每日自动财务高管简报：Google Sheets + GPT-4o-mini + Notion + Telegram

工作日 7 AM 定时触发 → 并行拉取 5 张 Google Sheets → 计算 KPI → GPT-4o-mini 生成 Telegram 格式简报 → Notion 存档 + Telegram 推送 → 收入跌幅超 10% 临界警报 → Slack 错误告警

给谁

需要每日向高管推送财务简报的运营/财务人员，或希望自动化 CFO 看板的小型团队

你会得到

工作日早晨自动收到一份 Telegram 格式的 AI 财务简报，关键指标异常时触发额外警报，执行记录保存至 Notion

你需要

n8n · google-sheets · notion · telegram · slack

步骤

1. 配置定时触发器

   在 n8n 中添加 Schedule Trigger 节点，设置 cron 表达式为 0 7 * * 1-5（周一至周五 7:00 执行）。手动点击「Test Workflow」确认触发器能正常启动。

2. 配置 5 张 Google Sheets 并行拉取

   为 CashFlow、Revenue、Payroll、Subscriptions、Expenses 各添加一个 Google Sheets 节点（typeVersion 4）。在每个节点中：选择或新建 googleSheetsOAuth2Api 凭证；将 documentId 从 YOUR_GOOGLE_SHEET_ID 替换为你的实际表格 ID；确认 sheetName 对应正确的工作表标签。5 个节点均连接到 Weekday 7 AM Trigger 的同一输出，实现并行拉取。

3. 合并所有表格数据

   添加 Merge 节点（typeVersion 3.2），设置 numberInputs 为 5。将 5 个 Google Sheets 节点的输出分别连接到 Merge 节点的 5 个输入端（index 0–4）。确认合并后数据包含各工作表的原始行。

4. 计算 KPI 与状态标签

   添加 Code 节点（typeVersion 2，JavaScript），粘贴以下代码。该节点按 sheet 名称或字段存在性过滤输入项，计算当前/上期收入、薪资、营销 ROI、现金余额、月燃烧率、跑道月数、订阅 MRR、总支出，并输出 GOOD/WARNING/ALERT 状态标签。

   运行提示词 · 来源原文。Calculate KPIs & Status Labels 节点的 JS Code 字段

   复制提示词

   const allItems = $input.all();
   
   const cashflowData = allItems.filter(i => i.json.sheet === 'CashFlow' || i.json.cash_in !== undefined);
   const revenueData = allItems.filter(i => i.json.sheet === 'Revenue' || i.json.revenue_current !== undefined);
   const payrollData = allItems.filter(i => i.json.sheet === 'Payroll' || i.json.payroll_current !== undefined);
   const subscriptionData = allItems.filter(i => i.json.sheet === 'Subscriptions' || i.json.subscription_revenue !== undefined);
   const expensesData = allItems.filter(i => i.json.sheet === 'Expenses' || i.json.expense_amount !== undefined);
   
   const totalRevenueCurrent = revenueData.reduce((sum, i) => sum + (parseFloat(i.json.revenue_current) || 0), 0);
   const totalRevenuePrevious = revenueData.reduce((sum, i) => sum + (parseFloat(i.json.revenue_previous) || 0), 0);
   const revenueChange = totalRevenuePrevious > 0 ? (((totalRevenueCurrent - totalRevenuePrevious) / totalRevenuePrevious) * 100).toFixed(1) : 0;
   
   const totalPayrollCurrent = payrollData.reduce((sum, i) => sum + (parseFloat(i.json.payroll_current) || 0), 0);
   const totalPayrollPrevious = payrollData.reduce((sum, i) => sum + (parseFloat(i.json.payroll_previous) || 0), 0);
   const payrollChange = totalPayrollPrevious > 0 ? (((totalPayrollCurrent - totalPayrollPrevious) / totalPayrollPrevious) * 100).toFixed(1) : 0;
   
   const totalMarketingSpend = expensesData.reduce((sum, i) => sum + (parseFloat(i.json.marketing_spend) || 0), 0);
   const totalMarketingRevenue = revenueData.reduce((sum, i) => sum + (parseFloat(i.json.marketing_attributed_revenue) || 0), 0);
   const marketingROI = totalMarketingSpend > 0 ? ((totalMarketingRevenue / totalMarketingSpend) * 100).toFixed(1) : 0;
   const marketingROIPrevious = 85;
   const marketingROIChange = (marketingROI - marketingROIPrevious).toFixed(1);
   
