Overview and teacher commentary will appear here.
User-Centred design in Topic B moves from theory to practice — planning research, applying methods, building personae, evaluating usability and mapping tasks. These notes supplement your classroom materials and textbook; they are not a substitute for them.
Students must be able toConstruct a plan for a UCD process based on research questions that engage with user-Centred research methods.
User-Centred design (UCD) is a design process that builds every decision around the real needs, behaviours and limits of the people who will use the product — not around the designer's assumptions.
Before you can research your users, you need a plan. A UCD research plan is a short written document that answers three questions: What do I need to find out? How will I find it out? What will I do with the findings?
Your plan begins with research questions — specific things you don't yet know about your users. Good research questions are about people's problems, habits and motivations. Weak research questions ask about product features. Compare:
Weak: "What features should the app have?"
Better: "What step in the current checkout process causes users to abandon their cart?"
Once you have your questions, choose methods that can actually answer them (more on methods in 1.1.2), decide how many participants you need and how you'll record findings, and set a realistic timeline.
以用户为中心的设计（UCD）是一种以产品最终用户的真实需求、行为和限制为核心的设计流程——而不是以设计师的假设为出发点。
在开始用户研究之前，你需要一份研究计划。UCD研究计划是一份简短的文件，回答三个核心问题：我需要了解什么？如何获得这些信息？如何将发现转化为设计决策？
计划的起点是研究问题——你尚未了解的、与用户相关的具体问题。好的研究问题关注用户的痛点、习惯和动机，而不是直接询问产品功能。比较以下两个例子：
弱："App应该有哪些功能？"
好："用户在结账流程的哪个步骤最常放弃购物？"
确定研究问题后，选择能够回答这些问题的研究方法，确定所需参与者数量、记录方式，并制定时间计划。
7-Step UCD Research Plan
- Define the design context — Who is the product for? What problem does it address?
- Identify your target user group(s) — Who experiences this problem most acutely?
- Write 3–5 specific research questions — Focus on behaviour, frustration and motivation.
- Choose 2–3 research methods — Match each method to the questions it can answer best.
- Plan your sample — How many participants? How will you recruit them?
- Decide how you will record and organise data — Notes, recordings, observation logs?
- Set a timeline — When will research be complete? When will you start designing?
Each product below presents a different kind of design challenge. Read the context, then study how the inquiry strategy would be structured differently in each case. Notice how the choice of research methods follows directly from the nature of the users and the decisions that need to be made.
- Observe the lift lobby during class transitions to map actual usage patterns
- Interview students with mobility needs and staff who use it daily with equipment
- Survey the broader community on signage clarity and door timing
- Conduct task analysis of the full journey for a wheelchair user, from ground floor to classroom
- Analyse existing usage data to identify where users hesitate or drop off
- Run separate unstructured interviews with power users and casual users — their needs diverge sharply
- A/B test two navigation prototypes with a live subset of real users
- Hold a focus group with content creators, who have the highest stake in how navigation changes affect reach
- Conduct field research at gyms, running events and recreational sports clubs
- Run structured taste-testing sessions with Likert scales on sweetness, aftertaste and colour preference
- Interview recreational athletes about when, where and why they select a drink — before exercise, during or after
- Use a focus group to test packaging and branding concepts against competitor products on a shelf mock-up
- Observe professionals using drills on active construction sites — not in a showroom
- Conduct structured interviews with tradespeople and DIY users separately, using the same questions to expose divergent priorities
- Apply task analysis to the bit-changing process, which preliminary research suggests is a shared pain point
- Ergonomic assessment: measure grip force range, one-handed operation and balance point across percentile hand sizes
Students must be able toApply a variety of user-centred research methods (field research, user observation, interviews, questionnaires and focus groups) and analyse data to establish users' characteristics, behaviours, and the wants and needs of the target population defined by their demographics.
Different research methods collect different kinds of information. The choice of method should be driven by your research questions — not by what is easiest to set up.
Field research — Observe users in their real environment (a kitchen, a bus, a hospital). Reveals the gap between what users say they do and what they actually do. Prone to the observer effect: people may behave differently when they know they're being watched.
User observation — Watch a participant perform a specific task. Can be in a lab or real setting. Good for spotting hesitation, workarounds and errors that users wouldn't think to mention.
