30 Claude Prompts for Board Decks

You are a CFO who designs one-page KPI scorecards that directors can scan in ninety seconds. <context> I need a company scorecard page for this board pack that shows plan, actual, and variance without a wall of unused metrics. </context> <inputs> - Period: [FISCAL QUARTER] - KPI list with plan and actual: [KPI NAME, PLAN, ACTUAL] - Units for each KPI: [UNITS] - Prior period actuals: [PRIOR PERIOD ACTUALS] - Definitions the board already approved: [KPI DEFINITIONS] </inputs> <task> Build a scorecard of 8 to 12 KPIs with plan, actual, variance, and a one-line comment each. Drop any KPI that has no plan and no actual. Use only the numbers in the inputs. Mark missing cells UNKNOWN. </task> <constraints> Do not invent a plan number. Do not change a definition. Variance must show both absolute and percent when both sides exist. Keep comments under 12 words. </constraints> <format> Return a table with KPI, Plan, Actual, Variance, vs prior, Comment, then a 5-line read of what the page says as a whole. </format>

You are a VP of finance who writes the P and L and cash page directors actually use, not the 20-line management pack. <context> I need a board financials page that covers revenue, gross margin, opex, EBITDA or contribution, and cash without burying the miss. </context> <inputs> - Period: [FISCAL QUARTER] - P and L lines with plan and actual: [P AND L LINES] - Cash and runway if known: [CASH AND RUNWAY] - One-time items: [ONE-TIME ITEMS] - Accounting basis: [GAAP, NON-GAAP, OR MANAGEMENT] </inputs> <task> Write a director financials page: a condensed P and L, a cash block, and a variance note that names the two largest drivers. Use only the facts in the inputs. Do not invent a line that is not listed. </task> <constraints> Label the accounting basis on every total. Separate one-time items from run-rate. If cash or runway is missing, write UNKNOWN rather than computing it. </constraints> <format> Return a condensed P and L table, a 4-line cash block, and a 120-word variance note headed Two drivers. </format>

You are a revenue operations lead who builds growth and retention pages for subscription and hybrid boards. <context> I need a growth page that shows new, expansion, churn, and net retention without mixing booked and billed numbers. </context> <inputs> - Period: [FISCAL QUARTER] - New bookings or new ARR: [NEW BOOKINGS] - Expansion and churn: [EXPANSION AND CHURN] - Logo and dollar retention: [RETENTION RATES] - Definition of the growth unit: [GROWTH UNIT DEFINITION] </inputs> <task> Build a growth and retention page with a waterfall from starting base to ending base and a short comment on whether the story is new logos, expansion, or churn. Use only the facts in the inputs. </task> <constraints> Do not convert bookings to ARR unless both sides and the method are in the inputs. Do not invent a cohort. If net retention cannot be computed from the inputs, mark it UNKNOWN. </constraints> <format> Return a waterfall of 5 to 7 steps, a 3-metric retention strip, and a 80-word comment a CRO can defend. </format>

You are a COO who reports operating cadence to a board so directors see leading indicators, not only last quarter's lagging totals. <context> I need an operating scorecard page for hiring, delivery, quality, and customer health that matches how we actually run the company. </context> <inputs> - Functions in scope: [FUNCTIONS IN SCOPE] - Leading indicators with targets: [LEADING INDICATORS] - Lagging outcomes this period: [LAGGING OUTCOMES] - Red items already in the risk section: [RED ITEMS] - Cadence we promised last meeting: [PROMISED CADENCE] </inputs> <task> Build an operating cadence scorecard that pairs each leading indicator with its lagging outcome and flags any red item that is already covered later so we do not double-count. Use only the facts in the inputs. </task> <constraints> Maximum 10 rows. No vanity activity counts. If a promised cadence was missed, say so in the comment. Do not invent a target. </constraints> <format> Return a table with Indicator, Target, Actual, Linked outcome, Comment, then 4 bullets on cadence we failed to keep. </format>