   const currentCash = cashflowData.reduce((sum, i) => sum + (parseFloat(i.json.cash_balance) || 0), 0);
   const monthlyBurnRate = expensesData.reduce((sum, i) => sum + (parseFloat(i.json.monthly_burn) || 0), 0);
   const cashRunway = monthlyBurnRate > 0 ? (currentCash / monthlyBurnRate).toFixed(1) : 'N/A';
   
   const totalSubscriptions = subscriptionData.reduce((sum, i) => sum + (parseFloat(i.json.subscription_revenue) || 0), 0);
   const totalExpenses = expensesData.reduce((sum, i) => sum + (parseFloat(i.json.expense_amount) || 0), 0);
   
   const today = new Date().toISOString().split('T')[0];
   
   const revenueStatus = revenueChange < -10 ? '🔴 ALERT' : revenueChange < 0 ? '🟡 WARNING' : '🟢 GOOD';
   const payrollStatus = payrollChange > 10 ? '🔴 ALERT' : payrollChange > 5 ? '🟡 WARNING' : '🟢 GOOD';
   const cashRunwayStatus = parseFloat(cashRunway) < 3 ? '🔴 CRITICAL' : parseFloat(cashRunway) < 6 ? '🟡 WARNING' : '🟢 HEALTHY';
   const marketingStatus = marketingROIChange > 0 ? '🟢 IMPROVING' : '🟡 DECLINING';
   
   return [{
     json: {
       date: today,
       revenue_current: totalRevenueCurrent,
       revenue_previous: totalRevenuePrevious,
       revenue_change_pct: parseFloat(revenueChange),
       payroll_current: totalPayrollCurrent,
       payroll_change_pct: parseFloat(payrollChange),
       marketing_roi: parseFloat(marketingROI),
       marketing_roi_change: parseFloat(marketingROIChange),
       cash_balance: currentCash,
       monthly_burn_rate: monthlyBurnRate > 0 ? monthlyBurnRate / allItems.filter(i => i.json.monthly_burn).length : 0,
       cash_runway_months: parseFloat(cashRunway),
       subscriptions_mrr: totalSubscriptions,
       total_expenses: totalExpenses,
       revenue_status: revenueStatus,
       payroll_status: payrollStatus,
       cash_status: cashRunwayStatus,
       marketing_status: marketingStatus
     }
   }];

5. 配置 AI Agent 生成简报

   添加 AI Agent 节点（@n8n/n8n-nodes-langchain.agent，typeVersion 3，promptType: define）。将 OpenAI GPT-4o-mini 子节点（@n8n/n8n-nodes-langchain.lmChatOpenAi，typeVersion 1.3，model: gpt-4o-mini）连接到该 Agent 的 ai_languageModel 输入。在 Agent 节点的 options.systemMessage 和 text 字段分别填入以下提示词。text 字段中的 {{ $json.* }} 变量会自动引用上一步 KPI 节点的输出。

   运行提示词 · 来源原文。Generate AI Outreach Message1 节点的 options.systemMessage 字段

   复制提示词

   You are a senior CFO assistant. Your job is to turn raw financial KPI data into a sharp, scannable executive briefing for Telegram. Rules: use Telegram Markdown (*bold* for labels), include emojis for each section, keep total length under 350 words, be direct and action-oriented — if a metric is at WARNING or ALERT status call out the specific recommended action, otherwise just confirm it is healthy. Never add filler sentences.

   运行提示词 · 来源原文。Generate AI Outreach Message1 节点的 text 字段（用户提示词）

   复制提示词

   =You are an expert CFO assistant. Write a concise daily financial briefing for the executive team based on the data below. Format it for Telegram using *bold* for section labels. Use emojis for quick scanning. Be direct, highlight what changed and what action (if any) is needed. Keep it under 350 words.
   
