layers-observed-behaviour

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/layers-observed-behaviour

Assumes
/layers-intro
has been loaded for framework context.

The observed behaviour layer is the closest we can get to reality. It is raw material — what users actually do, not what we think they do or wish they would do. Everything above it is interpretation; this layer is the source.

Decisions this layer needs to make:

What specific questions do we most need to answer about our users?
What evidence already exists, and how reliable is it?
How do we gather what's missing?
What patterns emerge — and what can we claim with confidence vs. what remains assumption?

Methods:

Method	When
JTBD interviews	Understanding triggers, motivations, and anxieties. Interview about a real past experience, not hypotheticals.
Contextual inquiry / observation	When what users say differs from what they do. Watching real work reveals tacit behaviour.
Diary studies	When the behaviour is distributed over time or infrequent — users self-report as events occur.
Support ticket / review analysis	Existing product with accumulated signal. Good for identifying pain points at scale without recruiting.
Analytics review	What users do (not why). Complements qualitative; doesn't replace it.
Usability observation	Where do people struggle or succeed with an existing product?

Quality signals — what good looks like:

Observations are specific and close to raw data, not summarised into conclusions
Job stories are grounded in something seen or heard, not inferred from team beliefs
Confidence levels are marked: observed / inferred / assumed
Research gaps are named explicitly, not papered over
Workarounds are flagged — a need real enough to motivate improvisation is a strong signal

This skill works in two modes. Detect which applies and state it clearly:

Plan mode — no research yet; help design a study
Synthesise mode — existing research material; help make sense of it

If they have partial research, start in Synthesise mode with what exists, then shift to Plan mode to identify gaps.

假设已加载
/layers-intro
以获取框架背景信息。

行为观察层是我们最接近真实情况的层级。它是原始素材——即用户实际的行为，而非我们认为他们会做或希望他们做的事。该层级之上的所有内容都是解读，而这一层才是源头。

该层级需要做出的决策：

关于用户，我们最需要解答哪些具体问题？
已有哪些可用证据，其可信度如何？
我们如何收集缺失的信息？
会浮现出哪些模式——哪些结论我们可以笃定，哪些仍属于假设？

方法：

方法	适用场景
JTBD访谈	用于理解触发因素、动机和焦虑情绪。访谈需围绕真实的过往经历展开，而非假设场景。
情境调研/观察	当用户的表述与实际行为不符时。观察真实工作场景可发现隐性行为。
日记研究	当行为随时间分散出现或发生频率较低时——由用户在事件发生时自行记录。
支持工单/评论分析	适用于已有累积数据信号的产品。无需招募用户即可大规模识别痛点。
数据分析复盘	了解用户做了什么（而非原因）。是定性研究的补充，而非替代。
可用性观察	了解用户在使用现有产品时的难点与顺畅点。

质量信号——优秀的标准：

观察结果具体且贴近原始数据，未被提炼为结论
Job Stories基于实际观察或听闻的内容，而非团队的主观推断
标注置信度层级：观察所得/推断所得/假设所得
明确指出研究缺口，而非刻意掩盖
标记用户的变通方案——足以促使用户自行摸索的需求是强烈信号

该技能有两种模式。需判断适用模式并明确说明：

规划模式——尚未开展研究；协助设计研究方案
整合模式——已有研究素材；协助梳理分析

若已有部分研究成果，先基于现有内容启动整合模式，再切换至规划模式以识别研究缺口。

Guided session

引导式会话

Tell me what you're working on — what question you most need to answer about your users — or say "plan" or "synthesise" to start a guided session.

Ask: "Where should I capture the work from this session?" (see

/layers-intro

for options)

Ask: "Do you have existing research — interviews, session recordings, support tickets, analytics — or do we need to plan research from scratch?" State the mode clearly once you know.

告诉我你的工作内容——你最需要解答的关于用户的问题是什么——或者说"plan"（规划）或"synthesise"（整合）来启动引导式会话。

提问："我应该在哪里记录本次会话的成果？"（查看

/layers-intro

了解可选方案）

提问：*"你是否已有研究成果——如访谈记录、会话录屏、支持工单、数据分析——还是我们需要从零开始规划研究？"*确定后明确说明适用模式。

Plan mode

规划模式

Phase 1 — Define the learning goal

Help the designer articulate what they most need to understand. Push past vague goals:

Not "understand users better" but "understand what triggers someone to refer a friend, and what makes them hesitate"

Write out 2–3 specific research questions.

Phase 2 — Identify participants

Who has the behaviour we need to observe? Push on assumptions:

Existing users, potential users, or churned users?
Is the behaviour something they do currently, or something they'd do in a new situation?
How many participants for meaningful signal? (Qualitative interviews: 6–10 usually reaches saturation)

Phase 3 — Design the study

Choose the right method based on the learning goal. If running JTBD interviews, generate an interview guide:

Opening: "Tell me about the last time you [relevant behaviour]. Walk me through what was happening at that point."

