Back to Details

mental-model-mismatch-and-onboarding

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Mental-model mismatches and onboarding

Mental-model 不匹配与用户引导流程

When the user's mental model and the system's actual behavior diverge, three things can be done: change the system to match the user's model; change the user's model through onboarding and teaching; or surface the divergence so the user knows when their model doesn't apply.
当用户的mental-model与系统实际行为存在差异时，可以采取三种措施：修改系统以匹配用户的心智模型；通过用户引导与教学改变用户的心智模型；或者明确指出这种差异，让用户知道自己的心智模型何时不再适用。

Diagnosing mismatches

诊断不匹配问题

Common signals of mental-model mismatch:
  • Repeated support tickets about the same misunderstanding ("Why doesn't X work?" when X works fine — the user's expectation didn't match).
  • Drop-off at the same flow step (the user reached this step expecting one thing; got another; left).
  • High undo / cancel rate at a particular action.
  • Forum / community questions asking how to do things the system doesn't actually do.
  • Negative reviews mentioning unmet expectations.
Each pattern points to a divergence between user model and system reality. The fix depends on which side to change.
mental-model不匹配的常见信号：
  • 重复的支持工单：关于同一误解的工单（比如用户问“为什么X无法工作？”但X实际运行正常——用户的预期与实际不符）。
  • 同一流程步骤的用户流失：用户到达该步骤时预期某一结果，实际得到的却是另一结果，于是离开。
  • 特定操作的高撤销/取消率。
  • 论坛/社区提问：询问如何实现系统并不支持的功能。
  • 负面评价：提及未被满足的预期。
每种模式都指向用户心智模型与系统实际情况的差异。解决方案取决于选择改变哪一方。

Three fix strategies

三种修复策略

1. Change the system

1. 修改系统

If the user's model is reasonable and many users share it, change the system to match. Often this is also the simpler design.
Example: users expect "save" to commit immediately. If your system actually queues the save for later, either make save immediate or rename the action to reflect the actual behavior.
如果用户的心智模型合理且被多数用户认同，修改系统以匹配该模型。通常这也是更简单的设计方案。
示例：用户期望“保存”操作立即生效。如果你的系统实际上是将保存操作排入队列延后执行，要么让保存立即生效，要么重命名该操作以反映实际行为。

2. Teach the model

2. 教授心智模型

If the system's behavior is genuinely better and users can learn it, invest in onboarding.
Examples:
  • Anti-lock brakes required teaching: "brake firmly, steer; don't pump." Manufacturer campaigns and driver-ed material taught the new model.
  • Spreadsheets taught a new computing model in the early 1980s; once learned, transferable.
  • Modal editors (vim) require learning that "modes" exist; users who learn become extremely productive.
Onboarding that teaches a model:
  • Walks the user through the first task with explicit guidance.
  • Provides explanation in context — not a separate tutorial page.
  • Shows the system's state as actions happen, so the user observes the model.
  • Uses progressive disclosure of complexity over time.
如果系统的行为确实更优，且用户可以学会，就投入资源做用户引导。
示例：
  • 防抱死刹车需要教学：“用力踩刹车，操控方向；不要反复踩踏。”制造商的宣传活动和驾驶培训材料教授了这种新模型。
  • 电子表格在20世纪80年代初引入了一种新的计算模型；一旦学会，即可迁移应用。
  • 模态编辑器（如vim）需要用户了解“模式”的存在；学会的用户会变得极高产。
教授心智模型的用户引导：
  • 引导用户完成首个任务：提供明确指导，而非“这里是功能介绍”，而是“让我们创建你的第一个项目”。
  • 提供上下文内的解释——不是单独的教程页面。
  • 操作时展示系统状态，让用户观察到模型的运作。
  • 逐步披露复杂功能：随着时间推移逐步展示复杂内容。

3. Surface the divergence

3. 明确指出差异

When the system genuinely differs and you can't (or shouldn't) change the system, surface the difference at the moment it matters.
Examples:
  • A "soft-deleted" item shows "Deleted (recoverable for 30 days)" — surfaces the system's actual model.
  • A subscription cancellation shows "Canceled — access continues until [date]" — surfaces what canceled actually means.
  • A search that ranks by relevance (not chronologically) shows "Sorted by relevance" — explains the unexpected order.
The point is to expose the divergence at the moment of the action, not buried in documentation.
当系统确实存在差异，且你无法（或不应）修改系统时，在关键节点明确指出这种差异。
示例：
  • 被“软删除”的项目会显示“已删除（30天内可恢复）”——明确展示系统的实际模型。
  • 订阅取消时显示“已取消——访问权限持续至[日期]”——明确说明“取消”的实际含义。
  • 按相关性排序（而非时间顺序）的搜索结果会显示“按相关性排序”——解释意外的排序方式。
关键在于在操作发生的时刻揭示差异，而非将其隐藏在文档中。