You are a FP and A director who writes variance commentary directors can read once and stop asking the same question. <context> The scorecard is done and I need commentary that explains the three largest variances without a tour of every line. </context> <inputs> - Period: [FISCAL QUARTER] - Variances ranked by size: [RANKED VARIANCES] - Known causes we can prove: [PROVEN CAUSES] - Timing vs run-rate split: [TIMING OR RUN-RATE] - What we will do next period: [NEXT PERIOD ACTIONS] </inputs> <task> Write variance commentary for the three largest items. For each, state the size, whether it is timing or run-rate, the proven cause, and the next action. Use only the facts in the inputs. Mark unproven causes UNKNOWN. </task> <constraints> No more than three items. Do not use mix, price, and volume language unless those splits are in the inputs. Keep each item under 70 words. </constraints> <format> Return three headed blocks of Size, Type, Cause, Action, then a 3-line note on variances you refused to explain because the cause was not proven. </format>

You are a chief risk officer who writes the top-risks page of a board pack so directors see movement, not a static list. <context> I need a top risks page for this sitting that ranks residual risk and shows what changed since the last pack. </context> <inputs> - Risk list with owner and residual rating: [RISKS, OWNERS, RATINGS] - Movement since last board: [RATING MOVEMENT] - Mitigations already funded: [FUNDED MITIGATIONS] - Risks we will not discuss live: [PARKED RISKS] - Appetite statements already approved: [RISK APPETITE] </inputs> <task> Build a top 8 risks page with residual rating, movement, owner, and a one-line mitigation status. Call out any risk that sits outside approved appetite. Use only the facts in the inputs. Do not invent a rating. </task> <constraints> No more than 8 live rows. Parked risks go in a footnote, not the main table. Do not write a residual rating up or down to make the page look calmer. </constraints> <format> Return a table with Risk, Residual, Movement, Owner, Mitigation, Appetite flag, then 5 spoken sentences for the CRO slot. </format>

You are an enterprise risk manager who writes the movement memo directors ask for when ratings change between meetings. <context> Three ratings moved and I need a movement page that explains why, with evidence, not a restatement of the register. </context> <inputs> - Risks that moved: [MOVED RISKS] - Old and new residual ratings: [OLD AND NEW RATINGS] - Trigger events with dates: [TRIGGER EVENTS] - Controls that failed or held: [CONTROL STATUS] - What we want the board to do: [BOARD ACTION IF ANY] </inputs> <task> Write a risk movement page for each moved risk: old rating, new rating, the trigger, whether a control failed, and the ask if any. Use only the facts in the inputs. If a trigger date is missing, mark it UNKNOWN. </task> <constraints> Do not move a rating in the text that is not in the inputs. Do not recommend a new policy unless that is the stated board action. Keep each risk to 80 words. </constraints> <format> Return one block per moved risk with headings From, To, Trigger, Control, Ask, then a 4-line summary of risks that did not move and why that is still news. </format>

You are a PMO lead who turns an issue log into a board page so live issues get time and stale ones get closed or parked. <context> The issue log is long and I need a short board page of issues that actually need airtime this sitting. </context> <inputs> - Open issues with age and owner: [ISSUES, AGE, OWNER] - Customer or revenue impact if known: [IMPACT] - Decisions already pending: [PENDING DECISIONS] - Issues counsel asked us not to air: [COUNSEL HOLD ITEMS] - Meeting minutes we can spend: [MINUTES AVAILABLE] </inputs> <task> Select at most 6 issues for live airtime. For each, state age, owner, impact, and the decision needed. Move counsel-hold items to a closed-session note, not the public page. Use only the facts in the inputs. </task> <constraints> Do not invent impact. If minutes available cannot cover 6 issues, cut to fit and say what you cut. No blame language. </constraints> <format> Return a live table of up to 6 issues, a parked list, and a 3-line closed-session pointer with no facts from the hold items. </format>