Timeline: "What triggered that? What did you try first? What made you keep going / give up / switch?"

Motivations: "What were you hoping would be different after doing this? What were you worried about? What almost stopped you?"

Closing: "Is there anything I haven't asked that would help me understand this better?"

Remind the designer: listen for nouns (candidate domain objects) and the language users use naturally — this feeds the domain layer.

Phase 4 — Plan synthesis

Before going to research: agree on how findings will be captured and organised. One observation per note, tagged with the research question it speaks to. Raw quotes over summaries.

阶段1——定义学习目标

协助设计师明确他们最需要了解的内容。摒弃模糊的目标：

不是"更好地了解用户"，而是"了解是什么触发用户推荐好友，以及是什么让他们犹豫"

撰写2–3个具体的研究问题。

阶段2——确定参与者

哪些人具备我们需要观察的行为？验证假设：

现有用户、潜在用户还是流失用户？
该行为是他们当前正在做的，还是在新场景下会做出的？
需要多少参与者才能获得有价值的信号？（定性访谈：通常6–10人即可达到饱和）

阶段3——设计研究方案

根据学习目标选择合适的方法。若开展JTBD访谈，需生成访谈指南：

开场："告诉我你最近一次[相关行为]的经历。详细描述当时的情况。"

时间线："是什么触发了这个行为？你首先尝试了什么？是什么让你坚持下去/放弃/切换方案？"

动机："你希望做完这件事后有什么不同？你当时担心什么？是什么差点阻止了你？"

收尾："有没有我没问到但能帮助我更好理解这件事的内容？"

提醒设计师：留意用户使用的名词（候选领域对象）和自然用语——这些内容将为领域层提供输入。

阶段4——规划整合环节

开展研究前：就研究成果的记录和整理方式达成一致。每条记录对应一项观察结果，并标注其关联的研究问题。优先使用原始引述而非总结内容。

Synthesise mode

整合模式

Phase 1 — Take stock

Ask the designer to share what they have — transcripts, notes, data summaries, whatever form. Ask: how was this collected, what questions was it designed to answer, how confident are you in its quality?

Phase 2 — Extract observations

Pull out concrete observations — things users said, did, or felt. Stay close to raw data; no interpretation yet.

If working from memory rather than documents, elicit by asking:

"What was the most surprising thing a user did or said?"
"What kept coming up across different users?"
"What did users struggle with that you didn't expect?"

Phase 3 — Identify patterns

Group observations by theme:

Recurring situations that trigger the behaviour
Common motivations — underlying goals that keep appearing
Shared anxieties — hesitations that show up across users
Workarounds — what users do when the product doesn't serve the job

Phase 4 — Draft candidate job stories

From the patterns: When [situation], I want to [motivation], so I can [expected outcome].

For each: is the "When" specific enough? Is the "I want to" a motivation or a solution? Mark confidence: observed / inferred / assumed.

Phase 5 — Flag research gaps

What questions do the observations not yet answer? What would a designer need to know that this research doesn't tell them? These become input for a follow-up Plan mode session.

阶段1——盘点现有素材

请设计师分享他们拥有的素材——如访谈transcript、笔记、数据摘要等，无论形式如何。询问：这些素材是如何收集的？旨在解答哪些问题？你对其质量的置信度如何？

阶段2——提取观察结果

提炼具体的观察结果——用户所说、所做或所感的内容。贴近原始数据，暂不进行解读。

若仅依赖记忆而非文档，可通过以下问题获取信息：

"用户的哪项行为或表述最令你惊讶？"
"不同用户反复提及的内容是什么？"
"用户遇到了哪些你未曾预料到的困难？"

阶段3——识别模式

按主题对观察结果进行分组：

触发行为的重复场景
共同动机——反复出现的潜在目标
共同焦虑——不同用户都存在的犹豫点
变通方案——当产品无法满足需求时用户的应对方式

阶段4——草拟候选Job Stories

基于识别出的模式：当[场景]时，我想要[动机]，这样我就能[预期结果]。

针对每个Job Story："When"部分是否足够具体？"I want to"部分是动机还是解决方案？标注置信度：观察所得/推断所得/假设所得。

阶段5——标记研究缺口

观察结果尚未解答哪些问题？设计师需要了解但现有研究未覆盖的内容是什么？这些将作为后续规划模式会话的输入。

Completion

交付成果

Produce:

Key observations — significant raw findings
Patterns — grouped themes with supporting observations
Candidate job stories — with confidence ratings
Research gaps — questions still unanswered

Close with: "These job stories are ready to refine at the user needs layer. Run
/layers-user-needs
to work through them."

交付以下内容：

核心观察结果——重要的原始发现
模式——带有支撑性观察结果的分组主题
候选Job Stories——带有置信度评分
研究缺口——尚未解答的问题

收尾："这些Job Stories已准备好在用户需求层进行细化。运行
/layers-user-needs
来处理它们。"