event-speaker-cold-outreach

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Event Speaker Cold Outreach

活动演讲者冷外联

You write the invitation and run it to an answer. Your input is the packet

samber/dev-event-organizer-skills@event-speaker-sourcing

hands you - ranked candidates, the evidence-backed "why" behind each name, a Hot/Warm/Cold/Skip score, availability and travel constraints, the honest budget line, the contact route, and the invited-review lane. Your output is a confirmed speaker with terms agreed in writing, or a recorded no.

This skill's boundaries:

Not sourcing or qualifying candidates: a name that arrives without its "why" goes back to sourcing, because the "why" is the message.

Not the open call design:

samber/dev-event-organizer-skills@event-cfp-design

owns that.

Not the invited-proposal review:
```
samber/dev-event-organizer-skills@event-talk-selection
```
owns that lane, and the packet already states which one applies.
Not what happens after a yes: briefing, travel and on-site care belong to
```
samber/dev-event-organizer-skills@event-speaker-experience
```
.

你负责撰写邀请并推进到获得答复。你的输入是

samber/dev-event-organizer-skills@event-speaker-sourcing

交给你的数据包——排名后的候选人、每个名字背后有证据支持的“原因”、Hot/Warm/Cold/Skip 评分、可用性与差旅限制、坦诚的预算额度、联系渠道，以及受邀提案评审通道。你的输出是已确认且条款书面约定的演讲者，或一条记录在案的拒绝。

本 skill 的边界：

不是寻找或筛选候选人：一个没有附带“原因”的名字应退回寻找环节，因为“原因”本身就是信息。

不负责公开征稿设计：

samber/dev-event-organizer-skills@event-cfp-design

负责该项。

不负责受邀提案评审：
```
samber/dev-event-organizer-skills@event-talk-selection
```
负责该环节，数据包已说明适用哪一项。
不负责同意之后的事务：简报、差旅和现场接待属于
```
samber/dev-event-organizer-skills@event-speaker-experience
```
。

Interview

访谈

Ask one question at a time, multiple-choice where possible. Questions 4-6 exist because the menus below diverge sharply on time-to-effect, durability and effort - those defaults cannot be picked for the user.

What does the sourcing packet contain for this slot: ranked candidates, the slot brief they were ranked against, each one's "why" and its evidence, the Hot/Warm/Cold/Skip score, availability and travel constraints, the honest budget line, the published contact route, and the invited-review lane? Anything missing goes back to sourcing rather than getting invented in the message.
Who owns the event: community conference or vendor/user conference? This changes what the invitation can honestly offer.
Which slot is this, and how many candidates are queued behind it? Decide now whether you approach them one at a time or in parallel - see failure modes, because two yeses for one slot are a worse problem than one no.
By what date do you need an answer, and what is the last date a yes is still usable once travel is counted? A hard date promotes the fast rungs and deletes the long cadence outright.
One-off edition, or a compounding annual program? A compounding mandate promotes the expensive personalization rungs: a candidate who declines this year with the relationship intact is next year's first call.
What is the effort ceiling: who actually writes these invitations, how many hours, and how much of your own name you are willing to spend on referral asks and personal introductions?
What can the budget honestly cover, phrased as a sentence you are willing to put in the first email - nothing beyond a ticket, travel, lodging, a fee? If it is unconfirmed, "unconfirmed" is what goes in the email.
What is negotiable if the answer is "yes, but": date, slot length, format, remote delivery, travel dates, a companion ticket, recording terms? Fix that list before the first message, never during the negotiation.
Who has authority to agree a fee or a condition, and is that person you? If not, name them and the turnaround, because a candidate waiting on your internal approval is a candidate cooling.

一次只问一个问题，尽可能用多选题。问题 4-6 存在的理由是：下面的选项在见效时间、持久性和投入上差异很大——这些默认值不能替用户决定。

这个档位的数据包包含哪些内容：排名后的候选人、他们据以排名的档位简报、每人的“原因”及其证据、Hot/Warm/Cold/Skip 评分、可用性与差旅限制、坦诚的预算额度、公开的联系渠道，以及受邀评审通道？任何缺失项都应回到寻找环节，而不是在消息中编造。
活动主办方是谁：社区会议还是厂商/用户会议？这会改变邀请函能诚实提供什么。
这是哪个档位，后面还排着多少候选人？现在就要决定是逐个联系还是并行联系——参见失败模式，因为一个档位出现两个“同意”比一个“拒绝”更麻烦。
你需要在什么日期前得到答复？计入差旅后，最晚什么日期之前“同意”仍可用？一个硬性日期会提升快速选项、直接删除长节奏。
是一次性活动，还是逐年累积的年度项目？累积型任务会提升高成本的个性化层级：今年婉拒且关系完好的候选人，是明年第一个要联系的人。
投入上限是什么：实际上由谁撰写这些邀请、多少小时，以及你愿意在转介请求和个人引荐上消耗多少自己的信誉？
预算能诚实地覆盖哪些内容，用你愿意写进第一封邮件的句子来表达——除门票、差旅、住宿、费用之外什么都没有？如果尚未确认，邮件中就写“未确认”。
如果答案是“可以，但是”，哪些可以协商：日期、档位时长、形式、远程演讲、差旅日期、同行票、录制条款？在第一次消息之前定下清单，绝不要在谈判过程中才定。
谁有权同意费用或条件，这个人是你吗？如果不是，说出人选和响应时间，因为等待你内部审批的候选人正在降温。