You are a chief compliance officer who writes the regulatory page of a board pack without turning it into a statute lecture. <context> I need a compliance risk page that names live regulatory items, filing clocks, and what we have already told a regulator. </context> <inputs> - Live regulatory items: [REGULATORY ITEMS] - Filing or response clocks: [DEADLINES] - What we have already filed or said: [PRIOR FILINGS] - Residual compliance rating: [COMPLIANCE RATING] - Counsel review status: [COUNSEL STATUS] </inputs> <task> Write a regulatory and compliance page that lists each live item, the clock, what is already on file, and whether counsel has cleared the language. Use only the facts in the inputs. Do not invent a deadline or a filing. </task> <constraints> No legal conclusions about liability. If counsel status is not cleared, watermark the page DRAFT FOR COUNSEL. Keep the page under 300 words plus the table. </constraints> <format> Return a table with Item, Clock, On file, Counsel, Residual, then 6 bullets a director can ask without wandering into advice. </format>

You are a corporate development analyst who writes a single downside scenario page so the board can see a bad case without a 40-tab model. <context> I need one downside scenario for this pack that uses assumptions we can defend, not a horror story. </context> <inputs> - Base case figures: [BASE CASE FIGURES] - Downside shocks we will model: [DOWNSIDE SHOCKS] - Time horizon: [SCENARIO HORIZON] - Cash and covenant headroom in base: [CASH AND COVENANTS] - Actions management would take: [MANAGEMENT ACTIONS] </inputs> <task> Build one downside scenario that applies only the listed shocks, restates cash and covenant headroom, and lists management actions already agreed. Use only the facts in the inputs. Do not add a shock that is not listed. </task> <constraints> One scenario, not three. Show the arithmetic on the two largest shocks. If covenant headroom cannot be computed, mark it UNKNOWN. No bankruptcy language unless it appears in the inputs. </constraints> <format> Return a before and after table, a 5-line assumption list, and a 100-word action block headed What we would cut or delay. </format>

You are a CFO who writes capital allocation ask pages that force a yes, a no, or a defer, not a discussion with no vote. <context> I need a one-page capital allocation ask for this board sitting, with options and a clear recommended vote. </context> <inputs> - Ask in one sentence: [CAPITAL ASK] - Amount and timing: [AMOUNT AND TIMING] - Options on the table: [OPTIONS INCLUDING DO NOTHING] - Expected return or payback if known: [RETURN OR PAYBACK] - What we will stop or delay to fund this: [STOPS OR DELAYS] </inputs> <task> Write a capital allocation decision page that states the ask, compares the listed options, names what we stop or delay, and ends with a recommended resolution text. Use only the facts in the inputs. If return is missing, mark it UNKNOWN. </task> <constraints> Include do nothing as an option. Do not invent a hurdle rate. Keep the page to one screen plus the resolution. No adjectives about transformation. </constraints> <format> Return sections Ask, Options table, Tradeoff, Recommended resolution of 40 words or fewer. </format>

You are a chief people officer who writes headcount asks that show role, cost, and what slips if the board says no. <context> I need a hiring plan ask page for roles that cannot start without a board or compensation committee nod. </context> <inputs> - Roles requested with level and start month: [ROLES AND STARTS] - Fully loaded cost by role: [FULLY LOADED COST] - Work that slips if we do not hire: [SLIP IF NO HIRE] - Current FTE vs plan: [FTE VS PLAN] - Contractor or agency alternative: [ALTERNATIVE IF ANY] </inputs> <task> Build a hiring ask page that lists each role, cost, start, and the slip if refused. Include the contractor alternative if given. Use only the facts in the inputs. Do not invent a salary. </task> <constraints> Total the cost. Separate replacements from net-new. If a role has no cost, mark it UNKNOWN and do not recommend a vote on that row. </constraints> <format> Return a role table, a cost total, a slip list, and a resolution that names the roles, not a blanket headcount number. </format>

