A strategy day rarely fails in the room. It fails on the following Monday, when everyone agrees that it went well and nobody can say what changed. Agreement is the cheapest output an executive session can produce, and it is the one it produces by default, because agreement costs nothing to give and leaves nothing behind that can later be found missing.
I have watched enough of these to know the shape. The material is good. The questions are intelligent. The close is warm. Weeks later the programme is still waiting on a budget line, a metric definition and a name against a role.
While going through a board deliverable I worked on for a consumer and SME banking group operating across several European markets, I stopped on a slide that most reviewers would page straight past. It is a section divider. Structurally it does nothing except announce that the deck has finished describing the operating model and the capability build, and is about to describe how the bank would deliver it. And in the middle of that transition, before any of the implementation content appears, it states the three decisions the board is being walked toward and what each one has to produce.
Put the ask at the top of the section, not only at the close
The conventional shape of a board deck saves the ask for last. You establish the case, walk the plan, show the money, and on the final slide you say what you need. It feels respectful. It is also the reason so many of these sessions end in warmth rather than in decisions, because for the whole preceding hour the room has no idea what it is being prepared to do, so it listens.
The bank's deck does close its board narrative that way, on a decision page carrying a signature block. What it adds is an earlier statement of the same job. At the moment the deck turns from design to delivery, the board is told that there are three things it will be asked for, here they are, and each one produces a document before the room breaks.
That changes the job of everything that follows. A roadmap presented to a room that has not been told what it will be asked is information. The same roadmap presented to a room that knows it will be asked to sign it before it leaves is an exhibit. People read exhibits differently. They interrupt earlier, they ask about the third pilot rather than nodding through the sequence, and they raise the resourcing objection while there is still time to answer it rather than in the final five minutes when the only available response is to take it away.
I want to be careful about what I can and cannot claim here. The deck is a design. I can tell you what the slide is engineered to do and why the engineering is sound. I cannot tell you how the room behaved, because a deliverable is not a minute.
- 3
- Board decisions the deck walks toward, each bound to a named output
- 4
- Sections previewed on the divider, then nothing else
- 90
- Days in the first sprint plan the board is asked to approve
- 0
- Things built at the point this was written; the deck is a proposal
The three asks are not three of the same thing
It would be easy to read the trio as one mechanism repeated three times. It is not. Each ask commits a different party to a different kind of cost, and the differences decide which one will quietly fail.
Approving the roadmap is the cheapest of the three. It commits the programme, not the board. A director who signs off a ninety-day sprint plan and a pilot ordering has agreed to a direction of travel that somebody else will execute, and if it goes wrong the plan was wrong. This is the ask most likely to be granted, and granting it feels like a decision.
Committing resources is the most visible. Budget envelopes and named role appointments attract scrutiny because money and headcount are the currency the room already knows how to argue about. This ask will not be waved through. It may be refused, deferred or trimmed, but it will be discussed properly, because everyone present has spent a career discussing exactly this.
Defining the KPIs is the fragile one. It is the only ask of the three that binds the board itself. A metric is an advance agreement about what evidence will be admissible later, which means it is an agreement about the terms on which this same group will be told, in two quarters, that the thing they funded is not working. That is an uncomfortable object to produce in a room full of sponsors, and it is trivially easy to defer with a sentence everyone accepts: we will work the metrics through with the team.
Once that deferral happens, the other two asks lose their teeth. An approved roadmap with no agreed measure of success cannot fail, it can only be revised. A funded programme with no agreed measure of success cannot be stopped, because stopping requires evidence and nobody defined what would count. The middle ask is the load-bearing one, and it sits in the position on the page that is easiest to skip.
An approved roadmap with no agreed measure of success cannot fail. It can only be revised, indefinitely, by the people it belongs to.
If I were running the session rather than writing the deliverable, I would take the metric ask first, while the room is fresh and before any money is on the table. Once a budget number is in the air, the metric conversation stops being about what good looks like and becomes a negotiation about how the spend will later be defended.
What has to be true before this works
There is a precondition on the whole design, and it is not stated on the slide.
You can only ask a board to produce three documents in a session if the three documents already exist in draft. Nobody authors a ninety-day plan in a board meeting. Nobody invents a dashboard from an empty page with the whole board watching. The ask is not really author this, it is ratify, amend and own this, and that requires that the sprint plan is already sequenced, the candidate metrics are already proposed with their sources named, and the budget envelope is already costed with the roles specified.
That is why this mechanism belongs to whoever prepares the session rather than to whoever chairs it. If you are the data or AI manager carrying this to a board, the work is not the ask slide. The work is having three defensible drafts in the pack so that the ask is answerable in the time available. Put an unanswerable ask in front of a board and you have not forced a decision, you have scheduled a second meeting.
The related discipline, that every heading in a governance framework has to resolve into an object somebody writes and owns, is argued in full in Turn Your Seven Governance Pillars Into Fifteen Documents With Named Owners. What the section divider adds is placement and order. Naming the artefact makes the ask concrete. Announcing it early makes the preceding material work, and sequencing it correctly decides which of the three survives contact with a busy room.
Reusing it on a smaller stage
Very little of this is specific to a bank, or to a board. The same failure mode runs a steering committee, an architecture review and a quarterly business review.
Write the ask page before you write the content, and hold it to three. Give each ask the name of the thing it produces, not the name of the topic it concerns. Put the page at the front of the section it governs. Say what happens to each output afterwards, because an artefact with no downstream consumer is not a decision, it is homework. Then check the precondition honestly: if you cannot hand the room a draft of each of the three, cut the ask rather than asking it badly.
None of this guarantees a good decision. It removes one specific way of avoiding a decision, which is the one that looks most like success while it is happening.
Source: a single section-divider slide from a board deliverable and workshop kit written for a consumer and SME banking group operating across several European markets. The slide previews the implementation half of the deck and states three board decisions with the output each must produce. The same deck carries a separate decision page with a signature block, referenced here only for contrast. It is a proposal and a design, not an executed programme, and it records what the board was to be asked rather than what the board decided. The reading of why the middle ask is the fragile one, and the recommendation to take it first, are mine and are not on the slide.
“Of the three asks, only one commits the board itself rather than the programme. Defining what success will be measured on is the cheapest thing in the room to defer and the only thing that makes the other two enforceable.”
Get in touch
Put RealAI’s applied-AI team on your hardest data problem.
We help enterprises move from pilots to production: sovereign models, governed data, and agents you can audit. Start with a value-first assessment.