Community-run vs vendor/user conference

社区主办 vs 厂商/用户会议

What changes the invitation is who owns the event, the same replacement the rest of this collection argues:

Community-run - the invitation trades on the audience and the peer group, because that is genuinely all it has. Budget is usually thin, which makes the honesty rule load-bearing rather than polite: an invitation that stays vague about money is asking someone to block a weekend on a guess. A sponsor never buys a speaker slot, so a candidate arriving through a sponsor relationship gets invited on the same terms as anyone else or not at all.
Vendor/user conference - invitation is the main intake rather than the backup channel, budget is usually real, and the ask can carry a fee and full travel. The offer changes; nothing about the disclosure rules or the cadence does. Published organizer guidance is written for the community pole, so treat every vendor-pole adjustment here as judged rather than established practice.

Say which pole a recommendation assumes whenever they differ.

决定邀请函内容的是活动主办方是谁，这与本系列其余部分的主张一致：

社区主办——邀请函依靠的是观众和同行群体，因为这确实是它所拥有的全部。预算通常很少，这使得诚实规则变得至关重要而非礼貌：对金钱含糊其辞的邀请，是在要求别人凭猜测空出一个周末。赞助商从不购买演讲者档位，因此通过赞助商关系出现的候选人，要么以与其他任何人相同的条件被邀请，要么根本不被邀请。
厂商/用户会议——邀请是主要入口而非后备渠道，预算通常真实存在，邀约可以包含费用和全额差旅。提供的条件会变；披露规则或节奏则完全不变。公开发布的组织者指南是为社区一端撰写的，因此请将这里所有针对厂商一端所做的调整视为判断，而非既定做法。

当两者不同时，说明建议假定的是哪一端。

Workflow

工作流

Run the interview. Confirm the packet is complete (Q1) and fix the pole (Q2), the answer date (Q4), the honest budget line (Q7) and the negotiable list (Q8).
Check the travel gate before writing anything. If the candidate has to travel and your answer date sits inside the four-to-six-week floor, you are about to send an invitation that can only be declined. Say so in the message rather than letting them discover it, or move to a candidate who does not travel.
Pick the first-touch channel from the menu below, per candidate, against the published contact route in the packet.
Pick the personalization depth from the menu below, tiered by the candidate's Hot/Warm/Cold score - effort scales with fit.
Write the invitation against the disclosure list below, then cut it to length. Every message-craft constraint - the five-part structure, the word cap, the banned vocabulary, the subject-line rules - is in references/invitation-message-craft.md, each with its transposition labelled.
Run the invitation past a humanizer pass in your preferred humanizer skill before sending. A model-shaped invitation to a practitioner who reads model-shaped email all day is the fastest way to be ignored.
Send through the chosen channel, one candidate per slot per thread. Never let two candidates for the same slot see each other.
Follow up on the cadence menu below, respecting its cap. Every touch carries something the last one did not; honor an opt-out the moment it arrives, mid-sequence, not at the next scheduled touch.
Run the fee-and-conditions conversation off the honest-budget line when the answer is "yes, but" - see below, and the scripted shapes in references/cadence-and-fee-conversation.md.
Close the loop: confirm a yes in writing with every agreed term, or record a no with its reason and ask for a referral. Then hand off or move to the next candidate.

Show the user the drafted invitation before it goes out. An invitation cannot be unsent, and a candidate asked by mistake has already been asked.

If your harness has persistent memory, record per candidate:

the channel used
the angle each touch carried
the answer and its date
the reason behind any decline
any condition you agreed

That record is what makes next edition's outreach yours rather than this skill's defaults.

执行访谈。确认数据包完整（Q1），并确定端（Q2）、答复日期（Q4）、坦诚的预算额度（Q7）和可协商清单（Q8）。
在写任何内容之前检查差旅门槛。如果候选人必须出行，而你的答复日期落在四到六周的最短时间窗内，你即将发出的邀请只会被拒绝。在消息中直接说明，而不是让他们自己发现；或者转向不需要出行的候选人。
针对每位候选人，对照数据包中公开的联系渠道，从下面的选项中选择首次接触渠道。
根据候选人的 Hot/Warm/Cold 评分分层，从下面的选项中选择个性化深度——投入随匹配度变化。
按照下面的披露清单撰写邀请，然后删减到合适长度。每一条信息撰写约束——五部分结构、字数上限、禁用词汇、主题行规则——都在 references/invitation-message-craft.md 中，并标注了对应的转换。
发送前，用你偏好的拟人化 skill 对邀请做一次拟人化处理。发给整天阅读模型风格邮件的从业者一封模型风格的邀请，是最快被忽略的方式。
通过所选渠道发送，每个档位每条线程只对应一位候选人。绝不要让同一档位的两位候选人互相看到。
按照下面的节奏菜单跟进，并遵守其上限。每次接触都要带来上一次没有的信息；一旦收到退出请求，立即在序列中途尊重它，而不是等到下一次预定接触。
当答案是“可以，但是”时，基于坦诚预算额度展开费用与条件对话——参见下文及 references/cadence-and-fee-conversation.md 中的脚本模板。
闭环：如果同意，以书面形式确认所有商定条款；如果拒绝，记录原因并询问是否可以转介他人。然后交接或转向下一位候选人。

在发出之前向用户展示草拟的邀请。邀请一旦发出就无法撤回，而被错误邀请的候选人已经收到了邀请。

如果你的运行环境有持久化记忆，请为每位候选人记录：

使用的渠道
每次接触携带的角度
答复及其日期
任何拒绝背后的原因
你同意的任何条件

这份记录让下一届的外联成为你的成果，而非本 skill 的默认行为。

What the first invitation must state

首次邀请必须说明的内容

The first message is incomplete without every one of these.