Interviews — A structured interview uses the same fixed questions for every participant, making results easy to compare. An unstructured interview follows the conversation wherever it leads, uncovering unexpected insights but producing harder-to-analyse data. Most real research uses a semi-structured format.
Questionnaires — Reach many people quickly. A Likert scale (typically 1–5, from Strongly Disagree to Strongly Agree) turns attitudes into numbers, giving you quantitative data. Always include a few open-ended questions alongside Likert items to capture the qualitative data that explains the numbers.
Focus groups — 5–8 people discuss a topic together. Useful for exploring social attitudes and generating ideas. Risk: one dominant voice can skew the group's responses.
不同的研究方法收集不同类型的信息。选择方法的依据应该是你的研究问题，而不是哪种方法最容易执行。
实地调研（Field research） — 在用户的真实环境（厨房、公交车、医院）中观察他们。能揭示用户口头描述与实际行为之间的差距。注意观察者效应：用户知道被观察时可能会改变行为。
用户观察（User observation） — 观察参与者完成特定任务。可在实验室或真实环境中进行。适合发现用户不会主动提及的犹豫、变通做法和操作错误。
访谈（Interviews） — 结构化访谈对每位参与者使用相同的固定问题，结果易于比较；非结构化访谈随对话自由发展，能发现意外信息，但分析难度较大。大多数真实研究采用半结构化形式。
问卷调查（Questionnaires） — 可快速收集大量数据。李克特量表（通常1–5分，从"非常不同意"到"非常同意"）将态度转化为数字，即定量数据。建议配合开放式问题，以获取能解释数字背后原因的定性数据。
焦点小组（Focus groups） — 5–8人共同讨论一个话题。适合探索社会态度和激发创意。风险：某个强势声音可能主导整个小组的讨论走向。
| Method | Best for finding… | 适合发现什么 | Main weakness |
|---|---|---|---|
| Field research | What users actually do (not what they say) | 真实行为（而非口头描述） | Time-consuming; observer effect |
| User observation | Specific errors, hesitation points, workarounds | 具体错误、犹豫节点、变通做法 | Artificial setting may not reflect real use |
| Structured interview | Consistent, comparable answers across many people | 可横向比较的标准化数据 | Rigid questions miss unexpected issues |
| Unstructured interview | Deep, unexpected insights | 深度、意外的发现 | Hard to analyse; small sample only |
| Questionnaire / Likert | Quantitative satisfaction or attitude scores | 量化满意度或态度分数 | Doesn't explain why |
| Focus group | Group opinions, social dynamics, idea generation | 群体意见、社会互动、创意激发 | One dominant voice can skew results |
A single project typically combines 2–3 methods. Field research or observation uncovers what problems exist; interviews and questionnaires explain why and how many people are affected.
Press the button to receive a field research assignment. Each assignment gives you a specific real-world setting, something to observe, and instructions for data to collect. Treat it as a genuine brief.
Type your focus group topic, then generate facilitator questions. Your topic is inserted into each question automatically — Mad Libs style. Press "New set" to get a different selection of six questions.
Answer the five usability questions as if you are a typical user. Then click "Get survey results" to see how 1,000 simulated respondents — drawn from a realistic distribution centred on your answers — responded to the same questions.
Students must be able toCreate a primary persona or personae based on user-centred research to aid design development.
A persona is a fictional but research-based character who represents a real user group. You build one from patterns found in your research — not from your own imagination or assumptions about users.
There are three types:
Primary persona — The main user the design is built for. If your design does not fully satisfy this person, it fails. You design for one primary persona first; everything else is secondary.
Secondary persona — An additional user with needs beyond the primary. Satisfy them if you can, but never at the cost of the primary persona's experience.
Anti-persona — A representation of someone who would misuse the product in ways that harm real users or the business (Nielsen Norman Group). Designing with them in mind reveals security gaps and abuse vectors that would otherwise be discovered too late.
What makes a good persona? Specific frustrations (not vague ones), realistic limits and habits, a concrete use scenario. A persona who "loves technology and is very organised" tells you nothing useful. A persona who "uses her phone one-handed while holding her coffee, drops it twice a week, and gives up on apps that require more than two taps to find what she needs" guides real design decisions.