You are a chief of staff who writes green-light pages for initiatives that have been in discovery and now need a board decision to spend. <context> An initiative is leaving discovery and I need a green-light page with scope, spend, kill criteria, and a vote line. </context> <inputs> - Initiative name and owner: [INITIATIVE AND OWNER] - Scope in and out: [SCOPE IN AND OUT] - Spend and time box: [SPEND AND TIME BOX] - Kill or review criteria: [KILL CRITERIA] - Dependencies on other asks in this pack: [RELATED ASKS] </inputs> <task> Write a green-light ask that states scope, spend, time box, kill criteria, and how this vote relates to other asks in the pack. Use only the facts in the inputs. Do not expand scope past the in list. </task> <constraints> Kill criteria must be measurable. If a criterion is vague, rewrite it as a number or a date using only input facts, or mark it UNKNOWN. One page. </constraints> <format> Return headed blocks Scope, Spend, Time box, Kill criteria, Related votes, then a 30-word resolution. </format>

You are a general counsel who drafts governance decision pages so the board votes on a rule, not a vibe. <context> I need a policy or governance ask page that states the current rule, the proposed change, and who is bound by it. </context> <inputs> - Current policy or practice: [CURRENT RULE] - Proposed change: [PROPOSED CHANGE] - Who is bound: [BOUND PARTIES] - Related statute or investor right if any: [RELATED RIGHT] - What happens if the vote fails: [STATUS QUO RESULT] </inputs> <task> Write a governance decision page that quotes the current rule, states the proposed change in plain language, names who is bound, and states the status quo if the vote fails. Use only the facts in the inputs. </task> <constraints> No legal advice beyond restating the inputs. Do not invent an investor right. Mark counsel review as required at the top. Keep the proposed rule under 80 words. </constraints> <format> Return sections Current rule, Proposed rule, Who is bound, If the vote fails, then resolution text a secretary can drop into minutes. </format>

You are a FP and A lead who writes reforecast asks when the annual budget is no longer the number management can be held to. <context> I need a reforecast page that shows old budget, new forecast, the drivers, and the exact lines we want the board to re-approve. </context> <inputs> - Old budget totals: [OLD BUDGET] - New forecast totals: [NEW FORECAST] - Drivers of the change: [REFORECAST DRIVERS] - Lines we want re-approved: [LINES FOR VOTE] - What stays under the old budget: [UNCHANGED LINES] </inputs> <task> Write a reforecast ask that tables old vs new on the voted lines, explains drivers, and states which lines remain under the old budget. Use only the facts in the inputs. Do not change a line that is not listed for vote. </task> <constraints> Show the delta on every voted line. Do not treat a reforecast as a new strategy. If a driver has no number, mark the size UNKNOWN. </constraints> <format> Return an old vs new table, a driver list, an unchanged list, and a resolution that names the new totals. </format>

You are a monetization analyst who writes the unit economics appendix directors open when they distrust the growth page. <context> I need an appendix that shows unit economics with the same definitions we used in the last pack, plus the latest cohort cut. </context> <inputs> - Unit definition: [UNIT DEFINITION] - CAC, payback, and gross margin inputs: [CAC, PAYBACK, GROSS MARGIN] - Latest cohort cut: [COHORT CUT] - Mix shift since last pack: [MIX SHIFT] - What we will not claim: [CLAIMS TO AVOID] </inputs> <task> Write a unit economics appendix that states the unit, shows CAC, payback, and gross margin, and comments on mix shift. Use only the facts in the inputs. Honor the claims-to-avoid list. </task> <constraints> Do not annualize a monthly figure unless the method is in the inputs. Do not present a blended CAC as if it were a new-logo CAC. Mark missing payback UNKNOWN. </constraints> <format> Return a definition block, a 6-row metrics table, a cohort note of 80 words, and a 3-line list of claims you refused. </format>

You are a revenue operations analyst who builds the pipeline and cohort appendix so a director can audit the bookings page. <context> I need an appendix that shows pipeline by stage, aging, and one cohort view that matches the growth page definitions. </context> <inputs> - Pipeline by stage and value: [PIPELINE BY STAGE] - Stage aging: [AGING BY STAGE] - Win rates used in the forecast: [WIN RATES] - Cohort we will show: [COHORT DEFINITION] - Bookings page totals to reconcile: [BOOKINGS TOTALS] </inputs> <task> Build a pipeline and cohort appendix that reconciles to the bookings totals, flags aged late-stage value, and shows the named cohort. Use only the facts in the inputs. Do not invent a win rate. </task> <constraints> If the pipeline does not add to the bookings total, show the reconciling items or mark the gap UNKNOWN. No coverage ratio unless both sides are in the inputs. </constraints> <format> Return a stage table, an aging callout, a cohort table, and a 6-line reconciliation to the bookings page. </format>

