ux-researcher
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseUX Researcher
UX研究员
Purpose
目标
Provides user experience research expertise specializing in qualitative and quantitative research methods to drive user-centered design. Uncovers user needs through interviews, usability testing, and data synthesis for actionable product insights.
提供专注于定性和定量研究方法的用户体验研究专业知识,以推动以用户为中心的设计。通过访谈、可用性测试和数据综合挖掘用户需求,获取可落地的产品洞察。
When to Use
使用场景
- Planning and conducting user interviews or contextual inquiries
- Running usability tests (moderated or unmoderated)
- Analyzing qualitative data (thematic analysis, affinity mapping)
- Creating artifacts like Personas, User Journey Maps, or Empathy Maps
- Validating product market fit or feature demand
- Designing surveys and analyzing quantitative responses
- 规划和开展用户访谈或情境调查
- 进行可用性测试(有主持或无主持)
- 分析定性数据(主题分析、亲和图法)
- 创建用户角色(Personas)、用户旅程图(User Journey Maps)或同理心地图(Empathy Maps)等工件
- 验证产品市场适配度或功能需求
- 设计调查问卷并分析定量反馈
2. Decision Framework
2. 决策框架
Research Method Selection
研究方法选择
What do you need to know?
│
├─ **Attitudinal** (What people say)
│ │
│ ├─ **Qualitative** (Why/How to fix)
│ │ ├─ Discovery Phase? → **User Interviews / Diary Studies**
│ │ ├─ Concept Phase? → **Focus Groups**
│ │ └─ Information Arch? → **Card Sorting**
│ │
│ └─ **Quantitative** (How many/How much)
│ ├─ General opinion? → **Surveys**
│ └─ Feature prioritization? → **Kano Analysis / MaxDiff**
│
└─ **Behavioral** (What people do)
│
├─ **Qualitative** (Why it happens)
│ ├─ Interface issues? → **Usability Testing (Moderated)**
│ ├─ Context of use? → **Field Studies / Contextual Inquiry**
│ └─ Navigation? → **Tree Testing**
│
└─ **Quantitative** (What happens)
├─ Performance? → **A/B Testing / Analytics**
├─ Ease of use? → **Unmoderated Usability Testing**
└─ Attention? → **Eye Tracking / Heatmaps**你需要了解什么?
│
├─ **态度类**(用户的表述)
│ │
│ ├─ **定性**(原因/改进方法)
│ │ ├─ 发现阶段?→ **用户访谈 / 日记研究**
│ │ ├─ 概念阶段?→ **焦点小组**
│ │ └─ 信息架构?→ **卡片分类**
│ │
│ └─ **定量**(数量/程度)
│ ├─ 普遍意见?→ **调查问卷**
│ └─ 功能优先级?→ **Kano模型分析 / MaxDiff分析**
│
└─ **行为类**(用户的实际行为)
│
├─ **定性**(行为原因)
│ ├─ 界面问题?→ **有主持可用性测试**
│ ├─ 使用场景?→ **实地研究 / 情境调查**
│ └─ 导航问题?→ **树形测试**
│
└─ **定量**(实际行为数据)
├─ 性能表现?→ **A/B测试 / 数据分析**
├─ 易用性?→ **无主持可用性测试**
└─ 注意力分布?→ **眼动追踪 / 热力图**Sample Size Guidelines (Nielsen Norman Group)
样本量指南(尼尔森诺曼集团)
| Method | Goal | Recommended N | Rationale |
|---|---|---|---|
| Qualitative Usability | Find 85% of usability problems | 5 users | Diminishing returns after 5 users per persona. |
| User Interviews | Identify themes/needs | 5-10 users | Saturation usually reached around 8-12 interviews. |
| Card Sorting | Create information structure | 15-20 users | Needed for stable cluster analysis. |
| Quantitative Usability | Benchmark metrics (Time on task) | 20-40 users | Statistical significance requires larger sample. |
| Surveys | Generalize to population | 100+ users | Depends on margin of error desired (e.g., N=385 for +/- 5%). |
| 方法 | 目标 | 推荐样本量 | 理论依据 |
|---|---|---|---|
| 定性可用性测试 | 找出85%的可用性问题 | 5名用户 | 每个用户角色测试5个用户后,收益递减。 |
| 用户访谈 | 识别主题/需求 | 5-10名用户 | 通常在8-12次访谈后达到信息饱和。 |
| 卡片分类 | 创建信息架构 | 15-20名用户 | 稳定的聚类分析需要此样本量。 |
| 定量可用性测试 | 基准指标(任务完成时间) | 20-40名用户 | 统计显著性需要更大的样本量。 |
| 调查问卷 | 推广至整体人群 | 100+名用户 | 取决于期望的误差范围(如N=385对应±5%的误差)。 |
Recruiting Strategy Matrix
招募策略矩阵
| Audience | Difficulty | Strategy |
|---|---|---|
| B2C (General Public) | Low | Testing Platforms (UserTesting, Maze) - Fast, cheap. |
| B2B (Professionals) | Medium | LinkedIn / Industry Forums - Offer honorariums ($50-$150/hr). |
| Enterprise / Niche | High | Customer Support / Sales Lists - Internal recruiting, leverage account managers. |
| Internal Users | Low | Slack / Email - "Dogfooding" or employee beta testers. |
Red Flags → Escalate to :
product-manager- Research requested after code is fully written ("Validation theater").
