acme Bakery, one oven and one board out of 520, got its delivery on the original date and said so publicly, at length, to other independent bakers.

What the numbers cannot show

The first is that the plan held because one person owned it. Tomasz owned the schedule, Ravi owned the objective and the escalations, separated on purpose. Crises are rarely lost to bad decisions; they are lost to four people making adjacent good ones from different spreadsheets.

The second is that the allocation survived because the rule was published rather than the answer. Anyone could see that Bellwether was protected and Peninsula deferred; the valuable part was seeing why. Publish only the outcome and twelve weeks become a lobbying campaign won by whoever has the most senior friend.

The third is the one Eleanor most wants the board to sit with: not one board arrived early enough to change who got an oven. Expedite bought two weeks at the end of the window and mostly turned lost revenue into slipped revenue; Kestrel qualified after the crisis was over; the redesign lands later still. Calder was not rescued by supply. It was rescued by allocation and four hundred honest phone calls.

Part two: what changes

The temptation to over-correct

Ingrid’s control plan is approved as written: availability built into design sign-off, buffer levels set as policy rather than as a favour to operations, one named owner per critical part, and a concentration figure in the quarterly board pack — covering the nine parts that are single-sourced, used across more than one line and longer than eight weeks to replace. New money is $120,000 of qualification work and roughly $200,000 of working capital held as buffers.

Two directors want more, and the instinct is decent: dual-source everything, hold six months of stock. Eleanor’s answer is arithmetic rather than principle. One qualification of one safety-approved part cost ten weeks and $180,000, so dual-sourcing 1,400 parts is not ambitious but imaginary, and six months of cover ties up cash in a business that has just learned how tight cash gets.

Then the real argument, about what happens to plans rather than what they cost. A twenty-item plan would be approved this month, unanimously and with feeling. By next March a third of the actions are closed, a third restated and a third quietly dropped, and nobody decided any of that. A shorter list survives the return of normality because every item has a name beside it and nowhere to hide.

The decision that costs nothing

The change Eleanor cares most about has no price tag. From this quarter, supplier concentration is reported to the board quarterly, beside the sales numbers, needing no commentary and no advocate: the share of weekly revenue passing through a component with no qualified alternative. Today that figure is 100%.

It needs no programme, no headcount and no consultant, and it is the only item that addresses what actually went wrong — because what went wrong was not a bad decision. It was two good ones, three years apart: consolidating three control boards into one, then awarding all the volume to a single supplier for $125,000 a year. Different people, different criteria, neither wrong, in a gap where nobody was looking at the combination.

No risk register catches that; the CB-40 sat on Calder’s for two years, rated amber, and stopped nothing. A standing number does what a register cannot. It makes the accumulation visible while it is still cheap to reverse, so that when the figure drifts from 12% to 30% through a series of sensible savings, somebody asks while the answer is still a conversation — rather than at 06:40 on a Tuesday in March, when the answer is $700,000 and 370 ovens. And once it is in the pack it cannot come out without a minuted decision to remove it.

Where this goes wrong

Recovery is the dangerous part, not the emergency. The buffer is consumed in the first tight quarter and never rebuilt; Kestrel’s unit price surfaces in a savings review as an obvious target, and the second source lapses because it has not been needed — which is exactly what a second source looks like when it is working. None of that requires anyone to decide the plan was wrong. It requires only that nobody decides it is still right, once the reason stops being memorable.

520 boards, 80 a week, 440 ovens

Go back to the arithmetic this course opened with: 520 boards, eighty a week, a twelve-week delay — 440 ovens that could not be built. That number never changed. What changed was the currency it was carried in.

Ravi turned it into an objective: protect first what cannot be bought back at any price. Hannah turned it into prices and lead times, and the discovery that every alternative arrives too late. Dev turned it into weeks and safety certification. Tomasz turned it into an allocation — 521 boards of protected orders against the 520 that exist. Grace turned it into hours on a line: eighty a week falling to 67, recovered to 76. Yusuf turned it into which orders were real, 960 down to 826. Claire turned it into dollars, and then into which quarter those dollars landed in. Marcus turned it into promises, ranked by whether a loss was permanent or merely painful. Siân turned those promises into four hundred conversations, and so into trust. Ingrid turned all of it into a mechanism and nine parts, and Eleanor into what the company permanently becomes.

Units, then hours on a line, then dollars, then promises, then trust — and nobody in that chain saw more than two links of it. Grace did not know a nine-day resequencing decision would put St Chad’s into a supplier review, and the buyer who consolidated Vantor’s volume had left before the email arrived.

Case one in this collection followed decisions that constrained the people who came after: a training budget agreed in week two quietly setting the precision of an answer in week twenty-eight, felt by colleagues, adjustable once noticed. These landed somewhere else. They landed on customers and on cash, within weeks rather than quarters, and could not be taken back — a line on a planner’s spreadsheet on the Wednesday was a phone call on the Friday, and a quarter of an account’s volume gone by the following year.

Spot the decision

Read each situation and decide how a chief executive should handle it, then tap a card to check.