Onboarding patterns that teach mental models

教授心智模型的用户引导模式

Guided first task

引导式首个任务

Walk the user through their first concrete task — not "here's a tour" but "let's create your first project."
html
<aside class="onboarding-guide">
  <h3>Let's create your first project</h3>
  <p>1. Click the + button at the top right →</p>
  <!-- The new-project button is highlighted; the user clicks; next step appears -->
</aside>
Active practice teaches the interaction model better than passive watching.
引导用户完成第一个具体任务——不是“这里是游览介绍”，而是“让我们创建你的第一个项目”。
html
<aside class="onboarding-guide">
  <h3>Let's create your first project</h3>
  <p>1. Click the + button at the top right →</p>
  <!-- The new-project button is highlighted; the user clicks; next step appears -->
</aside>
主动实践比被动观看更能有效地教授交互模型。

Inline explanation at the moment of action

操作时刻的内联解释

When the user encounters a feature whose model differs from common expectations, explain in line:
html
<div class="field">
  <label for="visibility">Project visibility</label>
  <select id="visibility">
    <option value="private">Private (default)</option>
    <option value="team">Team</option>
    <option value="public">Public</option>
  </select>
  <p class="hint">
    Private = only you. Team = everyone in your workspace. Public = anyone with the link.
  </p>
</div>
The explanation prevents knowledge mistakes by surfacing the actual model.
当用户遇到心智模型与普遍预期不同的功能时，提供内联解释：
html
<div class="field">
  <label for="visibility">Project visibility</label>
  <select id="visibility">
    <option value="private">Private (default)</option>
    <option value="team">Team</option>
    <option value="public">Public</option>
  </select>
  <p class="hint">
    Private = only you. Team = everyone in your workspace. Public = anyone with the link.
  </p>
</div>
这种解释通过展示实际模型，避免用户产生认知错误。

Progressive complexity

渐进式复杂度

Don't expose every feature on day one. Reveal them as the user encounters surfaces where they apply. The user's interaction model grows with their need.
不要在第一天就展示所有功能。随着用户遇到需要使用这些功能的场景时再逐步揭示。用户的交互模型会随着需求增长而逐步建立。

"What's new" surfaces

“新功能”展示

When you ship a feature whose model users wouldn't predict from prior versions, announce explicitly. A "what's new in v2" panel that explains how the new feature works, with brief examples, prevents widespread misunderstanding.
当你发布的功能其心智模型无法从旧版本预测时，要明确告知用户。一个“v2版本新功能”面板，解释新功能的工作原理并附带简短示例，可以避免广泛的误解。

Anti-patterns

反模式

  • Tour-only onboarding. Walk the user through a passive demo. They forget within hours; their interaction model isn't reinforced.
  • Documentation-as-onboarding. "If you have questions, see the help center." Most users don't.
  • Hidden divergences. A system that differs from expectation but doesn't say so. Users assume; produce errors.
  • Over-explanation in normal flows. If you have to constantly explain how things work, the design isn't matching mental models well; redesign rather than over-tutorialize.
  • 仅游览式引导：引导用户完成被动演示。用户几小时内就会忘记；他们的交互模型未得到强化。
  • 以文档作为引导：“如有疑问，请查看帮助中心。”大多数用户不会这么做。
  • 隐藏差异：系统与预期不符但未告知用户。用户会自行假设，进而产生错误。
  • 常规流程中过度解释：如果你必须不断解释功能的工作方式，说明设计与心智模型匹配度不佳；应重新设计而非过度教程化。

Heuristics

启发法

  1. The "first surprise" diagnostic. Watch a new user. Where do they say "huh, that's not what I expected"? Each surprise is a mental-model mismatch — fix the system or the surfaced model.
  2. The support-ticket pattern audit. Cluster recent support tickets by the misunderstanding behind them. Patterns reveal mismatches.
  3. The cancel-rate signal. Steps with high cancel rates often mean users reached them with the wrong expectation. Investigate.
  1. “首次意外”诊断：观察新用户的操作。他们在哪里说“嗯，这和我预期的不一样”？每一处意外都是mental-model不匹配——要么修改系统，要么明确展示模型。
  2. 支持工单模式审计：将近期的支持工单按背后的误解分类。模式会揭示不匹配问题。
  3. 取消率信号：取消率高的步骤通常意味着用户带着错误的预期到达该步骤。需要深入调查。

Related sub-skills

相关子技能

  • mental-model
     (parent).
  • mental-model-system-vs-interaction
     — distinguishing the two model types.
  • expectation-effect
     — mental models are the user's expectations.
  • affordance
     — affordance teaches interaction models at the per-element level.
  • mimicry
     — borrowing familiar patterns leverages existing models.
  • mental-model
    （父技能）。
  • mental-model-system-vs-interaction
    ——区分两种模型类型。
  • expectation-effect
    ——mental-model就是用户的预期。
  • affordance
    ——功能可见性在元素层面教授交互模型。
  • mimicry
    ——借鉴熟悉的模式可以利用用户已有的心智模型。