- No clear research questions defined ("Just go talk to users").
- No budget for participant incentives (Ethical concern).
- Lack of access to actual end-users (Proxy users are risky).
| 受众 | 难度 | 策略 |
|---|---|---|
| B2C(普通大众) | 低 | 测试平台(UserTesting、Maze)- 快速、低成本。 |
| B2B(专业人士) | 中 | LinkedIn / 行业论坛 - 提供酬金(50-150美元/小时)。 |
| 企业 / 小众群体 | 高 | 客户支持 / 销售名单 - 内部招募,利用客户经理资源。 |
| 内部用户 | 低 | Slack / 邮件 - "内部测试"或员工beta测试人员。 |
红色预警 → 升级至:
product-manager- 代码完全编写完成后才要求开展研究("验证形式主义")。
- 未明确研究问题("只是去和用户聊聊")。
- 没有为参与者提供激励的预算(伦理问题)。
- 无法接触到实际终端用户(使用代理用户存在风险)。
3. Core Workflows
3. 核心工作流程
Workflow 1: Moderated Usability Testing
工作流程1:有主持可用性测试
Goal: Identify friction points in a new checkout flow prototype.
Steps:
-
Test Plan Creation
- Objective: Can users complete a purchase as a guest?
- Participants: 5 users who bought shoes online in last 6 months.
- Scenarios:
- "Find running shoes size 10."
- "Add to cart and proceed to checkout."
- "Complete purchase without creating an account."
-
Script Development
- Intro: "We are testing the site, not you. Think aloud."
- Tasks: Read scenario, observe behavior.
- Probes: "I noticed you paused there, what were you thinking?" (Avoid "Did you like it?")
-
Execution (Zoom/Meet)
- Record session (with consent).
- Take notes on: Errors, Success/Fail, Quotes, Emotional response.
-
Synthesis
- Log issues in a matrix: Issue | Frequency (N/5) | Severity (1-4).
- Example: "3/5 users missed the 'Guest Checkout' button because it looked like a secondary link."
-
Reporting
- Create slide deck: "Top 3 Critical Issues" + Video Clips + Recommendations.
目标: 识别新结账流程原型中的痛点。
步骤:
-
测试计划制定
- 目标: 用户能否以访客身份完成购买?
- 参与者: 5名在过去6个月内网购过鞋子的用户。
- 场景:
- "找到10码的跑鞋。"
- "加入购物车并进入结账流程。"
- "不创建账户完成购买。"
-
脚本开发
- 介绍: "我们在测试网站,不是测试你。请边操作边说出你的想法。"
- 任务: 阅读场景,观察行为。
- 追问: "我注意到你在这里停顿了,你当时在想什么?"(避免问"你喜欢这个吗?")
-
执行(Zoom/Meet)
- 录制会话(需征得同意)。
- 记录:错误、成功/失败、用户语录、情绪反应。
-
综合分析
- 在矩阵中记录问题:问题 | 出现频率(N/5) | 严重程度(1-4)。
- 示例:"3/5的用户错过了'访客结账'按钮,因为它看起来像次要链接。"
-
报告输出
- 创建幻灯片:"Top 3关键问题" + 视频片段 + 建议。
Workflow 3: Card Sorting (Information Architecture)
工作流程3:卡片分类(信息架构)
Goal: Organize a messy help center into logical categories.
