mcpField
C
conversation-pause-intelligence

conversation-pause-intelligence

github.com/louisblythe营销与增长190405 次安装40 周前更新
代码仓库

关于 conversation-pause-intelligence

面向销售机器人开发者的沉默判断与响应设计技能,帮助结合互动记录、停顿时长和对话阶段推断潜在客户的状态。它提供等待、重新联系和跟进的策略,并考虑渠道与外部情境。

主要功能

  1. 01区分思考、流失与中性停顿
  2. 02结合互动与停顿时长评估状态
  3. 03按对话阶段制定响应策略
  4. 04参考渠道与历史回复模式
  5. 05规划分阶段跟进顺序

适用场景

  1. 01判断演示后无回复是否需要等待
  2. 02依据互动程度选择重新联系方式
  3. 03结合邮件与短信信号安排跟进
  4. 04按潜在客户的回复习惯评估异常沉默

安装

复制下面的命令,在终端中运行。

$npx skills add https://github.com/louisblythe/sales-skills --skill conversation-pause-intelligence

安装后在 Claude Code 中输入/conversation-pause-intelligence确认已加载。

详细介绍

对话停顿智能

你是构建销售机器人停顿智能系统的专家。你的目标是帮助开发者创建能够恰当解读潜在客户沉默并做出回应的机器人。

理解沉默

潜在客户沉默的原因

积极原因(正在考虑)

  • 处理信息
  • 内部讨论
  • 查看你发送的材料
  • 寻求批准
  • 比较选项
  • 忙于其他优先事项

消极原因(不再参与)

  • 失去兴趣
  • 选择了竞争对手
  • 通过其他方式解决了问题
  • 不再是优先事项
  • 为了避免说“不”而不再回复
  • 发现了你所提供方案中的致命问题

中性原因

  • 休假或不在办公室
  • 工作变动或过渡
  • 公司范围内的活动
  • 单纯忘记了
  • 正在等待外部因素

解读停顿信号

情境指标

可能在考虑(等待)

Pre-pause behavior:
- High engagement in conversation
- Asked detailed questions
- Requested materials/info
- Mentioned sharing with team
- Set expectation ("I'll review and get back")

Pause duration: 24-72 hours

可能已不再参与(重新互动)

Pre-pause behavior:
- Short, declining engagement
- Stopped asking questions
- Objections not resolved
- Vague about next steps
- Mentioned competitor

Pause duration: 5+ days
互动度评分衰减
Calculate engagement health:

Last interaction quality:
- Deep conversation: +50 points
- Brief exchange: +20 points
- One-word response: +5 points

Time decay:
- Day 1-2: Full score
- Day 3-4: -20%
- Day 5-7: -40%
- Day 8-14: -60%
- Day 15+: -80%

Engagement signals:
- Email opens: +5/open (max 2/day)
- Link clicks: +15/click
- Website visit: +25/visit

Score < 30 = Likely disengaged
Score 30-60 = Monitor closely
Score > 60 = Likely still thinking

应对策略

等待策略
When signs point to "thinking":

Day 1-2: Do nothing (respect their time)

Day 3-4: Light touch if appropriate
"Hey [Name], just floating to the top of your inbox.
No rush—let me know if any questions came up."

Day 5-7: More direct follow-up
"Hi [Name], wanted to check in on [topic we discussed].
Still relevant, or has something changed on your end?"
重新互动策略
When signs point to "disengaged":

Pattern 1: Add value
"[Name], came across this [article/insight] about [their
industry/problem]. Thought it might be useful—no reply
needed if not!"

Pattern 2: Direct acknowledgment
"Hi [Name], I haven't heard back and want to be respectful
of your time. Should I check back in a few weeks, or is
this not a priority right now?"

Pattern 3: The breakup
"[Name], it seems like the timing isn't right, which is
totally fine. I'll stop following up, but my info is
here if things change. Best of luck with [their goal]!"
跟进询问
When unclear:

"Hi [Name], checking in on our conversation from last week.
I know things get busy—are you still looking to move forward
with [project], or has something changed?

Either way, happy to help when the time is right."

停顿时长指南

SMS 对话
Expected response times:
- Hot lead: Minutes to hours
- Warm lead: Hours to 1 day
- Cold lead: 1-3 days

Action thresholds:
- 2 hours: Normal pause
- 24 hours: First follow-up
- 48 hours: Second follow-up
- 72 hours: Try alternate channel
- 7 days: Cool-down period
- 14 days: Re-engagement sequence
电子邮件对话
Expected response times:
- Engaged prospect: 24-48 hours
- Semi-engaged: 2-4 days
- Low engagement: 1-2 weeks

Action thresholds:
- 48 hours: Normal pause
- 4-5 days: Follow-up
- 7-10 days: Second follow-up
- 14+ days: Re-engagement or close out
多渠道解读
Combine signals:

Email sent → No response (2 days)
+ SMS sent → No response (1 day)
+ Website visit detected

Interpretation: Still engaged, reviewing
Action: Wait, they're doing research

vs.

Email sent → No response (5 days)
+ SMS sent → No response (3 days)
+ No website activity

Interpretation: Likely disengaged
Action: Direct check-in or breakup

高级停顿分析

模式识别
Historical behavior:
- How long do they usually take to respond?
- What time of day do they typically reply?
- Which channel do they prefer?

This prospect usually responds:
- Within 4 hours → 24+ hour pause is unusual
- Within 2-3 days → Same pause is normal
外部情境
Factor in:
- Time of year (holidays, fiscal year end)
- Day of week (Friday afternoons slow)
- Company news (funding round, layoffs)
- Industry events (conferences, seasonality)
- Their calendar (if you have visibility)
对话阶段
Pause interpretation varies by stage:

Early discovery:
- Silence = Low priority, may return later
- Action: Low-touch nurture

Post-demo:
- Silence = Evaluating, discussing internally
- Action: Provide supporting materials

Negotiation:
- Silence = Getting approval or cold feet
- Action: Direct check-in on concerns

Closing:
- Silence = Red flag, something's wrong
- Action: Immediate outreach to understand

实现

状态跟踪
conversation_state = {
  "last_message_at": "2024-01-15T10:30:00Z",
  "last_response_at": "2024-01-13T14:15:00Z",
  "pause_duration_hours": 68,
  "engagement_score": 45,
  "last_engagement_type": "email_open",
  "typical_response_time_hours": 8,
  "follow_ups_sent": 1,
  "stage": "post_demo",
  "interpretation": "monitoring",
  "next_action": "wait_24_hours"
}
决策逻辑
def assess_pause(conversation):
    pause_hours = hours_since_response(conversation)
    engagement = calculate_engagement_score(conversation)
    stage = conversation.stage

    if pause_hours < conversation.typical_response_time * 2:
        return "normal_pause", "wait"

    if engagement > 60 and pause_hours < 72:
        return "likely_thinking", "light_touch_48h"

    if engagement < 30 and pause_hours > 120:
        return "likely_disengaged", "breakup_sequence"

    if stage == "closing" and pause_hours > 48:
        return "concern", "direct_check_in"

    return "unclear", "value_add_check_in"
跟进序列
Pause-based follow-up ladder:

1. Value add (Day 3-4)
   "Thought this might help..."

2. Status check (Day 6-7)
   "Checking in on..."

3. Direct ask (Day 10-12)
   "Still interested, or...?"

4. Breakup (Day 14+)
   "Going to close this out..."

5. Reactivation (Day 30+)
   "Things change—wanted to reconnect..."

评论

支持 Markdown