   📅 Report Date: {{ $json.date }}
   
   💰 REVENUE
   - Current: ${{ $json.revenue_current }}
   - Previous: ${{ $json.revenue_previous }}
   - Change: {{ $json.revenue_change_pct }}%
   - Status: {{ $json.revenue_status }}
   
   👥 PAYROLL
   - Current: ${{ $json.payroll_current }}
   - Change: {{ $json.payroll_change_pct }}%
   - Status: {{ $json.payroll_status }}
   
   📣 MARKETING ROI
   - ROI: {{ $json.marketing_roi }}%
   - Change vs baseline: {{ $json.marketing_roi_change }}%
   - Status: {{ $json.marketing_status }}
   
   🏦 CASH POSITION
   - Balance: ${{ $json.cash_balance }}
   - Monthly Burn: ${{ $json.monthly_burn_rate }}
   - Runway: {{ $json.cash_runway_months }} months
   - Status: {{ $json.cash_status }}
   
   📦 SUBSCRIPTIONS MRR: ${{ $json.subscriptions_mrr }}
   💸 TOTAL EXPENSES: ${{ $json.total_expenses }}
   
   End with a 1–2 line executive summary highlighting the single most important insight today.

6. 合并 AI 简报与财务数据

   添加 Code 节点（typeVersion 2），粘贴以下代码。该节点将 AI Agent 的输出（output）与 KPI 数据合并，生成包含 telegram_report 字段的完整对象，供下游 Notion 和 Telegram 节点使用。

   运行提示词 · 来源原文。Merge AI Report + Financial Data 节点的 JS Code 字段

   复制提示词

   const aiOutput = $('Generate AI Outreach Message1').item.json.output || '';
   const financialData = $('Calculate KPIs & Status Labels').item.json;
   
   return [{
     json: {
       ...financialData,
       telegram_report: aiOutput
     }
   }];

7. 配置推送与存档

   从 Merge AI Report + Financial Data 节点引出 3 条分支： 1) Notion 节点（typeVersion 2，resource: databasePage）：设置 databaseId 为你的 CFO Report 数据库 ID；title 使用 =CFO Report — {{ $json.date }}；blockUi 和 propertiesUi 填入 {{ $json.telegram_report }} 与 {{ $json.marketing_status }}。 2) Telegram 节点（typeVersion 1.1）：设置 chatId 为目标 Chat ID；text 使用 {{ $json.telegram_report }}；additionalFields.parse_mode 设为 Markdown。 3) If 节点（typeVersion 2，Check for Critical Revenue Drop）：条件为 revenue_change_pct < -10（number/lt）。True 分支连接到第二个 Telegram 节点发送临界警报。

   运行提示词 · 来源原文。Send Critical Revenue Alert 节点的 text 字段

   复制提示词

   =🚨 *CRITICAL CFO ALERT — {{ $json.date }}*
   
   ⚠️ Revenue has dropped {{ $json.revenue_change_pct }}% — immediate attention required!
   
   💰 Cash Runway: {{ $json.cash_runway_months }} months
   🔥 Monthly Burn: ${{ $json.monthly_burn_rate }}
   
   Check the full report in Notion now.

8. 配置错误告警

   添加 Error Trigger 节点（typeVersion 1）和 Slack 节点（typeVersion 2.3）。Error Trigger 连接到 Slack 节点的输入。在 Slack 节点中：选择 channel 模式并指定目标频道；authentication 设为 oAuth2；text 当前为占位符 '=Error occured '，建议替换为包含工作流名称与错误详情的模板。该分支独立于主流程，用于捕获任意节点失败。

你正在复现一个 n8n 工作流模板：每日自动财务高管简报。请按以下顺序操作：1) 确认你已拥有 n8n 实例及所有必需凭证（Google Sheets OAuth2、OpenAI、Notion、Telegram Bot、Slack OAuth2）；2) 导入 workflow.json 或按步骤手动搭建；3) 将 YOUR_GOOGLE_SHEET_ID、YOUR_NOTION_DATABASE_ID、YOUR_TELEGRAM_CHAT_ID 替换为实际值；4) 手动触发一次，验证 5 张表数据能正常拉取、KPI 计算无误、AI 简报生成且 Telegram 消息送达；5) 检查收入下跌 10% 的临界警报分支（可通过临时修改测试数据验证）；6) 确认 Slack 错误告警频道能收到通知。如遇凭证权限问题，优先检查 Google Sheets 共享设置与 Notion 集成权限。