ingridleiria Board Deck
A board pack has two jobs and only two: tell the board the truth about the condition of the company, and get the decisions the company needs from the people empowered to give them. The failure this prevents is the pack that does neither while appearing to do both. It reports at length, looks complete, contains no visible untruth, and leaves the board with nothing to decide and no way to tell whether the business is in trouble. The meeting fills with clarifying questions about page 14, and the two items that actually needed the board are reached with eleven minutes left.
The cost compounds. A board trained to receive reports stops working as a check, which removes the only structural protection a chief executive has against their own blind spots. And a director who finds a miss for themselves, on a page management wrote, concludes they are being managed rather than informed. From then on every number is read as advocacy.
When to use this, and when not to
Use it for material going to a formal governance body: a board of directors, an advisory board, an investment committee, a trustee board, a joint venture steering group.
Do not use it for the work around the meeting: the annual calendar, pre-wiring difficult items one to one, running the room, minutes, and contact between meetings. That is board-and-investor-management, and it is the larger half of the job. A flawless pack that delivers bad news for the first time is a failure the pack cannot fix.
Three adjacent cases belong elsewhere. The periodic written note to shareholders who are not directors is investor-update, whose readers track a trajectory rather than decide. Persuading someone to commit money they have not committed is pitch-deck; a board is already committed, and that changes what evidence they need. Assembling what an investor demands in diligence is fundraise-readiness; both rest on the same numbers, but a pack is a narrative for people who know the company and a data room is an index for people checking it. Each decision paper inside the pack follows decision-memo.
What you need before starting
Board composition and what each director reads for. Investor directors read metrics against plan and the cash line. Independents read risk, governance and people. Observers read everything and repeat it to their firm. Missing: ask for the director list with affiliations, and meanwhile write for the most sceptical reader in the room.
Minutes and actions from the last meeting. Every open board request gets a status in this pack, including the uncomfortable ones. Missing: reconstruct from notes and the calendar, mark it as reconstructed, and ask the chair to confirm. Silently dropping a request teaches a board that its requests are optional.
Actuals against plan, closed to a stated date. A pack cannot be built on moving numbers. Missing: use the last closed period, state the close date on the dashboard, and show the current period separately and labelled as flash.
Written metric definitions. What counts as a customer, when revenue is recognised, whether churn is logo or value, what sits inside burn. Missing: write them into the appendix now, and flag any definition differing from the last pack, with the prior period restated.
The decisions being requested. Approvals, appointments, budget, financing, option pool, major contracts, policy. Missing: ask the chief executive directly, and if the answer is genuinely none, write "no decisions requested this meeting" on page one rather than leaving the board to guess.
The sensitive matters. Personnel, litigation, an approach from an acquirer, an investigation, compensation. Missing: ask the chief executive and the chair separately, because each knows things the other has not raised. These belong in executive session, never in a pack that reaches observers.
The method
Fix the close and freeze the numbers. Confirm the format and the material deadline first, defaulting to a pre-read of fifteen to twenty-five pages circulated five days ahead. Then pick the close date, pull every figure from one source, and record the source. From here no number changes without changing every other instance of it. Judgement: wait for a clean close if it lands before the material deadline, publish labelled flash if not, and never publish flash unlabelled.
Write the follow-up status first. Mark each of last meeting's actions done, in progress with a date, or not done with a reason. First rather than last, deliberately: it surfaces the item nobody wants to raise while there is still time to handle it properly.
Draft the dashboard and let it tell you the story. Ten to fifteen metrics, same ones in the same order every meeting, each with actual, plan, prior period and variance. Mark variance only, in three states: on plan, within tolerance, off plan, with the tolerance set numerically and stated. Then read it as a director would and write the three questions it raises. Those questions are the agenda.
Write the CEO letter in prose, in the chief executive's voice. One page, five paragraphs: what happened, what went well, what did not and why, what management is doing, what the board is being asked. Prose rather than bullets, because bullets let a writer imply a causal story without committing to one. Judgement on how much bad news leads: if a director would call it the most important fact of the quarter, it goes in the first paragraph.
Build the financials with causes attached. Profit and loss against budget, cash and runway, balance sheet summary, and the year forecast with the assumptions that changed since last meeting listed explicitly. Set and state a variance threshold, and give every variance above it a one-line cause rather than a restatement of the variance.
Write one page per function. Commercial, product, operations, people: a headline that is a finding, the key numbers, what changed, the risks. "Enterprise pipeline doubled while small business churn rose to four percent" is a headline; "Commercial update" is a label. If nothing changed in a function, say so in one line and save the page.
