Two years on, someone asks how it was decided.
Sector
Energy and utilities / Health systems / Financial services / Public infrastructure
The question
We write the question in one sentence.
Decision date
8 to 12 weeks for a full cycle
Record
112 files between 2019 and 2026
Plenary writes the record of the decision.
What leaves the building is a document, not a deck. It holds the question the board answered and the evidence that stood behind the answer. It also holds the alternatives that were on the table, and why this one was chosen over the rest. It is written for a reader who was not in the room and has no reason to trust you.
How the record is builtA record has four parts, and the fourth is the one that gets skipped.
What was known on the day of the vote.
carries the first part well
What was not known, and whether it could have been.
the second and third badly
Which alternatives were considered, including the ones dropped early and never written up.
Why this one was chosen over those.
the fourth not at all
Plenary has worked under nine sector regulators.
The question arrives in the same shape every time.
A regulator does not ask what you decided. It asks how.
The outcome is already on file. A review wants the path to it: what sat in front of the board, what was set aside, what nobody could have known then. A presentation cannot answer that. It was built to carry one conclusion into one room, and it drops everything that did not survive the argument. The companies that can answer it are the ones that wrote the answer down on the day.
The filesWe count these off the archive.
“We inherited four hundred slides and no memory of the choice. The next board inherits a document it can argue with.”
2
of the 4, one a partner
8 to 12
weeks for a full cycle
40
minutes, the board session
Every source carries a line saying how much weight it holds.
Board material usually presents evidence at one volume. A regulator filing, an internal model and a supplier estimate all read as though they deserved the same trust. We separate them. Each finding carries where it came from, how it was checked, and what would have to be true for it to be wrong. Where a number is an estimate, the page says so next to the number, not in a footnote.
Some questions will not be settled before the board sits.
Part of the page is blank on purpose.
01
A counterparty will not open its cost base.
02
A study will not read out before the vote.
03
A regime under consultation will still be under consultation on the decision date.
We write those down as gaps, each with the date it should close, or a note that it will not. A board that knows where the gaps are can decide around them. A board that does not know finds them later, in a hearing, described by somebody else.
Read the insight
How to say what you do not know
Advisers edit the uncertainty out in the final pass, because it reads as weakness. That edit is what makes the material fail at the audit.
Nadia Fournier, Principal
Three pieces of work. The first one often ends the matter.
Framing alone settles the question for some clients. Others take all three, eight to twelve weeks from the first meeting to the board session. The same two people stay on the file the whole way, and one of them is a partner.
Decision framing
Boards rarely lose time on the answer. They lose it on the question, which arrives too wide to answer and too loose to test. We narrow it to one the board can actually decide, then name the evidence that would change the answer. If nothing would change it, the decision has already been made and the work stops there. We say so.
Evidence review
We gather what exists, check it, and record where it is thin. Each source arrives with its provenance and its weight attached. We name what cannot be known before the vote, so the reader does not fill the gap with optimism. The output is not a finding. It is a body of evidence a third party can walk back through.
Board handover
The work is worth nothing if the board cannot read it. We cut the material to what a forty-minute board session can hold. We write the section that says what would make the recommendation wrong, and we run the session ourselves.
Every recommendation we write carries the case against itself.
The last section lists the conditions under which the recommendation fails: assumptions that have to hold, numbers that have to stay inside a range. It also names the event that would put the board back in the room. We write it before the recommendation is finished, because writing it afterwards produces a softer list. Clients come back to that section when one of the conditions moves.
Writing
Writing for the person who will disagree
Most advisory writing is built for a reader who already accepts the premise. That reader does not need it. The other one reads it at the appeal.
Mireille Osei, Partner, energy and utilities
The board signs off on the question before we gather a single document.
About the firm01
Two of us take the file, and one is always a partner. No juniors on the analysis, no handover to a new team halfway through.
That sets a limit on how much the firm takes on at once, and a full cycle runs eight to twelve weeks. The person who wrote the section on the limits of confidence answers for it when the board disagrees.
Who works on the file
02
We write the question in one sentence. If the board will not sign that sentence off, we have not found the question yet.
We began by narrowing the question the board had in front of it. The agenda said set the tariff. The real decision was whether to set it before the forecast firmed, and on what basis. We wrote that question down and tested it with the chair before gathering any evidence.
Energy and utilities / Tariff decision
03
We collect the evidence, check it and weight it. What we cannot check we mark unchecked and leave in.
The evidence review covered what could be checked at the time. That meant the vendor’s own data, the service record of the installed base, and the assumptions in the business case. Where confidence was thin, we said so next to the number it applied to.
Health systems / Equipment procurement
04
We write the alternatives up while they are still live, not at the end from what survived.
Every alternative went in, including the ones management had dropped early, with the evidence for and against each as it stood that month. The limits of confidence sat in the same sentences as the recommendation. A section named what would make the recommendation wrong, and what the board should watch for.
Energy and utilities / Tariff decision
05
We write the case against before we close the recommendation.
