retention-engagement

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Retention & Engagement

留存与参与度

Scope

适用范围

Covers
  • Diagnosing retention + engagement (cohorts/curves, frequency, segments, drop-offs)
  • Identifying the activation / “aha moment” and reducing time-to-value
  • Designing habit + re-engagement interventions (daily return, reminders, content loops)
  • Creating accruing value and ethical switching costs (“mounting loss”)
  • Turning insights into a prioritized experiment + measurement plan
When to use
  • “Improve retention / reduce churn”
  • “Increase engagement / DAU/WAU”
  • “Define our activation / aha moment”
  • “D1/D7 retention is low—fix onboarding and time-to-value”
  • “Create a retention experiment backlog and a 30/60/90 plan”
When NOT to use
  • You don’t have (or can’t assume) a stable value proposition / ICP (use
    problem-definition
    ).
  • You’re primarily deciding pricing/packaging/paywalls (this skill can add retention context but won’t replace pricing work).
  • You need acquisition loop design (use
    designing-growth-loops
    ).
  • You need to synthesize qualitative churn feedback before proposing experiments (use
    analyzing-user-feedback
    or interviews).
涵盖内容
  • 诊断留存与参与度(群组分析/留存曲线、使用频率、用户细分、流失节点)
  • 识别激活/“aha moment”并缩短价值实现时间
  • 设计习惯培养与再参与干预措施(每日回访、提醒、内容循环)
  • 创造“累积价值”和符合伦理的“转换成本”(“损失递增”)
  • 将洞察转化为优先级排序的实验与衡量计划
何时使用
  • “提升留存率/降低流失率”
  • “提高用户参与度/日活/周活”
  • “明确我们的激活/aha moment”
  • “D1/D7留存率偏低——优化引导流程与价值实现时间”
  • “创建留存实验待办清单及30/60/90天执行计划”
何时不使用
  • 你还没有(或无法确定)稳定的价值主张/ICP(请使用
    problem-definition
  • 你的核心需求是确定定价/包装/付费墙(该方法可补充留存相关背景,但无法替代定价工作)
  • 你需要设计获客循环(请使用
    designing-growth-loops
  • 你需要先分析定性流失反馈再提出实验方案(请使用
    analyzing-user-feedback
    或用户访谈)

Inputs

输入要求

Minimum required
  • Product + target user/ICP and 1–2 key segments
  • Current stage (pre-PMF / early PMF / growth / mature)
  • Best-available baseline metrics (even rough):
    • retention (D1/D7/D30 or weekly cohort), churn, engagement (DAU/WAU/MAU), activation rate, time-to-value
  • Onboarding flow summary (steps/screens + where users drop)
  • Constraints: timebox, engineering/design capacity, allowed channels (email/push/in-app), privacy/legal/brand limits
Missing-info strategy
  • Ask up to 5 questions from references/INTAKE.md, then proceed.
  • If metrics are missing, proceed with explicit assumptions and label confidence.
  • Do not request secrets or PII; prefer aggregated metrics and redacted funnels.
最低必要信息
  • 产品+目标用户/ICP以及1–2个关键细分群体
  • 当前产品阶段(产品市场契合前/早期产品市场契合/增长期/成熟期)
  • 现有基准指标(即使是粗略数据):
    • 留存率(D1/D7/D30或周群组)、流失率、参与度(日活/周活/月活)、激活率、价值实现时间
  • 引导流程摘要(步骤/页面+用户流失节点)
  • 约束条件:时间限制、工程/设计资源、允许使用的渠道(邮件/推送/应用内)、隐私/法律/品牌限制
缺失信息处理策略
  • references/INTAKE.md中最多提出5个问题,然后推进工作
  • 如果缺少指标,基于明确假设继续并标注置信度
  • 不得索要机密信息或个人身份信息(PII),优先使用聚合指标和脱敏转化漏斗

Outputs (deliverables)

输出成果(交付物)