Frame strategic topics as questions. One to three, each with the question, the options, management's current view, and what would change it. A topic qualifies only if the board's answer could change what management does. Everything else is reporting.
Give each decision its own page, with resolution language. The decision in one sentence, the rationale, the alternatives, the risk, and the exact wording the secretary can minute. Drafting the resolution converts a discussion into a decision, and its absence is why items get deferred. Everything that is not a letter, a dashboard, a functional page, a strategic question or a decision goes to the appendix, on one test: would a director's understanding change if this page were removed?
Build the live deck from the pack, not alongside it. Ten slides: the headline, the dashboard, the largest variances, the strategic questions, the decisions. Assume the pre-read was read; presenting it again teaches directors that reading it was unnecessary, and next quarter they will not.
Reconcile as a checklist before circulating. Every recurring number identical everywhere to the same rounding, every definition matching the appendix, every prior request carrying a status. Run it as a checklist rather than a read, because reading finds prose errors and misses arithmetic.
The section order, the agenda, and the charts
Section order is stable and should not vary between meetings without telling the board why: CEO letter, dashboard, financials, performance by area, strategic topics, decisions, people, risks and compliance, follow-ups, appendix.
At least half the meeting goes to strategic topics and decisions; reporting gets twenty minutes and no more. Put routine approvals on a consent agenda and pass them in one motion. Protect the last substantive item by scheduling it ahead of the item most likely to overrun, since the final slot is where important decisions get three rushed minutes.
Charts show trend over time with the plan line visible, one message each. Monochrome first, series separated by marker shape, dash pattern and fill texture rather than colour, so the pack survives printing and a reader with colour vision deficiency. Legend outside the plot area, one accent colour per page at most.
Worked example
Situation. A business software company of about 140 people, three quarters after a Series B. Six directors, including two investor directors and an independent who chairs audit and reads financials line by line. The quarter missed plan on new bookings by 18 percent, while net revenue retention rose from 104 to 112 percent because an expansion motion launched the prior quarter worked better than expected. Cash was 14 months. The chief executive wanted to lead with retention.
Task. A pack circulating in six days for a three hour meeting, needing two decisions: an increase to the option pool, and whether to open a second sales region in the first half of next year or wait.
Action. The first draft led with retention across two pages and put the bookings miss on page nine inside the commercial update. It was abandoned after one test: reading the page titles in order. They said retention was strong, product shipped on time, pipeline was building. A director reading only titles would not have learned that the company missed its primary plan metric by nearly a fifth.
The rewrite put the miss in the first paragraph of the letter, with the cause stated: two of four enterprise representatives hired the prior quarter had not reached quota, and the plan assumed a four month ramp where the observed ramp was closer to seven. Retention stayed in the second paragraph, as what had offset the miss.
The dashboard kept the same fifteen metrics in the same order, which required restating one prior figure: the definition of an active customer had changed during a billing system migration. The restatement was footnoted and both bases shown, which took four hours and prevented the audit chair finding an unexplained discontinuity.
The region question was reframed from an information item to a decision paper, because it committed roughly 1.1m of the next twelve months of spend. Three options: open in the first half, defer pending two quarters of ramp data, or contract coverage through an existing partner. Management recommended deferring, with a numeric trigger: proceed when the new representatives reach 70 percent of quota for two consecutive quarters.
Result. The pack circulated five days ahead at 21 pages plus a 30 page appendix. Reporting was held to its twenty minutes; the balance of the three hours went to the decisions and one strategic topic. The option pool passed on consent because the dilution table had already answered the only question. The region decision passed as recommended, with the trigger written into the minutes, so it did not recur the following quarter: management simply reported against the trigger.
A second scenario, where it goes differently
The same company two years later, in a quarter where the chief executive has begun exploring a sale and one investor director is conflicted because their fund holds a position in a likely acquirer.
The format does not change; what changes is what is in it and where it goes. The exploration is not in the main pack at all. It goes to the chair first in a call, then to an executive session item with a confidential note to directors only, not to observers and not through the usual circulation. The conflicted director is named in the note and a process for managing the conflict is proposed rather than assumed.
The main pack still reports the quarter honestly and requests its ordinary decisions. The temptation is to thin it because the real conversation is elsewhere. Resist it: a pack that goes suddenly light in the quarter before a transaction is exactly the artefact a future diligence process examines.
What changed is distribution and disclosure route, not the standard.