Event basics - size, location, dates. A speaker cannot evaluate an invitation without knowing when and where.
How you found them, in the first line - their talk at X, their post on Y, a colleague's referral. This is simultaneously the strongest personalization available and the consent-respecting move, which is why it is one instruction and not two: a directory listing or a public profile is consent to be found, never consent to be pitched.
The specific angle, drawn from the packet's "why" - the candidate's own recent work, not a compliment about their reputation.
Why this slot, this audience, this stage - the differentiator aimed at a speaker's incentive (a peer audience, a room worth being seen in, a topic they have been trying to get a hearing for), not at the organizer's need.
Honest terms, up front - what the budget covers, including when the answer is "nothing beyond a ticket" or "unconfirmed", plus the confirmation deadline. Silence on money reads worse to a speaker than to a sponsor - a sponsor expects a negotiation, a speaker is being asked to block a calendar.
A low-friction, specific ask - a proposed next step, never an open-ended "let me know."
No links, attachments, or images in the first email. Corporate phishing filters auto-trash mail carrying them, and a speaker's work address is exactly that kind of inbox. The CFP page, the program and your logo all wait for the second touch.

首次消息缺少以下任何一项都是不完整的。

活动基本信息——规模、地点、日期。演讲者不知道时间和地点就无法评估邀请。
你是如何找到他们的，放在第一行——他们在 X 的演讲、在 Y 的帖子、同事的推荐。这既是最强的个性化手段，也是尊重同意的行为，所以这是一条指令而非两条：目录列表或公开资料是同意被_找到_，绝不是同意被推销。
具体角度，来自数据包中的“原因”——候选人自己近期的工作，而不是对其声誉的恭维。
为什么是这个档位、这个观众、这个舞台——针对演讲者动机的差异化因素（同行观众、值得在其中亮相的会场、他们一直想获得倾听机会的话题），而不是组织者的需求。
坦诚的条款，事先说明——预算覆盖哪些内容，包括答案是“除门票外什么都没有”或“未确认”的情况，以及确认截止日期。对金钱保持沉默，对演讲者的观感比对赞助商更糟——赞助商期待的是谈判，而演讲者是被要求空出日历。
低摩擦、具体的请求——一个提议的下一步行动，绝不要开放式地“让我知道”。
**第一封邮件中不要有链接、附件或图片。**企业钓鱼过滤器会自动删除带有这些内容的邮件，而演讲者的工作邮箱正是那种收件箱。CFP 页面、议程和你的标志都等到第二次接触再说。

First-touch channel

首次接触渠道

Per candidate, the route the first message travels.

Every ranking below is a default, not a law. After the interview, re-rank all three menus against what you know about this organizer. Any of these can overturn a default rung:

a program chair the candidate already knows
a referrer willing to make the introduction
a confirmed travel budget
a conference you and the candidate are both attending next month

Ranking (default, not a law - Q3, Q4 and Q6 re-rank it):

effort (per candidate contacted: coordination, social capital spent, travel, your own hours):
```
in-person ask > referrer-relayed intro > direct email == published DM
```

value (a considered answer, not merely a reply):

in-person ask == referrer-relayed intro > direct email > published DM

compliance cost (review triggered, reversibility spent):

direct email == published DM > in-person ask > referrer-relayed intro

efficiency:

referrer-relayed intro > direct email > published DM > in-person ask

Effort tie: an email and a DM are the same act - one composed message to a route the person published. The channel changes the character limit, not the work.
Value tie: both top rungs arrive through a relationship rather than through an inbox, one carrying a third party's vouch and the other your physical presence, and neither has to survive the unread-mail filter that actually kills the two written rungs.
Compliance tie: the same act seen from the other side - unsolicited written contact to a named individual, with contact details retained and a message that cannot be unsent, is consent and e-marketing territory (GDPR, CAN-SPAM and kin) whichever pipe carries it.
Referrer-relayed introduction - the default wherever the path exists: the person who referred the candidate, or a mutual contact, forwards the ask or makes the introduction. Cheapest strong rung, because a vouch arrives attached and you contact nobody until the candidate answers. Exhaust it per candidate before anything colder, though it exists for only part of any shortlist.
Direct email to a published address - the default everywhere else, and the rung every message-craft rule in this skill was written for. Beats the DM on value because email is where a considered answer gets written, while a DM invites a fast reflexive one to a request that needs a calendar check and often an employer's sign-off (a judged ordering, not a measured one).
Published DM or social contact route - the right rung when it is the only route the candidate published, which is common for candidates sourced by community scouting. Same effort and same compliance exposure as email, lower value, so efficiency puts it below.
In-person ask - the starved option: ask at an event you are both attending. It ties for the top of value and tops effort outright, so efficiency never picks it. Promotion conditions: a keynote or headline slot where a no costs the program most, a candidate who has already let a written approach go unanswered, or a conference you are both attending anyway before the answer date (Q4).
Deleted, not demoted: guessed or scraped addresses, and the candidate's community space as a contact route. A guessed work address is not a cheap last resort, it is out. So is soliciting inside a project forum, chat or issue tracker: many community codes of conduct treat recruiting and solicitation in their spaces as off-limits, the breach is public, and it cannot be walked back. Use the contact route the person published instead; parked at the bottom of a menu, both quietly return as scope when the shortlist runs thin.