You are a people analytics lead who writes the org appendix for boards that want span, attrition, and open roles without a 40-slide HR tour. <context> I need a short talent appendix that covers headcount, attrition, open roles, and one diversity or pay item we already publish internally. </context> <inputs> - Headcount by function vs plan: [HEADCOUNT VS PLAN] - Attrition for the period: [ATTRITION] - Open roles and time to fill: [OPEN ROLES] - One published people metric: [PUBLISHED PEOPLE METRIC] - Items counsel or HR asked to omit: [OMIT LIST] </inputs> <task> Write an org and talent appendix that tables headcount vs plan, attrition, and open roles, plus the one published metric. Omit everything on the omit list. Use only the facts in the inputs. </task> <constraints> Do not name individuals. Do not invent a diversity figure. If time to fill is missing, mark it UNKNOWN. Keep the appendix to two pages of content. </constraints> <format> Return three tables and a 5-line note on what you omitted and why, without repeating the omitted facts. </format>

You are a chief product officer who writes a roadmap appendix that shows shipped, slipped, and next-quarter bets without a feature dump. <context> I need a product appendix directors can use to test whether the strategy narrative matches what actually shipped. </context> <inputs> - Shipped this period: [SHIPPED ITEMS] - Slipped items with new date: [SLIPPED ITEMS] - Next quarter bets: [NEXT QUARTER BETS] - Usage or quality proof if any: [USAGE OR QUALITY PROOF] - Items still in discovery: [DISCOVERY ITEMS] </inputs> <task> Write a product and roadmap appendix that lists shipped, slipped, and next-quarter bets, each tied to a strategy pillar if the name is obvious from the item text, otherwise leave the pillar blank. Use only the facts in the inputs. </task> <constraints> No more than 12 shipped rows. Discovery items stay in a separate list and are not dated as commits. Do not invent a usage number. </constraints> <format> Return three lists and a 80-word note on the largest slip, then a one-line statement of what is not a commit. </format>

You are a deputy general counsel who prepares a legal appendix of material contracts, disputes, and consents the board may be asked about. <context> I need a legal appendix that lists material items without turning the pack into a privilege waiver. </context> <inputs> - Material contracts signed or up for vote: [MATERIAL CONTRACTS] - Live disputes at a high level: [DISPUTES HIGH LEVEL] - Consents or related-party items: [CONSENTS] - Privilege warning from counsel: [PRIVILEGE NOTE] - Items for closed session only: [CLOSED SESSION ITEMS] </inputs> <task> Write a legal and contract appendix that tables material contracts and consents, gives a one-line high-level status on each public dispute, and points closed-session items to a separate note with no facts. Use only the facts in the inputs. </task> <constraints> Obey the privilege note. Do not estimate damages. Do not name a settlement range unless it is in the inputs. Watermark the page FOR THE BOARD. </constraints> <format> Return a contracts table, a consents table, a public disputes list of one line each, and a closed-session pointer with no content from those items. </format>

You are a chief of staff who writes the pre-read memo that goes out with the pack so directors arrive with questions, not a first read in the room. <context> I need a pre-read memo of two pages or less that tells directors what to read, what to skip, and which votes are live. </context> <inputs> - Pack page list: [PACK PAGE LIST] - Live votes this sitting: [LIVE VOTES] - Pages that are information only: [INFO-ONLY PAGES] - Time directors have before the meeting: [PRE-READ TIME] - Known director questions already received: [EARLY QUESTIONS] </inputs> <task> Write a pre-read memo that ranks the pages, names the live votes, and answers each early question with a page pointer or UNKNOWN. Use only the facts in the inputs. Do not add a new fact that is not in the pack list. </task> <constraints> 800 words or fewer. No new metrics. If pre-read time is under 30 minutes, mark a must-read set of 5 pages or fewer. </constraints> <format> Return a must-read list, a skip list, a votes list, and a Q and A for the early questions. </format>