人物画像（Persona）是一个虚构但基于真实研究的人物，代表真实的用户群体。它来自你在研究中发现的规律——而不是设计师对用户的想象或假设。
人物画像分为三种类型：
主要人物画像（Primary persona） — 设计优先服务的核心用户。如果设计无法完全满足此人的需求，设计就是失败的。你应先为一个主要用户画像而设计。
次要人物画像（Secondary persona） — 有额外需求的另一类用户。在不损害主要用户体验的前提下尽量满足其需求。
反人物画像（Anti-persona） — 代表可能滥用产品、损害真正用户或业务利益的人（尼尔森诺曼集团定义）。以此为参照进行设计，可提前发现安全漏洞和滥用风险。
优质人物画像的特征：具体的痛点（非模糊描述）、真实的限制和习惯、具体的使用场景。"喜欢科技、非常有条理"的人物画像毫无参考价值。"单手拿着咖啡使用手机、每周摔手机两次、超过两步找不到功能就放弃App"的人物画像才能指导真实的设计决策。
Interview questions that build good personas
Don't ask "What features do you want?" Ask about problems, habits and frustrations instead. 不要直接问"你想要什么功能？"——要问问题、习惯和痛点。
| English | 中文 |
|---|---|
| "What is the most frustrating thing about the current way of doing this task?" | "你觉得现在做这件事最烦人的地方是什么？" |
| "How do you currently solve this problem?" | "你现在是怎么解决这个问题的？" |
| "What would make this task faster or easier for you?" | "什么能让这件事做起来更快或更轻松？" |
| "When was the last time you gave up on a product like this? Why?" | "你上次放弃使用这类产品是什么时候？为什么？" |
| "What do you usually do first when you open this type of app?" | "你打开这类App时，通常第一步做什么？" |
Common persona-building mistakes
| Mistake | Why it fails | 中文说明 |
|---|---|---|
| Designing for "everyone" | "Everyone" is not a user group. No product fits all people. | "所有人"不是用户群体，没有产品能适合所有人。 |
| Inventing needs without research | You will design for your own imagination, not real users. | 不调研就编造需求，设计出来的只满足设计师自己。 |
| Making the persona too perfect | Real users have frustrations, limits and bad habits. | 真实用户有烦恼、限制和坏习惯，完美的画像不真实。 |
| Skipping the secondary persona | You miss easy improvements that serve more people. | 忽略次要画像会错过一些简单的改进机会。 |
Persona template — for your project
Choose a product context, fill in the fields based on your research findings, then click Finish to generate a formatted primary persona card. Use real or imagined research data — the goal is to practise making the persona specific and concrete.
Students must be able toExplain and apply five usability objectives (learnability, efficiency, memorability, errors and satisfaction) in order to evaluate a product.
Jakob Nielsen's five usability objectives — often called the "Five E's" — give you a structured framework for evaluating any product from the user's perspective.
The objectives are not independent. A product that scores well on Efficiency but poorly on Error tolerance may actually frustrate expert users when they recover from mistakes. Always evaluate all five together.
Measuring usability — the SUS:
The System Usability Scale (SUS) is a standardised 10-question Likert survey that produces a score from 0–100. The industry average is around 68. Below 50 indicates serious problems; above 85 is excellent. Scores below 68 are a signal to investigate which of the five objectives are being violated.
Poka-yoke and the Errors objective:
Poka-yoke (Japanese: "mistake-proofing") is a design strategy that directly targets the Errors objective. The key idea: instead of warning users after they make a mistake, the design makes the mistake impossible — or automatically corrects it. Examples: a USB-C plug that works in any orientation; a microwave that physically cannot run with the door open; a car that won't start in gear.
