An exec update that reads like a status report wastes the executive’s attention and your political capital. They don’t need “great progress on initiative Q” three weeks in a row; they need the call where progress is now at risk and the decision only they can make. A good update puts a decision in line 1, names risks honestly with a colour code that means something, lists the asks the exec must respond to, and skips pleasantries entirely. The templates below cover weekly skip-levels, monthly board updates, and investor reports. Pair them with the executive summary prompts when an update needs to compress a 10-page memo into a top-of-email TL;DR.
TL;DR
- Lead with the decision or ask, not the recap. Investors and the VC standard (intro, highlights, lowlights, metrics, asks) both reward this, per Visible.vc.
- Pick one hero metric and report it in every update so the trend line is readable.
- Keep three ask types separate: advice, an intro/referral, or money. Make each one explicit.
- AI drafts the structure; you supply the real numbers. AI invents plausible figures, so never let it fill a KPI cell.
- For tone, Claude Sonnet 4.6 or Opus 4.7 read most human on long-form prose as of June 2026; GPT-5.5 is the value pick. Either works for these short prompts.
Who this is for
Functional leads writing weekly updates, founders writing investor and board updates, and anyone reporting up to a busy executive.
When not to use these prompts
Don’t use them to bury bad news; execs find out anyway, and the next update lands with less trust. Don’t use them as a comprehensive log of what your team did all week.
Prompt anatomy
Every prompt should carry six elements:
- Role: who AI plays (chief of staff, manager, analyst).
- Context: team, org, scope, data.
- Goal: one deliverable (plan, memo, talking points, doc).
- Constraints: word count, must-include fields, audience seniority.
- Tone: confident, neutral, factual.
- Examples: one or two prior samples to anchor format.
12 copy-ready prompt templates
1. Weekly CEO update
Weekly update to CEO. Format: (1) Top 3 wins (with metric), (2) Top 2 misses (honest), (3) 1 decision I need from you, (4) 1 risk you should know. <= 200 words.2. Monthly investor update
Monthly investor update: (1) Hero metric, (2) 3 key KPIs vs target, (3) Wins (1-2 concrete examples), (4) Asks (intros, advice), (5) Risks. <= 400 words. Skip "we're grinding".3. Board-prep memo
Board meeting next week. Memo: (1) Quarter scorecard, (2) Strategic shifts, (3) Open questions for the board, (4) Pre-read links. 1 page max. Skip cheerleading.4. Skip-level update
Skip-level update to [name]. Cover: (1) What's going well in my team, (2) Where I need air cover, (3) One concern about cross-team friction, (4) A question for them. <= 200 words.Variables to swap: name
5. Status to action conversion
My draft update has 6 status bullets but no asks. Rewrite to surface decisions: convert at least 2 status items into "Decision needed from you on X by date Y".6. Pulling the alarm responsibly
Something is going wrong. Write a Sev-2 update: (1) What's happening, (2) Impact, (3) What we're doing, (4) What I need from you, (5) Next update timing. Calm tone, don't catastrophise.7. Wins that aren’t bragging
Frame these wins so they read as factual signals: [wins]. Skip "team is amazing", focus on what changed for the business. Each win <= 25 words.Variables to swap: wins
8. Misses without spin
I have 2 misses this period. Frame each: (1) What missed, (2) Why (honest, no scapegoats), (3) What I've changed, (4) Next signal to watch. Don't hedge.9. Hero-metric callout
My update should have 1 hero metric. Pick from: revenue, growth rate, conversion, NPS, headcount. Lead with it in the subject line. Body explains.10. Frequency calibration
I send weekly. Audit: (a) Am I sending too often (no real news), (b) Too rarely (surprise), (c) Right cadence. If too often, propose dropping to biweekly with a daily-by-exception rule.11. Update template per audience
I have 3 audiences (board, investors, internal exec). Build 3 templates with different sections / depth / tone. Don't use one template for all.12. Update hygiene audit
Audit my last 4 updates: (1) Average length, (2) Decisions actually asked for, (3) Wins-to-misses ratio (suspicious if all wins), (4) Same risk recurring without movement. Fix list.Color code that means something
A green-yellow-red status is only useful if the colors carry a rule, not a vibe. Use this in your prompt context so AI applies it consistently:
| Color | Meaning | What the exec does |
|---|---|---|
| Green | On plan, no action needed | Reads, moves on |
| Yellow | At risk, owner has a plan | Notes it, may ask one question |
| Red | Off plan or blocked, needs a call | Decides or unblocks now |
If everything is green four updates running, the exec stops reading. Yellow and red are what earn their attention.
Feeding real data to the model
These prompts produce structure, not facts. Paste your raw numbers, last update, and OKR doc, then let AI assemble them.
- Long pastes: As of June 2026, ChatGPT Plus, Pro, and Business turn any paste over ~5,000 characters into an attachment so it doesn’t eat the context window, per Notebookcheck.
- Context room: Claude Sonnet 4.6 and Opus 4.7 hold 1M tokens as standard; ChatGPT Plus is around 320 in-app pages, with the full 1M reserved for the $200 Pro tier. For a quarter of raw metrics either is plenty.
- Never trust generated numbers: AI will fill a blank KPI cell with a confident, wrong figure. Mark any cell it shouldn’t touch with
[VERIFY]and check it yourself before sending.
Common mistakes
- No specific context, so the output is generic.
- Skipping fact-check, so AI’s invented numbers ship.
- Vague audience, so the draft over- or under-shoots seniority.
- No word limit, so readers don’t finish.
- Same template every situation, so readers tune out.
- No “decision needed” framing.
- Forgetting to attach the source data.
How to push results further
- Specify the audience level (board, investor, direct manager).
- Cap length: 1 page tactical, 3 bullets executive.
- Lead with the ask or decision needed.
- Attach the source-data link.
- Read it aloud before sending; awkward sentences surface fast.
- AI drafts, you review and own every number.
- Save your best sent updates as examples and reuse the format, refreshing only the content.
FAQ
- How long should an exec update be? Tactical update: one page. Executive or board read: three bullets plus a pre-read link. Investor updates run longer but cap at ~400 words for the email body, with detail in attachments.
- Which AI model is best for this? For natural tone on longer memos, Claude Sonnet 4.6 or Opus 4.7 read most human as of June 2026; GPT-5.5 is the value pick at $20/month on ChatGPT Plus. For these short prompts the gap is small, so use what you already pay for.
- Can AI replace the analyst or chief of staff? It drafts structure and templates well. Judgment, the actual decision framing, and number accuracy stay with you.
- Should risks always be included? Yes. An all-green update reads as either spin or inattention, and pretending no risk exists erodes trust the moment reality lands.
- How do I fact-check the draft? Paste the source data, mark generated cells with
[VERIFY], and have a peer eyeball any number that drives a decision. - Can AI generate the data itself? No. It produces plausible-looking numbers with no source. Connect or paste real data every time.