针对每位候选人，首次消息所走的路线。

下面的每个排序都是默认值，不是铁律。访谈之后，根据你对该组织者的了解，重新排序所有三个选项。以下任何情况都可能推翻默认层级：

候选人已经认识的项目主席
愿意做引荐的推荐人
已确认的差旅预算
你和候选人下个月都会参加的会议

排序（默认值，不是铁律——Q3、Q4 和 Q6 会重新排序）：

effort（每位联系候选人的投入：协调、消耗的社会资本、差旅、你自己的时间）：
```
in-person ask > referrer-relayed intro > direct email == published DM
```

value（深思熟虑的答复，而不仅仅是回复）：

in-person ask == referrer-relayed intro > direct email > published DM

compliance cost（触发的审查、消耗的可逆性）：

direct email == published DM > in-person ask > referrer-relayed intro

efficiency：

referrer-relayed intro > direct email > published DM > in-person ask

**投入相同：**电子邮件和私信是同一行为——一封写好的消息发到对方公开的渠道。渠道改变的是字符限制，不是工作量。
**价值相同：**两个最高层级都通过关系而非收件箱到达，一个带有第三方的担保，另一个带有你的亲自到场，而且两者都不需要经过实际上会扼杀两个书面层级的未读邮件过滤器。
**合规成本相同：**从另一面看是同一个行为——未经请求地向某位具名个人发送书面联系，保留联系方式，且消息一旦发出无法撤回，无论通过什么渠道，都属于同意与电子营销范畴（GDPR、CAN-SPAM 及相关法规）。
推荐人转介——只要路径存在即为默认：推荐候选人的人或共同联系人转发请求或做引荐。最便宜的强层级，因为担保随之而来，而且在候选人答复前你不需要联系任何人。在采取更冷的方式之前，对每位候选人都先尝试这个渠道，尽管它只存在于部分名单中。
向公开地址发送直接邮件——其他情况下的默认方式，也是本 skill 中所有信息撰写规则所针对的层级。在价值上优于私信，因为深思熟虑的答复写在邮件里，而私信会诱发快速反射性的回复，而这一请求需要查看日历，往往还需要雇主批准（这是判断性的排序，而非实测结果）。
公开私信或社交联系渠道——当它是候选人公开的唯一渠道时的正确层级，这在通过社区发掘的候选人中很常见。与电子邮件相同的投入和相同的合规风险，但价值更低，所以效率将其排在下方。
当面请求——被冷落的选项：在你们都参加的活动上提出请求。它在价值上并列最高，而且投入最高，所以效率永远不会选择它。升级条件：主题演讲或头条档位，此时一个“不”对议程损失最大；候选人已经对书面联系置之不理；或者你们双方本来就会在答复日期（Q4）前参加同一会议。
**删除，而非降级：猜测或抓取的地址，以及将候选人的社区空间作为联系渠道。**猜测的工作邮箱不是便宜的最终手段，而是出局。在项目论坛、聊天或 Issue 追踪器中拉人同样如此：许多社区行为准则将在其空间中招募和拉人视为禁区，违规是公开的，且无法收回。改为使用对方公开的联系渠道；即使把它们放在选项底部，当名单变薄时，这两者都会悄悄作为活动范围回归。

Personalization depth

个性化深度

How much of the invitation is about this specific person. Ranking (default, not a law - Q5 and Q6 re-rank it):

effort (minutes per invitation, and whose minutes):

program-shaped ask > fresh-work personalization > sourced-"why" > slot-brief line

value (a reply that engages with the actual offer instead of declining politely):

program-shaped ask > fresh-work personalization > sourced-"why" > slot-brief line

compliance cost (review triggered, reversibility spent):

fresh-work personalization == program-shaped ask > sourced-"why" == slot-brief line

efficiency:

sourced-"why" > fresh-work personalization > slot-brief line > program-shaped ask

Dominance check: 4 rungs, 6 pairs, zero strict-dominance relations - clean only by construction, and by-construction is never a pass. Value and effort are the same list, so one mechanism blocks all six pairs: the deeper ask buys a better reply and costs strictly more minutes.

The compliance axis re-orders the menu and rescues nothing, because a dominance relation needs value ≥ and every cost ≤, and the effort half is already lost in every pair. The check catches no misordering; the efficiency line rests on the arguments below.

Both compliance ties are genuine and split the menu cleanly in half:

The two deep rungs go past the shortlist's recorded evidence into fresh research on a named individual, where the line between "read their public work" and "assembled a file on them" is yours to hold.
The two shallow rungs use only what sourcing already recorded and dated for exactly this purpose.

Efficiency inverts the value order at the top for one reason worth stating: the sourced-"why" is already paid for. Sourcing bought that evidence to qualify the candidate, and re-using it costs you one sentence.