Produce a Retention & Engagement Improvement Pack (Markdown in-chat; or as files if requested) containing:
  1. Context snapshot (goal, segments, constraints, timebox)
  2. Metric definitions + guardrails (how “retention” and “engagement” are measured)
  3. Retention + engagement diagnosis (cohorts/curves, segments, drop-offs, churn drivers)
  4. Activation / aha moment definition (candidate behaviors + threshold + validation plan)
  5. Lever hypotheses map (onboarding → habit → accruing value → re-engagement)
  6. Experiment backlog (prioritized; experiment cards with success metrics + guardrails)
  7. Measurement + instrumentation plan (events, dashboards, owners if known)
  8. 30/60/90 execution plan
  9. Risks / Open questions / Next steps (always included)
Templates and checklists:
  • references/TEMPLATES.md
  • references/WORKFLOW.md
  • references/CHECKLISTS.md
  • references/RUBRIC.md
生成留存与参与度改进包(聊天内以Markdown格式呈现;或按需生成文件),包含:
  1. 背景快照(目标、细分群体、约束条件、时间限制)
  2. 指标定义与防护规则(“留存率”和“参与度”的衡量标准)
  3. 留存与参与度诊断分析(群组/曲线、细分群体、流失节点、流失驱动因素)
  4. 激活/aha moment定义(候选行为+阈值+验证计划)
  5. 杠杆假设图谱(引导流程→习惯培养→累积价值→再参与)
  6. 实验待办清单(优先级排序;实验卡片包含成功指标+防护规则)
  7. 衡量与埋点计划(事件、仪表盘、已知负责人)
  8. 30/60/90天执行计划
  9. 风险/待解决问题/下一步行动(必须包含)
模板与清单:
  • references/TEMPLATES.md
  • references/WORKFLOW.md
  • references/CHECKLISTS.md
  • references/RUBRIC.md

Workflow (7 steps)

工作流程(7步)

1) Intake + goal framing

1) 需求收集与目标框架搭建

  • Inputs: User prompt; references/INTAKE.md.
  • Actions: Define the retention problem (segment, time horizon, metric) and the decision this work will drive (what will change). Confirm constraints (timebox, capacity, channels, privacy/brand).
  • Outputs: Context snapshot + metric definitions draft.
  • Checks: Goal is a sentence with a number and a date (e.g., “Improve paid D30 retention from 18%→24% by end of Q2”).
  • 输入: 用户需求;references/INTAKE.md
  • 行动: 明确留存问题(细分群体、时间范围、指标)及本次工作将推动的决策(具体改变内容)。确认约束条件(时间限制、资源、渠道、隐私/品牌要求)
  • 输出: 背景快照+指标定义草案
  • 检查项: 目标需包含数字与日期(例如:“截至Q2末将付费用户D30留存率从18%提升至24%”)

2) Data + instrumentation sanity check

2) 数据与埋点合理性检查

  • Inputs: Current tracking/events (or best guess), funnel steps, dashboards (if any).
  • Actions: List what you can/can’t measure today. Define the minimum event schema needed to learn (activation, engagement, churn). Identify 1–3 highest-impact instrumentation gaps.
  • Outputs: Instrumentation gap list + “minimum viable measurement” plan.
  • Checks: Every key metric in the goal has a data source or an explicit assumption.
  • 输入: 当前追踪事件(或最佳推测)、转化漏斗步骤、仪表盘(如有)
  • 行动: 列出当前可测量/不可测量的内容。定义学习所需的最小事件架构(激活、参与、流失)。找出1–3个影响最大的埋点缺口
  • 输出: 埋点缺口清单+“最小可行衡量”计划
  • 检查项: 目标中的每个关键指标都有数据源或明确假设

3) Diagnose: where retention fails (and why)

