Stakeholder update that gets read
Goal: Friday updates with traffic-light status + owned asks—facts yours, model only polishes.
Best for: Weekly updates to managers, clients, or cross-team stakeholders who skim on mobile.
Leaders do not need your diary. They need a traffic-light read: are we fine, are we delayed, and what do you need from them? AI is good at turning messy bullets into that shape. It is bad at inventing wins, owners, or dates you never said. Keep the facts yours.
Why beginners skip this
Leading with feelings then bolting on facts feels sincere—stakeholders need three decision lines. Facts first, then risk.
Step 1 — Steal a skim-friendly shape
Paste the empty win / risk / ask template every week. Fill brackets from *your* notes before you chat.
Audience: [manager / client / partner]
Project: [name]
Status color: Green / Yellow / Red — one sentence why
Wins (≤3): [fact bullets only]
Risks (≤3): [what could slip + earliest signal]
Asks (≤3): [person + decision + by when]
Boundaries: Max 150 words. No fluff openers. Do not invent metrics, owners, or dates.
Input notes:
[paste your raw bullets]
Step 2 — One worked week (API migration)
Example
Filled prompt
Audience: eng manager
Project: Checkout API migration
Status color: Yellow — cutover slipped 3 days; canary still green
Wins:
- Migrated read path for 40% of traffic
- Zero Sev-1 in canary week
Risks:
- Write path still on legacy; dual-write lag spikes at 18:00 HKT
- Two contractors end Friday unless extension approved
Asks:
- Approve 1-week contractor extension by Wed noon
- Confirm freeze window Thu 22:00–02:00
Boundaries: Max 150 words. No fluff. Do not invent % complete.
What good output looks like
- Opens with Yellow + one honest reason
- Asks name a person/decision/deadline
- Stays under ~150 words; no “hope you’re well”
- Does not invent a completion percentage you never gave
Step 3 — Kill the fluff before you send
After the draft, run one micro-pass: cut empty openers, fake greens, and any percentage you cannot point to.
Edit for skim only. Keep facts. Delete greetings, “just circling back”, and any metric not in my notes. Return the update + a 3-line “what I removed” list.
If the model adds a sunny Green when your notes said Yellow, reject it. Color is a judgment call—you own it.
Example
Filled prompt
Audience: product director (skims on phone)
Project: Mobile checkout beta
Status color: Yellow — internal beta live; public date blocked on design review
Wins:
- Internal beta live for staff cohort (n≈40)
- Crash-free for 3 days in canary
Risks:
- Public mobile date slips if design review misses Wed
- Copy freeze not signed by legal
Asks:
- Confirm design review Wed 15:00 with Mei
- Legal sign-off on freeze by Thu noon
Boundaries: Max 150 words. No fluff. Do not invent % complete or user counts beyond notes.
What good output looks like
- Opens Yellow with the design-review blocker named
- Asks name Mei + Legal with deadlines
- Stays ≤150 words; no “hope you’re well”
- Does not invent a public launch % or external user count
Example
Filled prompt
Yellow: API cutover +2 days; canary green.
≤3 wins/risks/asks with owners. ≤150 words. No invented % complete.
Context: real workstream this week (Asia/Shanghai). Name actual files/people.
Constraints: no invented metrics/owners/prices; label Unknowns; stop before send/delete/pay.
Done check: I can mark pass/fail in one minute on the output.
What good output looks like
- Matches outline/voice freeze for “Send a short status leaders can skim in about one ”
- No AI-slop openers; one concrete proof from your notes
- Numbers/names frozen — polish does not invent
- Pass/fail: phone-skim 20s still clear
Red flag
Unowned asks; invented metrics; model inventing status—traffic-light with your facts only.
Save this card
Card: Traffic-light status + owned asks—facts are yours; the model only polishes. Keep this beside the tool; read before you start.
Before you start
Name the reader and freeze facts on paper for “Send a short status leaders can skim in about.” If outline bullets do not exist yet, stop before prose. Guide: writing-stakeholder-update. Keep the real job visible the whole time; a blank box means stop and fetch the artifact first.
How to use this week
This week write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Pitfalls
| Pitfall | Fix |
|---|---|
| Pasting a week of Slack and asking “summarize” | Pre-sort into wins / risks / asks yourself |
| Letting the model invent “85% complete” | Ban new metrics; only reuse numbers you typed |
| Three soft asks (“thoughts?”) | Each ask needs owner + decision + date |
| Writing for yourself, not the skimmer | Read it on your phone in 20 seconds |
Example
Filled prompt
Temptation: skip the habit for writing-stakeholder-update to save five minutes.
Job still: Send a short status leaders can skim in
Smallest correct pass: Done visible · checkable artifact · Unknowns [?].
Boundaries: draft only; no silent send; no invented metrics.
Return: artifact + refuse-line said aloud.