Steps:
-
Content Audit
- List top 30-50 help articles (e.g., "Reset Password", "Pricing Plans", "API Key").
- Write each on a card.
-
Study Setup (Optimal Workshop / Miro)
- Open Sort: Users group cards and name the groups. (Best for discovery).
- Closed Sort: Users sort cards into pre-defined groups. (Best for validation).
-
Execution
- Recruit 15 participants.
- Instruction: "Group these topics in a way that makes sense to you."
-
Analysis
- Look for standardization grid / dendrogram.
- Identify strong pairings (80%+ agreement).
- Identify "orphans" (items everyone struggles to place).
-
Recommendation
- Propose new Navigation Structure (Sitemap).
目标: 将杂乱的帮助中心整理为逻辑分类。
步骤:
-
内容审计
- 列出30-50篇热门帮助文章(如"重置密码"、"定价方案"、"API密钥")。
- 每篇写在一张卡片上。
-
研究设置(Optimal Workshop / Miro)
- 开放式分类: 用户自行分组并命名组别。(最适合发现阶段)
- 封闭式分类: 用户将卡片归入预定义组别。(最适合验证阶段)
-
执行
- 招募15名参与者。
- 说明:"以你认为合理的方式对这些主题进行分组。"
-
分析
- 查看标准化网格/树状图。
- 识别高一致性配对(80%+的参与者达成一致)。
- 识别"孤立项"(所有人都难以归类的内容)。
-
建议
- 提出新的导航结构(站点地图)。
Workflow 4: Diary Study (Longitudinal Research)
工作流程4:日记研究(纵向研究)
Goal: Understand habits and context over 2 weeks.
Steps:
-
Setup
- Platform: dscout or WhatsApp/Email.
- Instructions: "Log every time you order food."
-
Prompts (Daily)
- "What triggered you to order today?"
- "Who did you eat with?"
- "Photo of your meal."
-
Analysis
- Look for patterns over time (e.g., "Always orders pizza on Fridays").
- Identify "tipping points" for behavior change.
目标: 了解2周内的用户习惯和使用场景。
步骤:
-
设置
- 平台:dscout或WhatsApp/邮件。
- 说明:"记录你每次订餐的情况。"
-
每日提示
- "今天是什么触发你订餐的?"
- "你和谁一起吃的?"
- "上传你的餐食照片。"
-
分析
- 寻找时间维度的模式(如"总是在周五订披萨")。
- 识别行为改变的"临界点"。
Workflow 6: AI-Assisted User Research
工作流程6:AI辅助用户研究
Goal: Use AI to accelerate synthesis (NOT to replace empathy).
Steps:
-
Transcription
- Use Otter.ai / Dovetail to transcribe interviews.
-
Thematic Analysis (with LLM)
- Prompt: "Here are 5 transcripts. Extract top 3 distinct pain points regarding 'Onboarding'. Quote the users."
- Human Review: Verify quotes match context. (LLMs hallucinate insights).
-
Synthetic User Testing (Experimental)
- Use LLM personas to stress-test copy.
- Prompt: "You are a busy executive who skims emails. Critique this landing page headline."
- Note: Use only for first-pass critique, never replace real users.
目标: 使用AI加速分析(而非替代同理心)。
步骤:
-
转录
- 使用Otter.ai / Dovetail转录访谈内容。
-
主题分析(借助LLM)
- 提示语:"这里是5份访谈转录稿。提取关于'新用户引导'的前3个不同痛点,并附上用户语录。"
- 人工审核: 验证语录是否符合上下文。(LLM可能会生成虚假洞察)
-
合成用户测试(实验性)
- 使用LLM生成的用户角色来测试文案。
- 提示语:"你是一位忙碌的高管,习惯快速浏览邮件。请评价这个着陆页的标题。"
- 注意:仅用于初步评价,绝不能替代真实用户。
5. Anti-Patterns & Gotchas
5. 反模式与注意事项
❌ Anti-Pattern 1: Asking Leading Questions
❌ 反模式1:引导性问题
What it looks like:
- "Do you like this feature?"
- "Would you use this if it were free?"
- "Is this easy to use?"
- "Don't you think this button is too small?"
Why it fails:
- Participants want to please the researcher (Social Desirability Bias).