3) 诊断:留存失效的节点与原因

  • Inputs: Baseline metrics, cohorts/curves, funnel drop-offs, segments, any churn feedback.
  • Actions: Build a diagnosis across three failure modes:
    • Activation failure (users never reach value)
    • Engagement decay (users get value once, don’t build a habit)
    • Monetization churn (value exists, but price/packaging/friction drives churn) Segment results (at least 2 segments) and identify the largest “leak.”
  • Outputs: Retention + engagement diagnosis table + primary failure mode(s).
  • Checks: Diagnosis points to one primary lever to test first (onboarding vs habit vs value vs comms).
  • 输入: 基准指标、群组/留存曲线、转化漏斗流失节点、用户细分、流失反馈(如有)
  • 行动: 从三种失效模式维度构建诊断分析:
    • 激活失效(用户从未触达核心价值)
    • 参与度衰减(用户体验过一次价值,但未形成使用习惯)
    • 付费流失(核心价值存在,但定价/包装/摩擦导致流失) 按细分群体拆解结果(至少2个细分群体)并找出最大“漏洞”
  • 输出: 留存与参与度诊断表+主要失效模式
  • 检查项: 诊断结果指向首个需测试的核心杠杆(引导流程vs习惯培养vs价值vs沟通)

4) Define the activation / “aha moment” (data-backed)

4) 数据驱动的激活/“aha moment”定义

  • Inputs: Candidate value behaviors + journey; usage events; retention outcome definition.
  • Actions: Propose 3–5 candidate “aha” behaviors, then define an activation threshold (e.g., “uses X feature twice within 7 days” or “invites 2 teammates + uses 2 key features within 14 days”). Document how you’ll validate (correlation with D30/D60 retention; holdout if possible).
  • Outputs: Activation/aha moment spec + validation plan + tracking requirements.
  • Checks: The activation definition is behavioral and measurable (not a survey response or opinion).
  • 输入: 候选价值行为+用户旅程;使用事件;留存结果定义
  • 行动: 提出3–5个候选“aha”行为,然后定义激活阈值(例如:“7天内使用X功能2次”或“14天内邀请2名同事+使用2个核心功能”)。记录验证方式(与D30/D60留存率的相关性;如有可能使用对照组)
  • 输出: 激活/aha moment规范+验证计划+追踪要求
  • 检查项: 激活定义基于行为且可测量(而非调查反馈或主观意见)

5) Generate lever hypotheses (convert insights → rules)

5) 生成杠杆假设(将洞察转化为规则)

  • Inputs: Diagnosis + activation spec; constraints.
  • Actions: Create a lever map with hypotheses tied to failure modes:
    • Onboarding/time-to-value: get users to aha faster and more reliably
    • Habit/daily return: design cues, routines, rewards; reduce friction to “come back tomorrow”
    • Accruing value + mounting loss (ethical): personalization, progress/history, saved work, identity/data repository
    • Re-engagement: lifecycle messaging, winback, content reminders, in-product nudges Convert each hypothesis into a rule + check (see references/SOURCE_SUMMARY.md).
  • Outputs: Lever hypotheses map + candidate interventions.
  • Checks: Every hypothesis ties to (a) a failure mode, and (b) a measurable leading indicator.
  • 输入: 诊断分析+激活规范;约束条件
  • 行动: 基于失效模式创建带假设的杠杆图谱:
    • 引导流程/价值实现时间: 更快、更稳定地让用户触达aha moment
    • 习惯培养/每日回访: 设计提示、常规流程、奖励;降低“次日回访”的摩擦
    • 累积价值+符合伦理的损失递增: 个性化、进度/历史记录、已保存工作、身份/数据存储库
    • 再参与: 生命周期消息推送、赢回策略、内容提醒、应用内提示 将每个假设转化为规则+检查项(参考references/SOURCE_SUMMARY.md
  • 输出: 杠杆假设图谱+候选干预措施
  • 检查项: 每个假设都关联(a)一种失效模式,以及(b)一个可测量的前置指标

6) Design + prioritize experiments (with measurement)

6) 设计并优先级排序实验(含衡量方案)

