Skip to content
W3S101

DESIGN ONBOARDING PROCESS 提示

提示词工程：设计用户引导流程

Section titled “提示词工程：设计用户引导流程”

我们可以利用大语言模型（LLM），如 ChatGPT，来为新用户或新员工设计有效且引人入胜的用户引导流程。这项技术使我们能够借助 LLM 庞大的知识库和生成能力，创建结构化的用户引导体验，帮助个人走向成功。

指导 LLM 设计用户引导流程

Section titled “指导 LLM 设计用户引导流程”

要让 LLM 生成用户引导流程，你需要提供一个清晰的提示词，详细说明你的需求。这包括指定目标受众（新用户或新员工）、产品或公司，以及用户引导的任何关键目标。

以下是你如何向 LLM 提供提示词的示例：

User: Can you design an onboarding process for new customers of our SaaS product, 'ConnectSphere,' which is a project management tool? The goal is to get them to create their first project and invite a team member within the first week.


LLM: Certainly! Here's a suggested onboarding process for new 'ConnectSphere' customers, aiming for first project creation and team member invitation within one week:


**Phase 1: Welcome & Initial Setup (Day 1)**
1.  **Welcome Email:**
    *   Personalized greeting.
    *   Clear call-to-action: 'Log in and Set Up Your Workspace.'
    *   Link to a quick start guide (2-minute video or concise webpage).
    *   Link to support resources.
2.  **First Login Experience:**
    *   Brief interactive tour highlighting key UI elements (Dashboard, Projects, Team Members).
    *   Guided workspace setup (e.g., naming workspace, setting preferences).


**Phase 2: Core Feature Adoption (Day 1-3)**
3.  **Guided Project Creation:**
    *   In-app prompt: 'Let's create your first project!'
    *   Simple template options for common project types.
    *   Tooltips explaining fields like 'Project Name,' 'Due Date,' 'Description.'
4.  **Educational Nudges:**
    *   Short email/in-app message on Day 2: 'Tip: Did you know you can assign tasks in ConnectSphere? Here’s how…'
    *   Highlight benefits of using specific features related to their first project.


**Phase 3: Collaboration & Engagement (Day 3-5)**
5.  **Guided Team Invitation:**
    *   In-app prompt: 'Projects are better with a team! Invite your first team member.'
    *   Simplified invitation flow.
    *   Brief explanation of roles (Admin, Member, Viewer) if applicable.
6.  **Value Reinforcement:**
    *   Email on Day 4: 'See what others are achieving with ConnectSphere teams!' (Include a short testimonial or case study snippet).


**Phase 4: Ongoing Support & Next Steps (Day 6-7 & Beyond)**
7.  **Check-in & Feedback:**
    *   Automated email: 'How's your first week with ConnectSphere going? Need help?'
    *   Link to a short feedback survey or support channel.
8.  **Advanced Feature Introduction:**
    *   Once first project and team invite are complete, introduce next-level features (e.g., reporting, integrations) via in-app messages or targeted emails.
9.  **Resource Hub:**
    *   Ensure easy access to a comprehensive knowledge base, FAQs, and video tutorials.


**Key Metrics to Track:**
*   Activation rate (logged in, completed setup).
*   Percentage of users creating their first project within 7 days.
*   Percentage of users inviting at least one team member within 7 days.
*   Feature adoption rates for core functionalities.

在此示例中，用户清晰地说明了产品、目标受众和目标。LLM 提供了一个结构化、可操作的用户引导计划。

设计用户引导流程提示词的最佳实践

Section titled “设计用户引导流程提示词的最佳实践”

为了最大化利用 LLM 协助设计用户引导流程，请考虑以下最佳实践：

  • 明确目标：新用户/员工应在用户引导结束时达到什么目标？要具体说明（例如，“理解公司文化”、“掌握核心产品功能”、“提交第一份报告”）。
  • 指定目标受众：是针对新的软件用户、特定部门的新员工，还是远程员工？不同受众的需求是不同的。
  • 提供上下文：提及公司、产品或服务。LLM 获得的上下文越多，生成的用户引导流程就越定制化。
  • 请求结构：要求用户引导流程分阶段、有时间表或包含特定组成部分（例如，“包括入职前、第一天、第一周和第一个月的活动”）。
  • 询问互动策略：请求使用户引导更具互动性、个性化或游戏化的想法。
  • 迭代和优化：将 LLM 的初步输出作为起点。提出后续问题以优化特定部分、请求更多细节或探索替代方法。

应用示例：Python 实现

Section titled “应用示例：Python 实现”