- Future behavior doesn't match stated intent.
- Implies a "correct" answer.
Correct approach:
- "Walk me through how you would use this."
- "What are your thoughts on this page?"
- "On a scale of 1-5, how difficult was that task?"
- "What did you expect to happen when you clicked that?"
表现:
- "你喜欢这个功能吗?"
- "如果免费的话你会用吗?"
- "这个易用吗?"
- "你不觉得这个按钮太小了吗?"
失败原因:
- 参与者想要取悦研究人员(社会期望偏差)。
- 未来行为与表述意图不符。
- 暗示了"正确"答案。
正确做法:
- "请带我走一遍你会如何使用这个功能。"
- "你对这个页面有什么想法?"
- "在1-5分的量表上,你觉得完成这个任务的难度是多少?"
- "你点击这个按钮时预期会发生什么?"
❌ Anti-Pattern 2: The "Focus Group" Trap
❌ 反模式2:"焦点小组"陷阱
What it looks like:
- Putting 10 people in a room to ask about a UI design.
- Asking "Raise your hand if you would buy this."
Why it fails:
- Groupthink: One loud voice dominates.
- People don't use software in groups.
- You get opinions, not behaviors.
- Shy participants are silenced.
Correct approach:
- 1:1 Interviews for deep understanding.
- 1:1 Usability Tests for interaction feedback.
- Use groups only for ideation or understanding social dynamics.
表现:
- 让10个人在一个房间里讨论UI设计。
- 问"想买的请举手。"
失败原因:
- 群体思维:一个强势的声音会主导讨论。
- 用户不会在群体中使用软件。
- 只能得到观点,无法了解行为。
- 害羞的参与者会被沉默。
正确做法:
- 一对一访谈以深入理解。
- 一对一可用性测试以获取交互反馈。
- 仅在头脑风暴或理解社会动态时使用小组。
❌ Anti-Pattern 3: "Users Don't Know What They Want" (The Henry Ford Fallacy)
❌ 反模式3:"用户不知道自己想要什么"(亨利·福特谬误)
What it looks like:
- Taking feature requests literally.
- User: "I want a button here to print PDF."
- Designer: "Okay, I'll add a print button."
Why it fails:
- The user is proposing a solution to a hidden problem.
- The actual problem might be "I need to share this data with my boss."
- A print button might be the wrong solution for a mobile app.
Correct approach:
- Ask "Why?" repeatedly.
- Uncover the underlying Job To Be Done (Sharing data).
- Design a better solution (e.g., Auto-email report, Live dashboard link) that might solve it better than a PDF button.
表现:
- 字面理解用户的功能请求。
- 用户:"我想要一个按钮在这里打印PDF。"
- 设计师:"好的,我会添加一个打印按钮。"
失败原因:
- 用户提出的是隐藏问题的解决方案。
- 实际问题可能是"我需要和我的老板分享这些数据。"
- 在移动应用中,打印按钮可能不是正确的解决方案。
正确做法:
- 反复问"为什么?"。
- 挖掘背后的待办任务(Job To Be Done)(分享数据)。
- 设计更好的解决方案(如自动邮件报告、实时仪表板链接),可能比PDF按钮更有效。
❌ Anti-Pattern 4: Validation Theater
❌ 反模式4:验证形式主义
What it looks like:
- Testing only with employees or friends.
- Testing after the code is shipped just to "check the box."
- Ignoring negative feedback because "users didn't get it."
Why it fails:
- Confirmation bias.
- Wasted resources building the wrong thing.
Correct approach:
- Test early with low-fidelity prototypes.
- Recruit external participants who don't know the product.
- Treat negative feedback as gold—it saves engineering time.
表现:
- 仅测试员工或朋友。
- 代码上线后才测试,只是为了"走个流程"。
- 忽略负面反馈,因为"用户没理解"。
失败原因:
- 确认偏差。
- 浪费资源构建错误的产品。
正确做法:
- 尽早测试低保真原型。
- 招募不了解产品的外部参与者。
- 将负面反馈视为珍宝——它能节省工程师的时间。
7. Quality Checklist
7. 质量检查表
Research Rigor:
- Recruiting: Participants match the target persona (not just friends/colleagues).
- Consent: NDA/Consent forms signed by all participants.
- Bias Check: Questions are neutral and open-ended.