  • Inputs: Hypotheses; measurement plan; capacity.
  • Actions: Turn top hypotheses into experiment cards (1–2 weeks each). Prioritize using a simple score (Impact × Confidence ÷ Effort). Define success metrics and guardrails; note required instrumentation and rollout/rollback.
  • Outputs: Prioritized experiment backlog + experiment cards + metric/guardrail spec.
  • Checks: Top 3 experiments are runnable with current constraints and have unambiguous “win/lose/learn” criteria.
  • 输入: 假设;衡量计划;资源情况
  • 行动: 将优先级最高的假设转化为实验卡片(每张1–2周)。使用简单评分(影响力×置信度÷投入成本)进行优先级排序。定义成功指标与防护规则;标注所需埋点及上线/回滚方案
  • 输出: 优先级排序的实验待办清单+实验卡片+指标/防护规则规范
  • 检查项: 排名前3的实验可在当前约束条件下执行,且有明确的“成功/失败/学习”判定标准

7) Build the 30/60/90 plan + quality gate

7) 构建30/60/90天计划+质量校验

  • Inputs: Draft pack; references/CHECKLISTS.md and references/RUBRIC.md.
  • Actions: Sequence work into a 30/60/90 plan (instrumentation, experiments, analysis cadence). Run the checklist and score the rubric. Always include Risks / Open questions / Next steps.
  • Outputs: Final Retention & Engagement Improvement Pack.
  • Checks: Next 2 weeks of work are unblocked; measurement is in place to learn.
  • 输入: 改进包草案;references/CHECKLISTS.mdreferences/RUBRIC.md
  • 行动: 将工作拆解为30/60/90天计划(埋点、实验、分析节奏)。执行清单检查并对照评分标准打分。必须包含风险/待解决问题/下一步行动
  • 输出: 最终版留存与参与度改进包
  • 检查项: 未来2周的工作无阻碍;已部署衡量方案以获取学习洞察

Quality gate (required)

质量校验(必填)

  • Use references/CHECKLISTS.md and references/RUBRIC.md.
  • Always include: Risks, Open questions, Next steps.
  • 使用references/CHECKLISTS.mdreferences/RUBRIC.md
  • 必须包含:风险待解决问题下一步行动

Examples

示例

Example 1 (B2C subscription, churn reduction):
“Use
retention-engagement
. Product: meditation app. Segment: paid subscribers. Baseline: D30 paid retention 22%, churn spikes after week 2. Constraint: 4-week sprint, no major redesign. Output: a Retention & Engagement Improvement Pack with an activation/aha definition, a diagnosis, and a prioritized experiment backlog + 30/60/90 plan.”
Example 2 (B2B SaaS, activation + habit):
“New users activate but don’t return weekly. Define our aha moment, identify the biggest engagement decay point, and propose 5 experiments (in-product + email) with success metrics and guardrails.”
Boundary example (upstream problem):
“Write a brand new value prop and pick an ICP for our product.”
Response: that’s upstream strategy/problem definition; use
problem-definition
(and optionally PMF measurement) before retention optimization.
示例1(B2C订阅,降低流失率):
“使用
retention-engagement
。产品:冥想应用。细分群体:付费订阅用户。基准数据:付费用户D30留存率22%,第2周后流失率骤升。约束条件:4周迭代周期,不允许重大重新设计。输出:包含激活/aha moment定义、诊断分析、优先级排序的实验待办清单+30/60/90天计划的留存与参与度改进包。”
示例2(B2B SaaS,激活+习惯培养):
“新用户完成激活但未每周回访。明确我们的aha moment,找出参与度衰减的最大节点,并提出5个实验方案(应用内+邮件)及成功指标与防护规则。”
边界示例(上游问题):
“为我们的产品撰写全新价值主张并确定ICP。”
回应:这属于上游战略/问题定义范畴;在优化留存前,请使用
problem-definition
(可搭配产品市场契合度衡量)。