Slot-brief line - the floor: name the specific slot, theme and audience, with nothing candidate-specific. Cheapest, and it sits below fresh-work on efficiency anyway, because cheap and efficient are different orderings - a generic invitation sent to someone you chose specifically throws away the qualification that got them onto the list.
Sourced-"why" personalization - the default for every candidate: open on the specific evidence the packet recorded, the talk or the post or the project, and connect it to the slot. Costs a sentence and returns most of what deeper research would.
Fresh-work personalization - read what they published this month, past the packet's evidence, and open on that instead. Promote every Hot candidate to this rung, and every candidate for a slot that concentrates the program's risk.
Program-shaped ask - the starved option: restructure the invitation around a talk only this person could give, naming the gap in the program their work fills, effectively co-designing the slot before they answer. It tops both value and effort, so efficiency never picks it. Promotion conditions: a keynote, a candidate who declined a previous edition, a compounding annual mandate (Q5) where the relationship outlives this slot, or a slot no shortlisted candidate quite fits.
Deleted, not demoted: the merge-field blast to the whole shortlist. One template, one mail-merge, every candidate at once. It is not the bottom rung of this menu; it is the thing that makes the entire shortlist worthless, and it is one slip away from the CC failure mode below.

邀请中有多少是关于这位具体的人。排序（默认值，不是铁律——Q5 和 Q6 会重新排序）：

effort（每封邀请的分钟数，以及是谁的分钟）：

program-shaped ask > fresh-work personalization > sourced-"why" > slot-brief line

value（与真实邀约互动的回复，而不是礼貌地婉拒）：

program-shaped ask > fresh-work personalization > sourced-"why" > slot-brief line

compliance cost（触发的审查、消耗的可逆性）：

fresh-work personalization == program-shaped ask > sourced-"why" == slot-brief line

efficiency：

sourced-"why" > fresh-work personalization > slot-brief line > program-shaped ask

支配关系检查：4 个层级、6 对组合、零严格支配关系——只有通过构造才干净，而通过构造从来不算通过。 价值和投入是同一个列表，因此一个机制阻断了所有六对：更深的请求换来更好的答复，但成本严格更高。

合规轴重新排序了选项但没有救回任何一对，因为支配关系需要价值 ≥ 并且所有成本 ≤，而投入这一半在每对中已经失败。检查没有发现错误排序；效率一行的依据来自下面的论证。

两个合规相同是真实的，并将选项干净地分成两半：

两个深层级超出名单中已记录的证据，进入对具名个人的新研究，此时“阅读他们的公开作品”和“整理了一份关于他们的档案”之间的界线由你来把握。
两个浅层级只使用寻找环节已经为此目的记录并注明日期的内容。

效率在最顶层颠倒了价值顺序，原因值得说明：来源“原因”已经付过费了。寻找环节购买该证据是为了筛选候选人，而重复使用它只花费你一句话。

档位简报行——底线：点名具体档位、主题和观众，不包含任何候选人特定内容。成本最低，而在效率上它本来就排在新鲜作品之下，因为廉价和高效是不同的排序——把一封通用邀请发给你特意选中的某人，等于浪费了让他们进入名单的那个筛选依据。
来源“原因”个性化——每位候选人的默认方式：以数据包记录的具体证据开头，即演讲、帖子或项目，并将其连接到档位。花费一句话，却能得到大部分深度研究会带来的效果。
新鲜作品个性化——阅读他们本月发布的内容，超越数据包中的证据，并以此开头。将所有 Hot 候选人提升到这一层级，并将所有承担议程集中风险的档位候选人提升到这一层级。
议程塑形请求——被冷落的选项：围绕只有这个人才能给出的演讲重构邀请，指出他们的工作填补了议程中的哪个空白，实际上是在他们答复之前共同设计档位。它在价值和投入上都最高，所以效率永远不会选择它。升级条件：主题演讲；候选人婉拒过上一届；累积型年度任务（Q5），此时关系比这个档位更长久；或者没有入围候选人真正适合的档位。
**删除，而非降级：向整个候选名单群发合并字段邮件。**一个模板、一次邮件合并、同时发给所有候选人。它不是这个选项的底层；它是让整个候选名单变得毫无价值的东西，而且离下面的抄送失败模式只有一步之遥。

Follow-up cadence

跟进节奏

After the first touch, how many times and how spaced. Ranking (default, not a law - Q4 and Q6 decide this menu almost alone):

effort (touches written, threads tracked, and how long the slot stays held open):
```
long-runway cadence > three-touch angle cadence > single ask
```
value (answers recovered from silence, including a no you can act on):
```
long-runway cadence > three-touch angle cadence > single ask
```

compliance cost (review triggered, reversibility spent):

long-runway cadence > three-touch angle cadence > single ask

efficiency:

three-touch angle cadence > single ask > long-runway cadence

Dominance check: 3 pairs, zero strict-dominance relations - clean only by construction, and that is not a pass. All three axes run in one order, so a single mechanism blocks every pair: the cadence that recovers more answers is strictly worse on both cost axes.

Compliance restates the effort order rather than re-ordering it, so it cannot rescue a pair either. The check verifies nothing here; argue with the ratios below.

The compliance axis needs no ties: it tracks touch count directly. Every additional unsolicited message to somebody who has not replied is more exposure, not less - silence is not consent to keep writing, and no touch can be recalled.

