A ten-point checklist for directors and sponsors to run against any decision reaching the board, covering evidence, alternatives, dissent and reversibility.
Board papers are written to be approved. This checklist exists to test whether they deserve to be. It is free to copy, with nothing gated behind a form, and it works as well for an investment committee as for a full board.
Last reviewed 3 July 2026 · Free and ungated
Put the board paper to selected senior operators from the Global Board before the vote is called and receive a confidential report on what the room is not seeing.
Pressure-test this decisionBy the time a proposal reaches the boardroom it has usually survived months of internal refinement, which means the weakest assumptions have been polished rather than tested. Directors are then asked to approve in ninety minutes what took a team ninety days to construct. A checklist does not replace judgement; it gives judgement somewhere specific to look.
| Check | What good looks like | The gap that usually appears |
|---|---|---|
| Decision statement | One sentence naming what is being approved, for how much, and by when | The paper describes a programme but never states the actual commitment |
| Alternatives considered | At least two genuine alternatives, including doing nothing, with reasons for rejection | A single option framed against a strawman |
| Evidence base | The three assumptions the case depends on are named, with their sources | Assumptions are buried in an appendix or inherited from the vendor |
| Financial downside | A quantified worst case, not just a base case with sensitivities | Only upside scenarios are modelled |
| Dissent surfaced | The paper records who disagreed during preparation and why | Unanimity is presented as evidence of quality |
| Reversibility | Explicit statement of what it costs to change course after approval | The commitment is treated as staged when contractually it is not |
| Sponsor exposure | The sponsor's incentives and history with this option are visible to the board | The strongest advocate is also the sole source of evidence |
| Implementation owner | A named executive who has accepted delivery accountability in writing | Ownership is assigned to a role that does not exist yet |
| Conditions of approval | Approval is tied to specific conditions the board can later verify | A blanket yes with no follow-up mechanism |
| Review trigger | A defined event or date at which the decision returns to the board | The board never sees the decision again unless it fails |
The checklist belongs at two moments: when the company secretary or sponsor assembles the board pack, and again in the pre-meeting where the chair decides whether an item is ready for a vote or should be tabled for discussion only. The sponsor completes it first; the chair or a non-executive then re-scores it independently. Divergence between the two scores is itself a finding.
A completed checklist tells you whether the paper is well constructed. It cannot tell you whether the underlying option is right, because everyone scoring it shares the same information and often the same incentives. When the decision is large, hard to reverse, or carried by a single forceful sponsor, put the completed checklist and the paper to people who have made a comparable call elsewhere. Digital Advisory routes exactly this kind of brief to selected senior operators on the Global Board, confidentially, before the vote rather than after.
Yes, as a completed one-page annex. Its presence changes how papers are written: authors who know the eleven checks are coming tend to answer them in the body of the paper rather than hoping nobody asks.
Table it for discussion without a vote. Most failed checks are curable in a fortnight (naming a downside, documenting the rejected alternatives), and a two-week delay is almost always cheaper than approving a case with holes in it.
They are companions. The interactive tool scores a decision in the browser; this template is the printable artefact you attach to the pack and mark up by hand in the meeting.