让我们通过一个实际示例，使用 Python 和 OpenAI API（应用程序编程接口）来引导 LLM 设计用户引导流程。

import openai
import os


# 推荐：通过环境变量设置 API 密钥以提高安全性。
# openai.api_key = os.getenv("OPENAI_API_KEY")
# 对于本教程，我们使用占位符。请替换为你的实际密钥。
openai.api_key = 'YOUR_API_KEY'


def generate_onboarding_process(product_name, user_type, goals):
    prompt_content = (
        f"Design a detailed onboarding process for {user_type} of our product, '{product_name}'. "
        f"The primary goals for onboarding are: {goals}. "
        f"Please structure the process in phases with actionable steps for each phase. "
        f"Include suggestions for communication touchpoints and key resources to provide."
    )


    messages = [
        {"role": "system", "content": "You are an expert in designing user and employee onboarding processes."},
        {"role": "user", "content": prompt_content}
    ]


    try:
        response = openai.chat.completions.create(
            model="gpt-3.5-turbo", # 或使用更新的模型，如 gpt-4o-mini
            messages=messages,
            max_tokens=1000, # 增加此值以获得更详细的计划
            temperature=0.7,
            n=1,
            stop=None
        )
        return response.choices[0].message.content.strip()
    except Exception as e:
        return f"An error occurred: {e}"


# 示例用法：
product = "FitTrack, a fitness and meal planning mobile app"
audience = "new users"
onboarding_goals = "Users should log their first workout and first meal within 3 days of signing up, and understand how to set weekly fitness goals."


onboarding_plan = generate_onboarding_process(product, audience, onboarding_goals)
print("Generated Onboarding Plan:")
print(onboarding_plan)

在这个 Python 脚本中，generate_onboarding_process 函数根据产品、用户类型和目标构建了一个详细的提示词。然后它调用 OpenAI ChatCompletion API 来生成用户引导计划。

预期输出

Section titled “预期输出”

当你运行脚本时，将收到 LLM 生成的一个全面的用户引导计划。对于示例中的 FitTrack 应用，输出可能看起来像这样（节略版）：

Generated Onboarding Plan:
Here's a detailed onboarding process for new users of 'FitTrack', a fitness and meal planning mobile app, with the primary goals of logging their first workout and meal within 3 days, and understanding how to set weekly fitness goals:


**Phase 1: Welcome & Profile Setup (Day 0-1)**


*   **Actionable Step 1: Personalized Welcome Screen**
    *   Brief, engaging welcome message upon first app open.
    *   Quick prompt for essential profile information (age, weight, height, fitness level, primary goals like weight loss, muscle gain, maintenance).
    *   Communication: In-app message.
    *   Resource: Link to 'Why this info helps us personalize your plan'.


*   **Actionable Step 2: App Tour & Key Features**
    *   Short, skippable interactive tour (3-5 screens) highlighting: Workout Logging, Meal Logging, Goal Setting, Progress Tracking.
    *   Communication: In-app interactive elements.
    *   Resource: 'Quick Start Video' (under 1 minute) accessible from tour.


**Phase 2: First Actions & Core Loop (Day 1-3)**


*   **Actionable Step 3: Guided First Workout Log**
    *   Prompt: 'Ready for your first workout? Let's log it!'
    *   Simplified workout selection (e.g., 'Beginner Full Body', 'Quick Cardio') or easy manual entry.
    *   Celebrate completion with positive reinforcement.
    *   Communication: In-app prompt, push notification reminder if not done by Day 2.
    *   Resource: 'How to Log Workouts Effectively' short article/video.


*   **Actionable Step 4: Guided First Meal Log**
    *   Prompt: 'Time to fuel your body! Log your first meal.'
    *   Easy food search (barcode scanner, large database, quick add calories).
    *   Show immediate impact on daily calorie/macro tracking.
    *   Communication: In-app prompt, push notification reminder.
    *   Resource: 'Tips for Accurate Meal Logging'.


*   **Actionable Step 5: Setting Weekly Fitness Goals**
    *   Guided process to set simple, achievable weekly goals (e.g., '3 workouts this week', 'Log meals for 5 days').
    *   Explain how these contribute to long-term success.
    *   Communication: In-app guided setup.
    *   Resource: 'The Power of Setting Weekly Goals'.


**Phase 3: Habit Formation & Exploration (Day 4-7)**


*   **Actionable Step 6: Introduce Progress Tracking**
    *   Show users where to view their progress (charts, summaries).
    *   Highlight how logged activities contribute to these visuals.
    *   Communication: In-app notification pointing to the progress section.
    *   Resource: 'Understanding Your FitTrack Dashboard'.


*   **Actionable Step 7: Explore More Features**
    *   Gently introduce other relevant features like recipe suggestions, workout plans, or community features (if any).
    *   Communication: Targeted in-app messages or short emails based on initial goals.
    *   Resource: 'Discover More in FitTrack' section in help center.


**Ongoing Support & Engagement:**
*   Regular tips and motivational messages (push notifications/emails).
*   Easy access to FAQs and customer support.
*   Solicit feedback after the first week.

结论

Section titled “结论”

在本章中，我们探讨了如何引导 LLM（如 ChatGPT）设计有效且引人入胜的用户引导流程。通过提供清晰的目标、上下文和期望的结构，你可以利用 LLM 的能力为新用户或新员工生成全面的用户引导计划，从而显著简化设计阶段并整合最佳实践。