Clients occasionally ask us to cut it, usually before a session where the decision is contested. We do not cut it. That section is what makes the recommendation legible to a reader who comes to the file cold. It also does something useful in the room. A board that can see the conditions under which the advice fails argues about those conditions.
Writing for the person who will disagree
06
We cut the pack to the forty-minute session, run it and take the questions.
A board gives an item about forty minutes, if the chair is firm. That is the session the material is written for, and everything in the pack competes for it. Set the pack against that length before it goes out. If a third of the session would go on context the board already has, the pack has failed.
What a board actually reads
07
The client keeps the files in editable form, and can cite the record later without us in the room.
Anders keeps the decision archive. The two on a file index it as it runs, not after it closes, in the form he set. A question asked today traces back to the evidence that existed then. The archive itself has been cited in cases.
Anders Vik, Principal, public infrastructure
Four people write the files. Two of them are on yours, and one is always a partner.
Mireille Osei, Partner.
Energy and utilities. Four years on an energy regulator’s appeal panel.
Tomas Reinholt, Partner.
Health systems. Built clinical trial protocols before he came to Plenary.
Nadia Fournier, Principal.
Financial services. Writes most of the method the firm works to.
Anders Vik, Principal.
Public infrastructure. Keeps the decision archive, the one cited in cases.
Four files. Two came back after the vote, and not for the same reason.
Names are withheld. The situations are not. In the first, a regulator reopened the matter and asked how the number had been reached. In the second, a study landed late and the board revised one section on a date the record had already set. In the third the split happened before the vote, between two committees. In the fourth there was nothing left to question it with.
Regulator wanted the reasoning, not the result.
A tariff decision was reopened nineteen months after the vote. The regulator did not contest the number. It asked how the company had reached it and which options had been set aside on the way. The record written at the time answered both questions in the form they were asked. Nothing was reconstructed from memory, no new analysis was commissioned, and the file closed on documents that already existed.
The evidence arrived after the vote.
A hospital network had to commit to an equipment programme before the study that bore on it read out. The record named that gap in advance and set the date it would close. It set out what the board would do in each direction the study could go. When the evidence landed, the board revised one section instead of reopening the whole decision. The minute already explained why the timing had been accepted.
Two committees, one decision.
A bank was leaving a segment. The risk committee and the strategy committee read the same data and reached opposite conclusions. We wrote both readings into one document and marked the point where they split. It was not the data. It was a difference in what each committee treated as an acceptable loss, which nobody had written down. The board decided that question in one session, and the dissent sits in the record under its own heading.
A decision nobody could reconstruct.
An eleven-year investment programme was under way and nobody could say why it had been shaped that way. The previous contractor had left four hundred slides and no record of the alternatives. We rebuilt what could be rebuilt from minutes and correspondence, and marked the rest unrecoverable rather than guessing at it. The programme now carries a record from the month we started. The period before that stands in the file as a gap, not as a reconstruction.
Notes on the pack, the minute, and evidence that lands late.
Pieces by the people who write our board material. They come out of work we have done, not out of a content plan, so they appear irregularly.
What a board actually reads.
A board pack is written in the order the work happened. It is read in the order the risk sits, and those two orders are almost never the same.
Where attention goes in a board pack, and where the hard part has to sit.
Mireille Osei, Partner, energy and utilitiesRead: What a board actually reads
The minute that matters.
Most minutes record attendance, motions and results, and nothing more is asked of them. The minute that protects a board records the reasoning, and it runs two paragraphs longer.
The minute is the part still being read two years later. Most are written as though nobody will read them again.
Anders Vik, Principal, public infrastructureRead: The minute that matters
When the evidence arrives late.
Evidence does not arrive on the board calendar. Decide before the vote what would count as a reason to look again, and the late arrival becomes a check instead of an argument.
How to take a decision that can absorb a study landing after the vote.
Tomas Reinholt, Partner, health systemsRead: When the evidence arrives late
Send us the question the board cannot agree on.
All four of us see what arrives. The one who covers your sector writes back, and that is the person who would carry the file. You get a view on whether the question is framed tightly enough to be worth the work. If the work is not ours, we say so, and where we can we name whose it is. No pitch document follows.
No intake team and no account manager. Nobody writes a summary of your message for someone senior to skim.
Before you write
Three things that change the first reply
The decision, in one sentence. Not the subject, the decision. “Close the northern plant or run it to end of life” is one sentence, and it says more than a briefing pack does. “Generation strategy” names a subject, and it tells us almost nothing. The date the board sits. A full cycle takes eight to twelve weeks. If your vote comes sooner than that, the first reply says so. It names the part that still fits: usually the framing, sometimes the handover on evidence you already hold. What has already been written, and by whom. A previous review, a regulator’s letter, an internal paper that went nowhere. One file reached us as 400 slides from a departed contractor, with no record of the alternatives anywhere in them. Knowing that at the start changes the shape of the work and the price of it.