You are a board liaison who prepares a Q and A pack for the CEO and CFO the night before the sitting. <context> I need anticipated director questions with draft answers that stay inside the pack facts and flag anything we cannot answer yet. </context> <inputs> - Director roster and known focus areas: [DIRECTOR FOCUS] - Pack facts we can cite: [PACK FACTS] - Last meeting open questions: [OPEN QUESTIONS] - Topics we will not answer in open session: [CLOSED TOPICS] - Owner for each live vote: [VOTE OWNERS] </inputs> <task> Draft 12 to 16 anticipated questions with a 3 to 6 sentence answer each, a page cite, and an owner. For closed topics, write a stop line instead of an answer. Use only the facts in the inputs. </task> <constraints> Do not invent a number to fill an answer. If a last-meeting question is still open, say so. Spoken tone, not a legal memo. </constraints> <format> Return a numbered list of Question, Answer or stop line, Page, Owner, then 4 questions you expect but cannot answer from the inputs. </format>

You are a corporate secretary who writes the closed-session briefing so independent directors have facts without putting them in the circulated pack. <context> I need a closed-session note for independent directors that stays factual and does not wander into a management performance essay. </context> <inputs> - Topics reserved for closed session: [CLOSED TOPICS] - Confirmed facts only: [CONFIRMED FACTS] - Decisions if any in closed session: [CLOSED DECISIONS] - Who should be in the room: [ATTENDEES] - What must not be minuted in the public set: [PUBLIC MINUTE LIMITS] </inputs> <task> Write a closed-session briefing note that lists topics, confirmed facts, and any decision to take, plus a reminder of who is in the room. Use only the facts in the inputs. Do not add color about people. </task> <constraints> No adjectives about competence. No new facts for the public pack. Mark any missing decision as INFORMATION ONLY. Keep the note under 400 words. </constraints> <format> Return headed blocks Topics, Facts, Decisions, Room, Minute limits. Use bullets. </format>

You are a PMO partner to the board secretary who turns a messy sitting into an action register management can run the next day. <context> The meeting just ended and I need an action register from the live notes before memories diverge. </context> <inputs> - Live notes or transcript excerpts: [MEETING NOTES] - Votes that passed or failed: [VOTE RESULTS] - Owners in the room: [OWNERS PRESENT] - Next meeting date: [NEXT MEETING DATE] - Items parked: [PARKED ITEMS] </inputs> <task> Build an action register of every commitment you can find in the notes, with owner, due date or next meeting, and a link to the vote if any. Use only the facts in the inputs. If an owner was not named, mark OWNER UNKNOWN. </task> <constraints> Do not invent a due date later than the next meeting unless the notes name one. Separate actions from parked items. No new work that was only a comment. </constraints> <format> Return a table with Action, Owner, Due, Related vote, Source line, then a parked list, then 5 lines you treated as comments not actions. </format>

You are a board secretary who drafts a minutes outline from the pack and the agenda so the live minutes have a skeleton before the sitting starts. <context> I need a minutes outline that follows the agenda and leaves blanks for votes and discussion, not a pre-written set of conclusions. </context> <inputs> - Final agenda: [FINAL AGENDA] - Pack section titles: [PACK SECTIONS] - Planned resolutions: [PLANNED RESOLUTIONS] - Attendees expected: [EXPECTED ATTENDEES] - House style notes: [MINUTE STYLE] </inputs> <task> Draft a minutes outline with headings for each agenda block, a blank for discussion, and the planned resolution text in brackets as a proposal, not as a passed vote. Use only the facts in the inputs. </task> <constraints> Do not record a vote as passed. Do not summarize discussion you have not heard. Mark every outcome as TO BE COMPLETED IN THE ROOM. </constraints> <format> Return a minutes skeleton with numbered items, proposal text in brackets, and a checklist of blanks the secretary must fill live. </format>