What good output looks like
- Names the shortcut refused (specific to writing-stakeholder-update)
- Still delivers checkable output for “Stakeholder update that gets read”
- Unknowns labeled; draft-only
- Pass/fail: refuse-line dated (writing-stakeholder-update #4)
Practice now
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Delta only: lead with what changed since last update (Asia/Shanghai).
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Risk line: one risk + mitigation or “none” (Asia/Shanghai).
Ask line: every update ends with one explicit ask or “no ask” (Asia/Shanghai).
Stakeholder drill: one update with decision-needed line + owner + date; no fluffy status (Asia/Shanghai).
Tonight write a real update: progress / risk / ask in three lines each; move feeling sentences to the end or cut them.
Close the loop
Write one line: next time I lead with feelings then bolt on facts, I refuse send.
One sentence takeaway
Progress / risk / ask three lines each; feelings to the end or cut.
Next
Outline before prose · Reject AI slop · Office weekly routine
能被读完的干系人进展更新
目标: 干系人更新开头就是状态色加一句诚实原因:请求写清对象、决定与截止。
适合: 周报或项目邮件爱写「您好感谢支持」、又爱编绿色指标的人。
经理扫三十秒要知道拍什么板。没有你提供过的完成度百分比,就不要出现。
为什么新手跳过
先写感受再补事实像真诚;干系人要的是决策所需的三行。先事实再风险。
步骤 1——先固定一个好扫的骨架
每周都贴这份空模板。先自己填好括号里的事实,再让模型润色语气。
读者:[经理 / 客户 / 合作方]
项目:[名称]
状态颜色:绿 / 黄 / 红,一句话说明原因
进展(≤3):[只写事实要点]
风险(≤3):[可能滑点的事 + 最早信号]
请求(≤3):[找谁 + 要什么决定 + 截止何时]
边界:最多约 150 词。不要空开场。不要编造指标、负责人或日期。
原始笔记:
[粘贴你的零散要点]
步骤 2——一个真实周例(API 迁移)
实例
填好的提示词
读者:工程经理
项目:结账 API 迁移
状态:黄,切换日推迟 3 天;金丝雀仍绿
进展:
- 读路径已切 40% 流量
- 金丝雀周零 Sev-1
风险:
- 写路径仍在旧系统;双写在香港时间 18:00 附近延迟升高
- 两名外包周五到期,不延期就走人
请求:
- 周三中午前批准外包延期一周
- 确认周四 22:00–02:00 冻结窗口
边界:约 150 词内。不要空开场。不要编造完成百分比。
像样的输出大概长这样
- 开头就是黄灯状态,加一句诚实原因(例如切换推迟三天)
- 每一个请求都写清对象、要拍的板、以及截止时间
- 没有「您好感谢一直支持」这类空开场
- 不出现你材料里没提供的完成度百分比
- 工程经理扫三十秒就知道今天要拍什么板
- 一分钟内能对照清单判过/不过;缺项标红再继续
步骤 3——发出前再砍一次空话
草稿出来后,再跑一轮极短编辑:砍空开场、假绿灯、指不出的百分比。
只为“好扫”而改。保留事实。删掉问候、“顺带同步一下”,以及笔记里没有的指标。返回正文 + 三行“我删了什么”。
如果你的笔记是黄,模型却写成绿:直接否决。颜色是你的判断,不是模型的乐观。
实例
填好的提示词
请把这些要点收成干系人更新邮件:进展、风险、请求。最多 150 词。不要编造绿色指标。要点:[内测已内部上线 | 移动端有延期风险 | 需要周三设计评审]。
边界:不编造;未知标 [?];只出草稿。
完成检查:一分钟内能判过/不过。
像样的输出大概长这样
- 全文压在约一百五十字内,进展 / 风险 / 请求三段分清
- 不编造绿色指标或「一切顺利」
- 移动端延期风险和周三设计评审请求都写在请求段
- 内测已在内部上线,不夸成对外正式大捷
- 读起来像能直接发出的干系人邮件
- 一分钟内能对照清单判过/不过;缺项标红再继续
红旗
先写感受再补事实,干系人找不到决策点;先事实再风险。
保存这张卡
记下:编辑轮或译→本地化里管用的一句改法。标上日期。
开始前
先写清读者与事实清单,再让模型动笔。
本周怎么用
本周写一份真更新:进展/风险/需要各三行;感受句一律移到末段或删。
常见坑
| 坑 | 改法 |
|---|---|
| 空开场感谢信 | 开头状态色 + 一句原因 |
| 编绿色指标 | 只用你提供的进展/风险 |
| 请求无截止 | 每个请求有对象/决定/截止 |
现在就练一次
今晚就练:写一份真更新:进展/风险/需要各三行;感受句移到末段或删。
闭环
写下一行:下次先写感受再补事实,我拒绝发出。
一句话带走
进展/风险/需要各三行;感受句移末段或删。
下一步
先提纲后成文 · 主动干掉 AI 套话 · 一周办公例程
← Back to home · Learn · Tools · Compare · Start