Case Study 40.1: The Plan That Passed Every Review
📜 Tier 3 — Illustrative. Kwabena Osei-Duah, the Thornfield event, and every figure here are constructed. The failure mode is common; the details are the author's.
The plan
A 300-person charity gala. A converted mill. Kwabena Osei-Duah, eleven years in, running his fourth event at this venue.
The production book was excellent and everybody who read it said so.
| Read-through with the venue | Two hours, three weeks out. Four discrepancies found and fixed |
| Read-through with the AV producer | Ninety minutes. Two cue timings corrected |
| Internal audit against his own template | All items present |
| The committee | Approved it, twice |
Four reviews. Eight findings, all fixed. Nobody had a reservation.
And on the night, at 19:12, the entire event stopped for twenty-six minutes.
What happened
🚨 19:12, traced
The paddle raise is at 19:51. The lead-gift donor — the $25,000 opening paddle secured in month two, Chapter 32's whole mechanism — is not in the room.
He is at the venue. He arrived at 18:40. And he is upstairs in the mezzanine bar with four colleagues, because that is where he was directed by a steward who did not have a seating list.
19:12 Development director notices table 3 has an empty seat 19:18 Somebody is sent to look. Nobody knows what he looks like 19:31 Found, upstairs, having a good evening 19:38 Seated 19:51 The paddle raise runs. It is fine Twenty-six minutes of the development director's attention, at the single most consequential hour of the year, on a problem that did not affect the outcome.
And that is the good version. Had he been found at 19:53, the ladder opens to silence — and Chapter 32's compounding loss runs down the whole thing.
Why four reviews missed it
🚪 Because nothing in the plan was wrong.
The seating plan was correct. The lead gift was secured, documented, and named in section 03. The stewards were briefed. The paddle raise's run sheet was right to the minute.
Every reviewer read the plan and every reviewer found it internally consistent — because it was.
A read-through asks does this document agree with that one?
The question nobody asked was: what happens if this particular person is not where the plan assumes he is at 19:45?
That is an injection, and there is no review that produces it.
⚡ What an injection would have found, in about four minutes
Injection: the lead-gift donor is not in his seat at 19:45.
Who notices? Nobody, until 19:12 by accident. Nothing in the plan checks it Who could find him? Nobody knows what he looks like. Not the stewards, not the auctioneer, not the development director's own team beyond her Point of no return 19:49 — after which the ladder opens without him What is the fallback? There is none. The plan has one lead gift and no second Four minutes of tracing, at a desk, three weeks out, and it produces three fixes that together cost nothing:
A named person greets him on arrival — Chapter 38's Delphine problem in reverse. A photograph in section 03, which is slightly awkward and entirely normal. And a second lead gift at $10,000, which Chapter 32 recommended and this plan did not have.
What Kwabena changed
✅ Three things, and the third is the one that generalises
1. A "critical people" block in section 03 Names, photographs, arrival times, and who greets each — lead-gift donors, the officiant, key speakers, anybody without whom a segment cannot run 2. Two lead gifts, always Chapter 32 §32.5, which he had read and not implemented 3. The five-injection set, run at −3 weeks, in writing Forty minutes, before the read-through rather than after Item 3 is the change and the ordering is the point.
He had assumed injection was a more rigorous review and would therefore duplicate it.
It is a different instrument that finds different things — and running it before the read-through means the read-through checks a plan that has already been attacked.
🎤 Kwabena, afterwards
"Four people read it. All four are good. Not one of them was ever going to find it, and I want to be clear that's not a criticism of them."
"You cannot find a fragility by reading. You can only find it by breaking something."
"And the thing that still bothers me is that I'd read Chapter 32. I knew about the second lead gift. It was in my notes."
Discussion Questions
DQ1. Four competent reviewers missed it. Is that a failure of the reviewers?
Consider
**No, and Kwabena is right to say so explicitly.** **A read-through has a defined question: does this document agree with that one, and is anything missing against the template?** **On both, the plan passed** — the lead gift was documented, the seating was right, the run sheet was correct to the minute. **The fragility was not a discrepancy. It was an assumption** — *the donor will be in his seat* — **and an assumption is invisible to a method that checks consistency.** **Which is Chapter 28's finding in a different form.** **An audit finds contradictions and cannot find a consistent error**, and here there was not even an error: **there was a correct plan with an unexamined dependency.** **And the general shape is worth carrying: reviews find what is wrong; injections find what is fragile, and those are different sets.**DQ2. "I'd read Chapter 32. I knew about the second lead gift. It was in my notes." What is this an instance of?
Consider
**Chapter 39's entire subject, arriving as a lived cost.** **He had the finding. It was written down. It never became a change to an artefact** — and the reason is §39.1's second mechanism: **the change landed in his notes rather than in his gala template.** **A note is not an artefact.** **It is not opened at −3 weeks, it has no owner, and it produces no obligation** — and Case Study 39.1's Rasheeda had ninety-one of them. **What would have closed it: a line in the gala template's revenue section reading *"lead gifts: two, secured by −6 weeks, named in section 03."*** **Four words in a document he opens every time.** **And the specific pain of this case is that the finding was correct, recorded, and understood** — **which is the exact profile of the fifty-eight assignable entries Rasheeda had never actioned.**DQ3. The event was fine. The paddle raise ran and produced its number. So what actually went wrong?
Consider
**Twenty-six minutes of the development director's attention at the most consequential hour of her year** — which is Chapter 28's second-order failure, and it is invisible because nobody can see an absent action. **What she was not doing between 19:12 and 19:38: watching the room, positioned for the appeal, and having the two conversations with major prospects that the entire evening exists for** (Chapter 32 §32.2). **And the counterfactual matters more than the cost.** **Found at 19:53 instead of 19:31, the ladder opens to silence, the auctioneer drops a rung, and Chapter 32's compounding loss runs down the whole ladder** — **a five-figure difference caused by a steward who did not have a seating list.** **Which is why the case study is titled as it is.** **The plan passed every review and was twenty-two minutes away from a materially worse evening**, and nobody in the building would have known why.DQ4. The fix required a photograph of a donor in the production book. Is that appropriate?
Consider
**Yes, with a caveat that Chapter 39 supplies.** **It is entirely normal practice at scale** — production books for large events routinely carry photographs of speakers, VIPs, and anybody who must be recognised — **and it exists precisely because "somebody will know him" fails.** **The caveat is §39.7.** **A photograph of a named individual in a document distributed to fourteen people is personal data**, and the same rules apply: **collect the minimum, share only with those who need it, control access, and delete on a schedule.** **Which points at a better implementation than putting it in the main book: the critical-people block goes to the four people who need it** — the greeter, the two stewards, and the development director — **and not to the caterer.** **And that is Chapter 26's distribution matrix, doing exactly what it was built for.**DQ5. Kwabena now runs injections before the read-through. Why does the order matter?