Output
The pre-read pack, in the fixed section order above. The dashboard:
| Metric | Definition ref | Actual | Plan | Variance | Prior period | Same period last year | Status | | Net new ARR | A.1 | | | | | | off plan | | Net revenue retention | A.2 | | | | | | on plan | | Cash and runway (months) | A.7 | | | | | | within tolerance |
Each decision page:
DECISION REQUESTED: [one sentence, in the form the board will vote on]
Recommendation: [management's recommendation, one sentence]
Why now: [what forces the timing]
Options considered: [each with cost, what it gives up, what must be true]
Risk: [two or three, each with its early signal]
Financial effect: [figures, including dilution or cash effect]
Resolution: "RESOLVED, that ..." [exact language for the minutes]
The follow-up page:
| Meeting date | Request | Owner | Status | Note |
The live deck, ten slides at most, each titled with a sentence stating its finding. The executive session list, to directors only, with the reason each item is in session.
Failure modes
The pack that reports without deciding. Count pages: if the decisions section is shorter than the functional updates and no motion is proposed, the board has become an audience. Ask what management wants to be different after this meeting, and write that as a decision paper.
Bad news buried mid-pack. Read the page titles in order and ask whether a director would learn the worst fact of the quarter. This is rarely deliberate; it is the natural output of writing each section separately. Fix by making the letter carry it.
Numbers that do not tie. Revenue at 4.2m on the dashboard and 4.18m in the financials. A director who finds one discrepancy checks everything else, so reconcile against a single source as a checklist rather than a read.
A definition that moved without notice. Recognise it as a trend line changing slope at a system migration or reorganisation. Restate the prior period, show both, footnote the change. Never smooth it.
The strategic topic that is really a report. Recognise it because there is no question, or the question has an obvious answer. Cut it, or find the open question underneath. The same test catches a decision page with no resolution language: if the minutes could only record a discussion, it was not a decision item.
Edge cases
A crisis between meetings. Do not wait for the pack. Contact directors when it happens, then record it factually with timeline, response and what changed. The pack is not the notification channel; crisis-and-incident-comms covers that.
First meeting after a financing. New directors have no context. Add a one-time orientation appendix: metric definitions, the operating model, the organisation chart, the plan, and the terms of anything they will be asked to approve.
An advisory board with no fiduciary duty. Same structure, no decision pages, much heavier weight on strategic questions. Asking advisers to approve things confuses everyone about what they are for.
A board that does not read the pre-read. Do not lengthen the pack or present it in full. Shorten it, put the asks on page one, and have the chair set the expectation; if it persists, the problem is the chair, and a page that is unintelligible without narration is your half of it.
Litigation or an investigation. Nothing goes in the main pack. Take advice on privilege first, then handle it in executive session with counsel present.
Quality bar
- A director reading only the CEO letter and the dashboard knows the condition of the company and what is being asked of them.
- Every miss against plan is stated by management before a director could find it, with a cause rather than a description.
- The page titles, read alone and in order, carry the honest story of the quarter.
- Every decision requested has resolution language ready to be minuted.
- Every request from the previous meeting has a status, including those not done.
- Every recurring number is identical everywhere, to the same rounding, from one stated source.
- Metric definitions are unchanged, or the change is flagged and the prior period restated.
- The main pack contains nothing that belongs in executive session.
Adapting this to your context
The shape here comes from a venture-backed board of six directors meeting quarterly for three hours on a fifteen to twenty-five page pre-read. Adjust it before adopting it.
- The twenty-minute reporting cap. Set against a three-hour meeting where the pre-read is read. Where the board meets for ninety minutes, or will not read ahead, cut the pack rather than raising the cap, and hold the ratio at roughly one part reporting to two parts deciding.
- Ten to fifteen dashboard metrics. Suited to a single-product software company. A services business tracks utilisation, backlog and realisation; a charity tracks restricted and unrestricted funds and months of reserves. Keep the count, change the rows.
- Fifteen to twenty-five pages, five days ahead. Move both to fit your directors' actual reading, and where a notice period is constitutional, the notice period wins.
- Resolution language. Written for a board that passes motions. Advisory boards and steering groups resolve nothing, so drop the decision pages and put the weight on strategic questions instead.
- What not to change. Management states every miss before a director could find it, and every recurring number is identical everywhere, from one stated source.
Related skills
board-and-investor-management owns everything around this document: the calendar that sets the deadline, the pre-wiring that means nothing in the pack is a surprise, the conduct of the meeting, and the minutes. Build the pack inside that process, not instead of it. financial-model-builder produces the plan the pack reports against and revenue-forecast the commercial numbers on the dashboard. decision-memo is the method for each decision page. investor-update is the shorter periodic letter to non-director shareholders, and must agree with the pack without repeating it. fundraise-readiness reuses these numbers in a data room, where inconsistency with old packs will be found. principal-simulator rehearses how a specific director will react before circulation.