雅各布·尼尔森（Jakob Nielsen）提出的五个可用性目标——通常称为"五个E"——是从用户角度评估任何产品的结构化框架。
这五个目标并非相互独立。一个在效率上表现优秀但容错性差的产品，可能会在用户犯错后恢复过程中带来严重挫败感。评估时应综合考虑全部五个目标。
衡量可用性——系统可用性量表（SUS）：
SUS是一份标准化的10题李克特量表，生成0–100分的可用性评分。行业平均分约为68分。低于50分表明存在严重问题；高于85分为优秀。低于68分是需要深入调查哪些可用性目标被违反的信号。
防错设计（Poka-yoke）与容错性目标：
防错设计（日语：poka-yoke，"防止错误"）直接针对容错性目标。核心理念：与其在用户犯错后发出警告，不如通过设计让错误根本无法发生——或自动纠正。例子：USB-C接口可以任意方向插入；微波炉门未关时实体无法运行；汽车挂挡状态下无法启动。
The Five E's are also written as: Effective, Efficient, Engaging, Error tolerant, Easy to learn. The IB syllabus uses Learnability, Efficiency, Memorability, Errors and Satisfaction — these map directly onto the Five E's.
Each card below shows one of the Five E's in action. For each example, ask yourself: how would you measure whether this product succeeds on this objective? What would a failure look like?
Students must be able toApply task analysis techniques to break down a process into steps and identify the critical points for design improvement.
Task analysis is the process of breaking down a user's overall goal into a hierarchy of smaller, observable steps. The most common approach is Hierarchical Task Analysis (HTA).
Users rarely know exactly what makes a process frustrating — they just stop using the product. By mapping every micro-step, you find precisely where users hesitate, make errors, or give up entirely.
How to conduct a basic HTA:
1. Start with the user's overall goal (e.g., "Pay for an online order").
2. Break it into 4–7 main stages.
3. Break each stage into sub-steps — go until the steps are actions a user can actually perform.
4. Flag each step where users might: (a) hesitate, (b) make an error, or (c) abandon.
Cognitive load is the amount of mental effort required at any given step. When a step asks users to hold too many things in mind at once, cognitive load spikes — leading to errors and abandonment. The fix is usually one of three things: split the step into smaller pieces, add a visual cue, or reduce the number of choices.
任务分析（Task analysis）是将用户总目标拆解为可观察的逐级子步骤的过程。最常见的方式是层级任务分析（HTA，Hierarchical Task Analysis）。
用户通常无法清楚描述某个流程为何令人沮丧——他们只是停止使用产品。通过梳理每一个微小步骤，设计师可以精准定位用户犹豫、出错或放弃的节点。
基本HTA操作步骤：
1. 从用户的总目标出发（例如："完成一次网上支付"）。
2. 拆解为4–7个主要阶段。
3. 将每个阶段进一步拆解为子步骤，直到步骤是用户能实际执行的具体动作为止。
4. 标注每个可能出现以下情况的步骤：(a) 犹豫、(b) 出错、(c) 放弃。
认知负荷（Cognitive load）是用户在某一步骤需要投入的脑力总量。当一个步骤要求用户同时在脑中记住太多信息时，认知负荷急剧上升——导致出错和放弃。解决方案通常是以下三种之一：将该步骤拆分为更小的部分、添加视觉提示，或减少选项数量。
Worked example — Mobile shopping app checkout
| Stage | Sub-steps | ⚠ Flag |
|---|---|---|
| 1. Open cart | Tap cart icon → review items → adjust quantities | |
| 2. Enter details | Confirm shipping address → choose delivery speed | |
| 3. Payment | Select payment method → if not logged in: manually type card number, expiry, CVV → system prompts: "Create an account before continuing" | ⚠ High cognitive load. Users must stop, decide, and often abandon. |
| 4. Confirm | Review order summary → tap "Place Order" |
Design change identified: Move account creation to after order confirmation. Add a "Save my info for next time?" toggle as the final optional step. This removes an interruption mid-task, reducing abandonment at the highest-friction point.
Enter your user's goal, build the hierarchy of steps, flag the friction points — then generate a visual HTA map you can rearrange and export. Try it →
Nine questions covering all five learning objectives. Select one answer per question, then click "Check all answers" to see your score and the explanations.
Show example answer
Easy to learn (Learnability):
The interface should require minimal instruction to get started. For example, the watch could have only two physical buttons: one to cycle through functions (steps, heart rate, time) and one to confirm. A short narrated audio tutorial that plays the first time the watch turns on guides the user without requiring them to read a manual — reducing frustration for users unfamiliar with complex technology.
