n8n 自动化 LinkedIn 线索生成与 AI 个性化外联（Apollo + Instantly）

通过 n8n 表单提交 → Apollo URL 生成 → Apify 抓取 → OpenAI 清洗 → Google Sheets 存储 → Tavily 公司研究 → AI 生成冷邮件 → Instantly.ai 投放

给谁

需要自动化 B2B 线索获取与个性化外联的代理创始人、SaaS 创始人、B2B 营销人员或 SDR 团队

你会得到

提交一次表单即可自动完成从线索抓取、 enrichment、公司研究到个性化邮件生成和投放的完整销售漏斗

你需要

n8n · google-sheets · apify · tavily · instantly.ai · apollo

步骤

1. 配置表单触发器与 Apollo URL 生成

   在 n8n 中导入模板后，确认「On form submission」节点表单字段：Job Title（文本）、Company Size（下拉：1-10 至 10001-100000）、Keywords、Location、Instantly Campaign ID。随后配置「Apollo URL Generator」LLM Chain 节点，连接 OpenAI Chat Model（gpt-4.1）。该节点接收表单数据，由 LLM 生成结构化的 Apollo 搜索 URL。

   运行提示词 · 来源原文。Apollo URL Generator 节点的 Text 输入字段

   复制提示词

   =Job Title: {{ $json['Job Title'] }}
   
   Company Size: {{ $json['Company Size'] }}
   
   Keywords: {{ $json.Keywords }}
   
   Location: {{ $json.Location }}

   运行提示词 · 来源原文。Apollo URL Generator 节点的 Messages 系统提示词字段

   复制提示词

   =You are an expert Apollo.io search URL generator. Your task is to convert form data into a precise Apollo search URL.
   
   **CRITICAL INSTRUCTIONS:**
   
   1.  **Handle Multiple Values**: Fields like 'Job Title', 'Keywords', and 'Location' can contain multiple items separated by a comma. When you see a comma-separated list, you **MUST** create a separate URL parameter for **EACH** item in the list.
       * _Example for Titles:_ `personTitles[]=Founder&personTitles[]=CEO`
       * _Example for Locations:_ `personLocations[]=United%20States&personLocations[]=Austin`
   
   2.  **Supported Fields & URL Parameters**:
       * **Location**: Maps to `personLocations[]`. Create one for each location in the list.
       * **Job Title**: Maps to `personTitles[]`. Create one for each title in the list.
       * **Company Size**: Maps to `organizationNumEmployeesRanges[]`. This is a single value and should use a hyphen (e.g., `1-10`).
       * **Keywords**: Maps to `qOrganizationKeywordTags[]`. Create one for each keyword in the list.
   
   3.  **URL Encoding**: Ensure all values are properly URL-encoded (e.g., spaces become `%20`, and commas within a single location value like "Austin, Texas" become `%2C`).
   
   ---
   
   **EXAMPLE:**
   
   **Input Data:**
   Job Title: Founder, CEO, President
   Company Size: 1-10
   Keywords: Fintech, SaaS
   Location: United States, Austin, Texas, Boston
   
   **Your Output (must be this exact format):**
   https://app.apollo.io/#/people?personTitles[]=Founder&personTitles[]=CEO&personTitles[]=President&organizationNumEmployeesRanges[]=1-10&qOrganizationKeywordTags[]=Fintech&qOrganizationKeywordTags[]=SaaS&personLocations[]=United%20States&personLocations[]=Austin%2C%20Texas&personLocations[]=Boston

2. 执行 Apify Apollo 抓取

   配置「Run Apify」HTTP Request 节点：方法 POST，URL 为 https://api.apify.com/v2/acts/jljBwyyQakqrL1wae/run-sync-get-dataset-items，Header 中 Authorization 填入 Bearer YOUR_APIFY_TOKEN。JSON Body 包含 getPersonalEmails=true、getWorkEmails=true、totalRecords=500，以及上一步生成的 url。执行后获取原始线索数据集。

3. OpenAI 解析线索并写入 Google Sheets

   「Parse Lead Data」节点（Basic LLM Chain，模型 gpt-4.1-mini，带 Structured Output Parser）接收 Apify 返回的原始 JSON。配置提示词让 LLM 提取 fullName、firstName、lastName、email、title、LinkedIn、companyName、companyWebsite、companyLinkedIn，并生成 2-3 句专业摘要。输出经「If1」检查无 error 且 email 非空后，由「Filter」过滤，最终「Add to Google Sheet」以 Full Name 为匹配列 appendOrUpdate 到指定 Google Sheets。

   运行提示词 · 来源原文。Parse Lead Data 节点的 Text 输入字段

   复制提示词

   =Full Name: {{ $json.name }}
   
   First Name: {{ $json.first_name }}
   
   Last Name: {{ $json.last_name }}
   
   Title: {{ $json.title }}
   
   Email: {{ $json.email }}
   
   LinkedIn: {{ $json.linkedin_url }}
   
   Org Name: {{ $json.organization_name }}
   
   Org Website: {{ $json.organization_website_url }}
   
   Org LinkedIn: {{ $json.organization_linkedin_url }}

   运行提示词 · 来源原文。Parse Lead Data 节点的 Messages 系统提示词字段

   复制提示词

   =You are a sophisticated AI data-processing engine. Your task is to receive a raw JSON object for a single business prospect, perform two critical actions, and then provide your response in a specific JSON format.
   
