feat: Educational CRM conversion rate improvement system

[Copilot]

The repo lacked any implementation addressing the education-industry CRM conversion rate strategies discussed in context — lead lifecycle management, follow-up automation, trial-class funnel tracking, and parent-school communication.

What's added: crm.html

A fully self-contained single-page CRM demo (pure HTML/CSS/JS, zero dependencies) covering all key scenarios:

🗂 Lead Management (线索管理)

• Kanban board across 5 stages: 新线索 → 已接触 → 已邀约 → 已试听 → 已签单
• Multi-channel source tags, hot/warm/cold scoring, real-time search + channel filter
• Click any lead card to advance its stage; manual lead entry via prompt

📋 Follow-up Reminders (跟进提醒)

• Tasks bucketed by urgency: overdue / due-today / done — color-coded with border accent
• One-click mark-done; batch weekly plan generation

🎯 Trial Class Funnel (试听转化)

• 5-step visual funnel with per-step conversion rate and drop-off counts
• Key metric progress bars (到访率, 成单率, 满意度)
• Toggleable automation rules (e.g. "48h post-trial no-convert alert")
• Trial schedule table with confirm/feedback actions

💬 Parent Communication (家校沟通)

• Chat interface with contact list, unread indicators, and message bubbles
• Enter to send; each send triggers a "synced to CRM" toast
• Bulk notification and archive-sync buttons

📈 Analytics Dashboard (数据看板)

• KPIs: overall conversion rate, avg response time, trial→sign rate, referral share
• Channel source bar chart, sales advisor conversion leaderboard
• Automation ROI table (before/after metrics per rule), 7-day stacked trend chart

Code quality fixes (from review)

• Magic baseline count extracted to named constant BASELINE_LEAD_COUNT
• Negative margin-top stacked bar technique annotated with explanatory comment

Original prompt

请为基于教育行业 CRM 的产品明确给出提升“转化率”的策略和实现方法，包括产品功能、销售流程、自动化工具等方面的优化方案。重点阐述线索管理、沟通跟进、试听课转化、家校沟通等教育行业关键场景，并结合CRM产品落地实现提建议。

The following is the prior conversation context from the user's chat exploration (may be truncated):

User: If you were asked to write a CRM system, what tech stack would you choose?

Assistant: If I were building a CRM (Customer Relationship Management) system from scratch today, my technology choices would prioritize data integrity, scalability, and developer velocity. CRMs handle complex relational data (contacts, companies, deals, tasks) and require high-performance search and background processing.

Here is the modern tech stack I would choose, broken down by component:

1. The Core Stack: TypeScript (End-to-End)

Using TypeScript across the entire stack allows for shared types, faster iteration, and fewer bugs.

• Frontend Framework: Next.js (React)

  • Why: Excellent for building complex dashboards. You get a mix of server-side rendering for fast initial loads and client-side interactivity.
  • UI/Styling: Tailwind CSS combined with shadcn/ui or MUI. These provide highly customizable, accessible building blocks for the data-dense tables, forms, and modals a CRM needs.
  • State Management: React Query (TanStack Query) for data fetching and caching, combined with Zustand for lightweight global state.

• Backend Framework: NestJS (Node.js)

  • Why: NestJS provides a highly opinionated, scalable, and enterprise-ready architecture out of the box. It uses decorators and dependency injection, which is perfect for maintaining the complex business logic, role-based access control (RBAC), and modularity a CRM requires.
  • Alternative: If the team prefers Python, Django is phenomenal for CRMs due to its built-in ORM, admin panel, and rapid development capabilities.

2. Database & Data Layer

• Primary Database: PostgreSQL
  • Why: A CRM is inherently relational (A User owns a Deal, which is tied to a Company, which has many Contacts). Postgres handles this beautifully and supports JSONB for when you need schema-less "custom fields" for users to add their own attributes to contacts.
• ORM: Prisma or Drizzle ORM (if using TypeScript) for type-safe database queries.

3. Search & Caching

• Search Engine: Meilisearch or Elasticsearch
  • Why: Users expect instantaneous global search across thousands of contacts, notes, and emails. Querying Postgres directly for this is too slow. Meilisearch provides a typo-tolerant, blazing-fast search experience.
• Caching & In-Memory Store: Redis
  • Why: Essential for caching frequent queries, managing user sessions, and rate-limiting APIs.

4. Background Jobs & Queues

• Queueing System: BullMQ (Node.js) + Redis
  • Why: CRMs rely heavily on async tasks: importing massive CSVs of contacts, sending bulk email campaigns, syncing with Google Calendar, and triggering webhooks. These must be processed in the background to keep the UI responsive.

5. Authentication & Security

• Auth Provider: Clerk or Auth0
  • Why: Building auth from scratch is risky. CRMs often sell to B2B clients who eventually demand features like SAML, SSO, and multi-factor authentication. These providers handle complex B2B multi-tenant logic effortlessly.

6. Infrastructure & Deployment

• Hosting: AWS (ECS or EKS for containers, RDS for Postgres) or Render / Railway for faster, PaaS-style deployments in the early stages.
• CI/CD: GitHub Actions for automated testing, linting, and container builds.
• Monitoring: Sentry for error tracking and Datadog for performance monitoring.

Summary

This Next.js + NestJS + PostgreSQL + Redis stack strikes the perfect balance. It is robust enough to handle the complex relational data of a CRM, modern enough to attract top developer talent, and scalable enough to grow from a small startup to an enterprise SaaS.

User: 我开发出来的CRM系统，你觉得开源好还是闭源好

Assistant: 开发 CRM 系统，选择开源（Open Source）还是闭源（Closed Source），这是一个非常核心的商业策略问题。没有绝对的好坏，主要取决于你的商业目标、目标客户群体、团队资源以及你希望如何变现。

针对 CRM 这种对“数据隐私”和“定制化”要求极高的系统，我为你梳理了两种模式的利弊，以及目前最流行的第三条路：

1. 开源模式（Open Source）

开源通常意味着你将代码公开在 GitHub 上，允许任何人下载、修改和自行部署。

👍 优势：

• 极低的早期获客成本： 开发者和中小型企业（SMB）非常喜欢开源工具。你可以通过在 GitHub 获取 Star、在开发者社区（如 Hacker News, V2EX）曝光，快速获得第一批种子用户。
• 建立信任（数据安全）： CRM 存储着企业的核心命脉——客户数据。许多企业不信任小公司的 SaaS 产品，担心数据泄露或被“卡脖子”。开源意味着代码透明，企业可以私有化部署（Self-host），这能极大地消除信任门槛。
• 社区生态建设： CRM 通常需要对接各种外部系统（邮件、ERP、支付网关等）。开源可以吸引社区开发者为你贡献插件和集成，帮你丰富生态。

👎 劣势：

• 变现困难： 大量用户会选择免费“白嫖”自行部署，不愿意付费。
• 竞争风险： 其他公司（甚至是云厂商）可能会直接拿着你的开源代码，打包成他们自己的服务去卖钱。
• 维护心智成本高： 你需要花费大量时间处理社区的 Issues、PR，以及解答各种千奇百怪的部署问题，这可能会分散你做核心业务的精力。

2. 闭源模式（传统 SaaS / Proprietary）

代码保密，用户只能通过网页登录使用你提供的云服务，按月/按年付费。这比如现在的 Salesforce 或 HubSpot。

👍 优势：

• 商业模式极其清晰： 纯粹的 SaaS 模式，控制权完全在你手里。只要产品能解决痛点，客户就会持续...

This pull request was created from Copilot chat.