Error tolerant (Errors):
The design should prevent or forgive common mistakes. For example, if the user accidentally touches the screen with a shaky hand, the watch could require a 1-second long press to change any setting rather than responding to brief contact. Additionally, if the user forgets to stop a workout session, the watch could auto-detect 5 minutes of inactivity and display: "Did you finish exercising?" — preventing inaccurate data recording without requiring any user action.
Show example answer
A designer would conduct task analysis by breaking the checkout process into a hierarchy of sub-tasks:
- Open cart
- Review items
- Select shipping address
- Choose shipping speed
- Enter payment information
- Confirm order
Pain point discovered: Users frequently abandon checkout after realising they must manually re-enter credit card details because they are not logged in, and a "Create account" screen appears after entering their email address but before payment — interrupting the task mid-flow and increasing cognitive load at the highest-friction point.
Design change: Add a guest checkout option that processes payment without requiring account creation. Reposition the account creation prompt to after the order is confirmed, with a simple "Save my info for next time?" toggle. This removes the mid-task interruption and reduces the steps between intent and completion.
Show example answer
Scenario: A team is designing an AI-assisted moderation dashboard for a large social media platform. Human moderators use the dashboard to review flagged posts for hate speech, bullying and misinformation.
Primary persona — Elena, 34, professional content moderator:
Goals: Quickly and accurately review flagged posts; minimise false removals; maintain personal wellbeing by avoiding repeated exposure to graphic content.
Needs: Clear visual indicators of violation type, batch processing tools, keyboard shortcuts, automatic blurring of graphic images.
Anti-persona — "Bad Actor" Ben:
Represents a malicious user or coordinated group who submits mass false reports to overwhelm moderators, or posts borderline content designed to confuse the AI filter. He is not a moderator — he is someone who exploits the system.
How the anti-persona prevents misuse: By designing with Ben in mind, the team can add safeguards: rate-limiting reports from a single account per hour; requiring a verified account before submitting reports; flagging accounts that repeatedly submit false reports for human review. These safeguards protect Elena from being flooded with junk reports, and protect legitimate users from coordinated false-flagging campaigns.
Show example answer
| Method | Advantage in this context | Disadvantage in this context |
|---|---|---|
| Field research | Observes real behaviour: watching students collaborate in a library or dorm reveals workarounds, distractions and silent frustrations (e.g., one student retyping instead of using comments) that users would never mention in an interview. | Time-consuming and unpredictable: group projects happen irregularly, and the observer's presence may change behaviour (observer effect — students may appear more productive than normal). |
| Structured interview | Produces consistent, comparable data across many students. Identical questions ("On a scale of 1–5, how easy is it to recover a deleted file?") allow statistical comparison between year groups or between tools. | Misses unconscious issues: students may say "collaboration is easy" while field research would show them spending 10 minutes per session figuring out who edited what. Structured questions cannot uncover what users don't think to mention. |
Recommended approach: Conduct structured interviews first to identify broad satisfaction patterns, then use targeted field observation with a small group to uncover the specific usability problems that interviews missed.
Show example answer
Research question 1: "At what specific point during the docking process do users fail to properly lock the scooter, and what physical or cognitive factors contribute to that failure?"
Method: User observation at an existing kiosk, watching 8–10 real users return a scooter.
Justification: Users are unlikely to be aware of or articulate the precise moment of failure in an interview. Observation at the actual kiosk captures the physical interaction in context — including hesitation, incorrect grip and missed feedback cues — that a questionnaire would miss.
Research question 2: "How satisfied are users with the current touchscreen interface, and which specific steps feel most confusing?"
Method: A short questionnaire with a 5-question Likert scale (satisfaction rating per screen) plus two open-ended follow-up questions.
Justification: A questionnaire can be distributed digitally to a large number of users quickly (e.g., via QR code at the kiosk), producing quantitative satisfaction scores that can be compared across different kiosk locations. The open-ended questions capture qualitative context that the Likert scores alone cannot provide.
Linking Questions
- To what extent does UCD rely on a strong foundation of ergonomics? (A1.1)
- How important is a good understanding of user-centred research methods to ensure effective UCD? (A2.1)
- To what extent can the UCD process be influenced by the quality of modelling and prototyping of potential design solutions? (B2.2)
- To what extent should a UCD process focus on ensuring inclusive design? (C1.2)
- What influence can product analysis and evaluation have on the effectiveness of UCD? (C3.1)