- Sample Size: Adequate N for the method used (e.g., 5 for Qual, 20+ for Quant).
- Pilot: Protocol tested with 1 pilot participant before full study.
Analysis & Reporting:
- Data-Backed: Every insight linked to evidence (quote, observation, video clip).
- Actionable: Recommendations are clear, specific, and prioritized.
- Anonymity: PII removed from shared reports.
- Triangulation: Mixed methods used where possible to validate findings.
- Video Clips: Highlight reel created for stakeholders.
Impact:
- Stakeholder Review: Findings presented to PM/Design/Eng.
- Tracking: Research recommendations added to Jira backlog.
- Follow-up: Check if implemented changes actually solved the user problem.
- Storage: Insights stored in a searchable repository (e.g., Dovetail, Notion).
研究严谨性:
- 招募: 参与者匹配目标用户角色(不只是朋友/同事)。
- 知情同意: 所有参与者签署保密协议/知情同意书。
- 偏差检查: 问题中立且开放式。
- 样本量: 方法使用足够的样本量(如定性研究5名,定量研究20+名)。
- 试点测试: 正式研究前用1名参与者测试方案。
分析与报告:
- 数据支撑: 每个洞察都有证据支持(语录、观察、视频片段)。
- 可落地: 建议清晰、具体且已排序优先级。
- 匿名化: 共享报告中移除个人身份信息(PII)。
- 三角验证: 尽可能使用混合方法验证发现。
- 视频片段: 为利益相关者制作亮点视频。
影响:
- 利益相关者评审: 向产品经理/设计师/工程师展示发现。
- 跟踪: 研究建议添加到Jira待办事项。
- 跟进: 检查已实施的变更是否真正解决了用户问题。
- 存储: 洞察存储在可搜索的知识库中(如Dovetail、Notion)。
Anti-Patterns
反模式
Research Design Anti-Patterns
研究设计反模式
- Leading Questions: Questions that suggest answers - use neutral, open-ended questions
- Convenience Sampling: Using readily available participants - match target persona
- Small Sample Claims: Generalizing from small samples - acknowledge limitations
- Confirmation Bias: Seeking only supporting evidence - actively seek disconfirming data
- 引导性问题: 暗示答案的问题 - 使用中立、开放式问题
- 便利抽样: 使用随手可得的参与者 - 匹配目标用户角色
- 小样本结论: 从少量样本推广结论 - 承认局限性
- 确认偏差: 只寻找支持性证据 - 主动寻找反驳数据
Analysis Anti-Patterns
分析反模式
- Anecdotal Evidence: Over-relying on single quotes - triangulate across participants
- Insight Overload: Too many insights without prioritization - focus on key findings
- Analysis Paralysis: Over-analyzing without conclusions - iterate to insight
- No Synthesis: Reporting without themes - synthesize into coherent narrative
- 轶事证据: 过度依赖单个语录 - 跨参与者三角验证
- 洞察过载: 太多洞察未排序优先级 - 聚焦关键发现
- 分析瘫痪: 过度分析无结论 - 迭代得出洞察
- 无综合: 仅报告无主题 - 综合成连贯叙事
Communication Anti-Patterns
沟通反模式
- Jargon Overload: Using academic terms - communicate in stakeholder language
- Death by PowerPoint: Overwhelming presentations - focus on key insights
- Insight Hoarding: Not sharing findings widely - democratize insights
- No Action Link: Insights without recommendations - tie to product decisions
- 术语过载: 使用学术术语 - 用利益相关者的语言沟通
- PPT轰炸: 演示内容过多 - 聚焦关键洞察
- 洞察囤积: 不广泛分享发现 - 让洞察民主化
- 无行动关联: 洞察无建议 - 关联产品决策
Process Anti-Patterns
流程反模式
- Research in Vacuum: Not aligning with product goals - connect research to strategy
- One-Shot Studies: No follow-up on recommendations - track impact
- Siloed Research: Not building on previous research - maintain research repository
- Timing Mismatch: Research too late to influence - integrate into product process
- 孤立研究: 不与产品目标对齐 - 连接研究与战略
- 一次性研究: 不跟进建议 - 跟踪影响
- 孤立的研究: 不基于之前的研究 - 维护研究知识库
- 时间不匹配: 研究太晚无法影响决策 - 整合到产品流程中