Single ask, then stop - send once and read silence as a no. Its whole cost is one message, and it frees the slot the same day for the next candidate.
Three-touch angle cadence - the default: Day 0, Day 3 only if you have new information, Day 7 with a genuinely fresh hook, then stop. Disciplined enough for a volunteer organizer and short enough that the slot does not rot behind it.
Long-runway cadence - the starved option: one to two weeks between touches, capped at three, tightening to three or four days as the answer date closes in. It tops value and effort together, so efficiency never picks it - yet it is the only cadence shaped for the candidate who has to get an employer's sign-off and arrange travel, which is most invited speakers.
- Promotion conditions: a candidate who needs internal approval, an international candidate inside the four-to-six-week travel window, or a slot with no viable next candidate behind it.
- Efficiency ranks it below even the single ask: its real cost is not the touches, it is the weeks a slot stays held open behind someone who may never answer, and an open slot is the scarcest thing in an invited program.
Deleted, not demoted: the bump. "Bumping this up", "did you see my email", the same pitch reformatted - deleted here rather than ranked. Never bump; always add an angle. In a community whose members talk to each other, a bumped invitation costs the event's name more than the slot is worth.

Rotate a genuinely new angle into every touch:

a confirmed co-speaker
a program detail that landed since
a specific audience question their work answers

The full angle table is in references/cadence-and-fee-conversation.md.

首次接触之后，跟进多少次、间隔多久。排序（默认值，不是铁律——Q4 和 Q6 几乎单独决定这个选项）：

effort（写出的接触次数、跟踪的线程数，以及档位保持开放的时间）：
```
long-runway cadence > three-touch angle cadence > single ask
```
value（从沉默中找回的答复，包括你可以行动的拒绝）：
```
long-runway cadence > three-touch angle cadence > single ask
```

compliance cost（触发的审查、消耗的可逆性）：

long-runway cadence > three-touch angle cadence > single ask

efficiency：

three-touch angle cadence > single ask > long-runway cadence

支配关系检查：3 对组合、零严格支配关系——只有通过构造才干净，而这不算通过。 所有三个轴都按同一顺序排列，因此一个机制阻断了每一对：能找回更多答复的节奏在两个成本轴上都严格更差。

合规只是复述了投入顺序而不是重新排序，所以它也无法救回任何一对。检查在这里没有验证任何东西；请用下面的比率来论证。

合规轴不需要相同项：它直接跟踪接触次数。每一条发给未回复者的额外未经请求的消息都是更多的风险暴露，而不是更少——沉默不等于同意继续写信，而且任何接触都无法撤回。

一次请求，然后停止——发送一次，将沉默视为拒绝。它的全部成本是一条消息，并在同一天释放档位给下一位候选人。
三次接触角度节奏——默认方式：第 0 天、第 3 天_仅当你有新信息时_、第 7 天带一个真正新鲜的角度，然后停止。对志愿组织者来说足够自律，而且足够短，档位不会在其后腐烂。
长跑道节奏——被冷落的选项：每次接触间隔一到两周，上限三次，随着答复日期临近缩短到三四天。它在价值和投入上同时最高，所以效率永远不会选择它——然而它是唯一为需要获得雇主批准并安排差旅的候选人设计的节奏，而大多数被邀演讲者都是如此。
- 升级条件：需要内部批准的候选人、处于四到六周差旅窗口内的国际候选人，或者后面没有可行替补候选人的档位。
- 效率将其排在甚至低于一次请求的位置：它的真实成本不是接触次数，而是档位在可能永远不会答复的人身后保持开放的数周时间，而空置档位是受邀制议程中最稀缺的资源。
删除，而非降级：催促式跟进。“把这个顶上来”“你看到我的邮件了吗”、重新排版后的同一套说辞——在这里被删除而非排名。永远不要简单催促；总是添加一个角度。在一个成员互相交流的社区里，一封催促式邀请对活动名声的损害超过了档位的价值。

在每次接触中轮换一个真正新鲜的角度：

已确认的共同演讲者
之后确定的议程细节
他们的工作所回答的某个具体观众问题

完整的角度表格见 references/cadence-and-fee-conversation.md。

The fee and conditions conversation

费用与条件对话

A sponsor negotiation has a rate card behind it. A speaker invitation has nothing equivalent - there is no published price, no tier, and no norm. What you have instead is the honest-budget line fixed at Q7, and it is fixed before the first message, stated in it, and not rediscovered under pressure.

When the candidate names a fee, you have no market rate to check it against, and inventing one is the single most tempting failure in this skill. Two named community events publish a flat per-speaker honorarium - !!Con's stable "$256 (USD)" and ffconf's "£500" - but both pay a fixed amount to every accepted CFP speaker alike, which is a different shape from a negotiated invited-speaker fee; treat them as calibration for what a small honorarium looks like, never as a rate to counter-offer with. Answer from your own budget and your own authority (Q9), and route any actual payment through whoever handles the event's money and contracts. Direct payment to speakers can bring complex issues around taxes, visas, and conflict of interest or employment agreements, so involve them rather than routing payment yourself.
When money is not available, trade conditions instead - the list fixed at Q8, never improvised. None of these is a convention of event practice: they are ordinary concessions with a real cost to you and real value to them, which is the only test a concession has to pass.
- travel dates that let them add a personal day
- a shorter or longer slot
- remote delivery
- a co-speaker seat
- a companion ticket
- recording and publication terms
- the time of day their session runs
- an introduction to someone in the audience they want to meet
Never promise what you cannot deliver - the pitch-killer that damages an event longest, because the candidate finds out after they have already said yes.
Escalate rather than stall. If the ask exceeds your authority, say who decides and by when. A candidate waiting on an unexplained silence is cooling, and the cadence menu has no rung for "waiting on us."

赞助商谈判背后有价目表。演讲者邀请没有与之相当的东西——没有公开价格、没有等级、没有规范。你拥有的是在 Q7 确定的坦诚预算额度，它是在_第一次消息之前_就确定、在_消息中_说明的，并且不会在压力下重新发现。

当候选人提出费用时，你没有市场价可参照，而编造一个市场价是本 skill 中最诱人的失败方式。两个知名的社区活动公布了固定的每位演讲者酬金——!!Con 稳定的“$256（美元）”和 ffconf 的“£500”——但两者都对每位被接受的 CFP 演讲者支付相同的固定金额，这与协商制的受邀演讲者费用形态不同；把它们视为小规模酬金的校准参考，绝不要作为还价的基准。从你自己的预算和自己的权限（Q9）出发回答，并通过负责活动资金和合同的人处理任何实际付款。直接向演讲者付款可能带来税务、签证、利益冲突或雇佣协议等复杂问题，因此要让他们参与，而不是自己安排付款。
当没有资金可用时，改用条件交换——使用在 Q8 确定的清单，绝不临时发挥。这些都不是活动惯例：它们是对你有实际成本、对他们有实际价值的普通让步，这是让步必须通过的唯一检验。
- 让他们能额外安排一天私人时间的差旅日期
- 更短或更长的档位
- 远程演讲
- 联合演讲者席位
- 同行票
- 录制与发布条款
- 他们环节在一天中的时段
- 引荐观众中他们想见的人
绝不要承诺你无法交付的东西——这是对活动伤害最久的推销杀手，因为候选人是在已经同意之后才发现的。
**升级而不是拖延。**如果请求超出你的权限，说明由谁决定以及何时决定。一个在不明沉默中等待的候选人正在降温，而节奏菜单中没有“等待我们”这一层级。

Closing the loop

闭环

On a yes, confirm in writing the same day, then hand the confirmed speaker to
```
samber/dev-event-organizer-skills@event-speaker-experience
```
:
- slot, date, time, length, format
- what the event pays and - explicitly - what it does not
- the travel and lodging terms
- the recording terms
- the review lane the packet stated (
```
samber/dev-event-organizer-skills@event-talk-selection
```
  needs that answer, not a guess)
On a yes-with-conditions, agree only from the Q8 list, then confirm exactly as above with the condition written into it.
On a no, record the reason. That reason is worth more than the slot: it feeds
```
samber/dev-event-organizer-skills@event-speaker-sourcing
```
's channel record and next edition's shortlist. Ask whether they would suggest someone else - a referral is the cheapest strong signal in the sourcing menu, and a declining candidate is unusually well placed to give one, but never re-approach the same person for the same slot.
On silence past the cadence cap, treat it as a no, move to the next candidate, and release the slot. A slot held open out of hope is the most expensive thing in this workflow.

如果同意，当天书面确认，然后将已确认的演讲者交给
```
samber/dev-event-organizer-skills@event-speaker-experience
```
：
- 档位、日期、时间、时长、形式
- 活动支付什么，以及明确说明不支付什么
- 差旅与住宿条款
- 录制条款
- 数据包中说明的评审通道（
```
samber/dev-event-organizer-skills@event-talk-selection
```
  需要这个答案，而不是猜测）
如果同意但有条件，只能从 Q8 清单中同意，然后按上述方式确认，并将条件写入其中。
如果拒绝，记录原因。这个原因比档位更有价值：它反馈给
```
samber/dev-event-organizer-skills@event-speaker-sourcing
```
的渠道记录和下一届的候选名单。询问他们是否会推荐其他人——转介是寻找菜单中最便宜的强信号，而拒绝的候选人特别适合给出一个，但绝不要为同一个档位再次联系同一个人。
超过节奏上限仍无回应，视为拒绝，转向下一位候选人，并释放档位。出于希望而一直空置的档位是这个工作流中最昂贵的东西。

Failure modes

失败模式

CCing two candidates for the same slot. Both see they were second choice, or worse, both see each other. One candidate per thread, always.
Running parallel threads for one slot without deciding what happens on two yeses. Sequential asks cost time; parallel asks cost a retraction. Pick at Q3, and if you go parallel, know in advance which second yes you can honestly place elsewhere.
Treating a directory listing or a public profile as consent to be pitched. It is consent to be found. Use the route they published, say how you found them, and keep the ask tied to a real slot.
Hiding the budget until late. A candidate who blocks travel time and then learns nothing is covered will decline, tell people, and not answer next year. The budget line goes in the first message even when the answer is "nothing" or "unconfirmed".
Bumping without new information. Deleted from the cadence menu for a reason; it converts a pending answer into a remembered annoyance.
Inviting without the sourcing "why". An invitation that cannot answer "why me, why this event, why now" is a form letter to someone you specifically chose. Send it back to sourcing rather than writing around the gap.
Links, attachments or images in the first email. Auto-trashed unread by corporate filters. Everything visual waits for touch two.
Inventing a speaker fee, a market rate or an "industry standard". Saying "I don't have a benchmark, here is our budget" is the honest and stronger move.
Inviting inside the travel window without saying so. The four-to-six-week travel floor is the earliest a travelling candidate could plausibly commit; an invitation arriving under it is functionally a request to decline unless you name the constraint yourself.
Promising a slot, a format or a fee you do not control. See Q9 - check your authority before the sentence, not after.

**将同一档位的两位候选人抄送在同一封邮件中。**两人都会看到自己是第二选择，更糟的是会互相看到。每条线程永远只有一位候选人。
**为同一个档位并行推进多个线程，却没有决定出现两个“同意”时怎么办。**顺序请求消耗时间；并行请求消耗一次撤回。在 Q3 时做出决定，如果选择并行，提前知道第二个“同意”可以诚实地安排在哪个别处。
**将目录列表或公开资料视为同意被推销。**那是同意被找到。使用他们公开的渠道，说明你是如何找到他们的，并让请求与真实档位绑定。
**迟迟不透露预算。**候选人空出差旅时间后才发现什么都不能覆盖，会拒绝、告诉别人，并且明年不再回复。预算额度要放在第一条消息中，即使答案是“什么都没有”或“未确认”。
**没有新信息就催促。**从节奏菜单中删除是有原因的；它会把一个待答复变成一段被记住的烦扰。
**没有来源“原因”就发出邀请。**一封无法回答“为什么是我、为什么是这个活动、为什么是现在”的邀请，就是发给一个你特意选中的人的形式信件。把它退回寻找环节，而不是绕过这个缺口硬写。
**第一封邮件中的链接、附件或图片。**会被企业过滤器不读直接删除。所有视觉内容都等到第二次接触。
**编造演讲者费用、市场价或“行业标准”。**说“我没有基准，这是我们的预算”是诚实且更有力的做法。
**在差旅时间窗内发出邀请却不说明。**四到六周的差旅最低期限是出行候选人可能承诺的最早时间；在此期限之内到达的邀请实际上是在要求对方拒绝，除非你自己点明这一限制。
**承诺一个你不掌控的档位、形式或费用。**参见 Q9——在说出那句话之前检查你的权限，而不是事后。

Measurement

衡量

Acceptance rate, reply rate and time-to-answer are all self-set here: quote your own numbers rather than importing a sales figure, and expect them to mean something only across editions.

Invitation-to-acceptance rate (self-set) - accepted over sent, per slot type. Edition two turns this into the only speaker-outreach benchmark you will ever have that is actually about your event.
Time-to-answer by channel (self-set) - days from first touch to any answer, split by channel. This is what re-ranks the first-touch menu for you.
Decline-reason capture - the one metric with a real threshold: every decline carries a recorded reason, or the next edition's sourcing repeats the mistake that produced it.
Slot-holding time (self-set) - days a slot spends behind an unanswered candidate. A number nobody tracks, and the one that quietly decides whether a program fills.

Pick two, write down what each would change next edition, and record both before the event.

接受率、回复率和答复时间在这里都是自行设定的：引用你自己的数字，而不是引入销售数据，并预期它们只有跨届比较才有意义。

邀请到接受率（自行设定）——按档位类型统计的接受数除以发送数。第二届会把它变成你拥有的唯一真正针对你自己活动的演讲者外联基准。
按渠道的答复时间（自行设定）——从首次接触到任何答复的天数，按渠道拆分。这是为你重新排序首次接触菜单的依据。
拒绝原因捕获——唯一有真实阈值的指标：每次拒绝都带有一条记录在案的原因，否则下一届的寻找环节会重复产生该拒绝的错误。
档位占用时间（自行设定）——一个档位在未答复候选人身后度过的天数。这是一个没人跟踪的数字，却悄悄决定议程是否能填满。

选择其中两个，写下每个会在下一届改变什么，并在活动前把两者都记录下来。

Invocation examples

调用示例

"Write the invitation to the keynote candidate at the top of our shortlist - we can cover travel but no fee."
"Our first invitation went out nine days ago with no answer. What do we send now, if anything?"
"A speaker we invited came back asking for €2,000. We have no budget line for fees - how do I answer?"
"We have three candidates for one slot and six weeks until the program is announced. How do I approach them?"

Expected output: a drafted invitation carrying the packet's "why", the honest budget line and the answer deadline, plus a per-candidate plan covering:

the channel and cadence, with the angle each follow-up will carry
the negotiable list fixed in advance
the written confirmation or the recorded decline that closes the loop

“为我们名单顶部的主题演讲候选人撰写邀请——我们可以承担差旅，但没有费用。”
“我们的第一封邀请九天前发出，没有回复。现在该发什么，或者什么都不发？”
“我们邀请的一位演讲者回复要求 2,000 欧元。我们没有费用预算——我该怎么回答？”
“我们一个档位有三位候选人，距离议程公布还有六周。我该怎么联系他们？”

预期输出：一封携带数据包“原因”、坦诚预算额度和答复截止日期的草拟邀请，外加一份每位候选人的计划，包含：

渠道和节奏，以及每次跟进将携带的角度
事先确定的可协商清单
闭环的书面确认或记录在案的拒绝

References

参考

references/invitation-message-craft.md - the five-part message structure, the length cap and banned vocabulary, subject-line rules, the full disclosure checklist, and a worked positive and negative invitation.
references/cadence-and-fee-conversation.md - the three cadences touch by touch with an angle table, the close-out message, the fee-and-conditions conversation scripts against the honest-budget line, the concession list with its costs, and the yes/no confirmation shapes.