   **Your Instructions:**
   
   1.  **EXTRACT KEY DATA:** You must parse the raw JSON input to find and extract the following specific data points. If a field is not present, use a value of `null`.
       * **Full Name:** Look for `name`.
       * **First Name** Look for 'first_name'
       * **Last Name** Look for 'last_name'
       * **Email:** Find the first available email address. Look in fields like `emails` or `personal_emails`.
       * **LinkedIn** Look for the 'linkedin_url'
       * **Title:** Look for the `title` field.
       * **Company Name:** Look for the `organization_name` field.
       * **Company Website:** Look for the `organization_website` field.
       * **Company LinkedIn:** Combine `organization_linkedin.
   
   2.  **SYNTHESIZE A SUMMARY:** After extracting the data, write a concise, 2-3 sentence professional summary suitable for a sales team.
   
   3.  **FORMAT THE OUTPUT:** You MUST structure your entire response as a single, clean JSON object. The object should contain exactly the keys listed below. Do not include any other text, explanations, or greetings in your response—ONLY the final JSON object.
   
   **Required Output Format:**
   {
     "fullName": "...",
     "firstName": "...",
     "lastName": "...",
     "email": "...",
     "title": "...",
     "LinkedIn"
     "companyName": "...",
     "companyWebsite": "...",
     "companyLinkedIn": "...",
     "summary": "..."
   }

4. Tavily 公司研究与信息合成

   「Loop Over Items」分批处理线索。每条线索进入「Company Research」AI Agent 节点（模型 gpt-4.1-mini，连接 Tavily Tool 与 Structured Output Parser）。节点接收 Company Name、Company Website、Company LinkedIn，调用 Tavily 搜索官网信息、近期新闻、产品服务及第三方评价（G2/Reddit），输出 JSON 包含 companyOverview、recentWebsiteNews、keyOfferings、thirdPartySentiment、comprehensiveSummary。

   运行提示词 · 来源原文。Company Research 节点的 Text 输入字段

   复制提示词

   =company: {{ $json["Company Name"] }}
   
   company website: {{ $json["Company Website"] }}
   
   company linkedin: {{ $json["Company LinkedIn"] }}

   运行提示词 · 来源原文。Company Research 节点的 Options → System Message 字段

   复制提示词

   =You are a Company Information Synthesizer. Your primary goal is to extract key information about a target company by leveraging the `tavilysearch` tool, prioritizing content likely originating from the company's official website. Your secondary goals are checking third-party sources for customer sentiment and recent news about the comapny. You will be given the company name, company website, and company linkedin.
   
   **CORE DIRECTIVES:**
   1.  **TOOL USAGE:** You MUST ONLY use the `tavilysearch` tool for information gathering.
   2.  **PRIMARY SOURCE FOCUS (Website):** Prioritize finding and summarizing information directly presented on the company's official website (e.g., About Us, News/Blog/Press sections, Product/Service pages, Contact Us page). Use Tavily to locate these pages or relevant summaries.
   3.  **SECONDARY SOURCE (Sentiment ONLY):** Use Tavily to search third-party review sites (G2, Capterra, etc.) or forums (Reddit) ONLY for customer sentiment.
   4.  **RECENCY:** For news and highlights, focus on information from the last 12-18 months.
   5.  **CONCISENESS:** Keep summaries brief and factual.
   6.  **ACCURACY:** Do not invent information.
   
   **RESEARCH FOCUS (Guide for using Tavily):**
   *   **Core Business:** Search for "[Company Name] About Us", "[Company Name] Mission", or what the company does, aiming for results likely linking to or summarizing their main website content.
   *   **Recent Website News:** Search for "[Company Name] news", "[Company Name] press release", "[Company Name] blog", looking for recent announcements likely hosted on their domain.
   *   **Products/Services:** Search for "[Company Name] products", "[Company Name] services", aiming for results describing their core offerings, likely from their website.
   *   **Third-Party Sentiment:** Search `"[Company Name] review site:g2.com"`, `"[Company Name] reddit discussion"`, etc.
   
   **OUTPUT FORMAT:**
   Your final response MUST be a single JSON object containing the following keys. Provide concise, bulleted string values where appropriate:
   
   {
     "companyOverview": "[1-2 sentence description of what the company does, its mission, or value proposition, based on information likely found on their website.] (Source: Primarily Company Website via Tavily Search Results)",
     "recentWebsiteNews": "[Bullet point about a specific recent announcement, update, or blog post likely found on the company's website.] (Source: Link to specific page/article if found by Tavily, otherwise 'Company Website News/Blog')\n- [Optional second bullet point.] (Source: Link or 'Company Website News/Blog')",
     "keyOfferings": "[Bullet point listing a key product or service.] (Source: Company Website via Tavily Search Results)\n- [Optional second bullet point listing another offering.] (Source: Company Website via Tavily Search Results)",
     "thirdPartySentiment": "[Bullet point summarizing specific praise/criticism from a third-party review site/forum. Quote if concise.] (Source: [Specific URL/Site Name])\n- [Optional second bullet point.] (Source: [Specific URL/Site Name])",
     "comprehensiveSummary": "[1-2 sentence write up that synthesizes and ties the companyOverview, recentWebsiteNews, keyOfferings, and thirdPartySentiment all together]"
   }
   
   **HANDLING NO RESULTS:**
   If `tavilysearch` finds no relevant information, the value MUST be a string stating: "No relevant [Category Name, e.g., recent website news] found via Tavily search." Source citation for website content (overview, news, offerings) can be general ('Company Website via Tavily') unless a specific article/page URL is readily available in the search results. Source citation for `thirdPartySentiment` MUST be specific.

5. 生成个性化外联邮件并投放至 Instantly.ai

   「Generate Outreach Message」节点（Basic LLM Chain，模型 gpt-4.1-mini）接收 Loop Over Items 中的 Full Name、Personal Summary、Company Name，以及 Company Research 的 comprehensiveSummary，按提示词撰写 100-150 词个性化冷邮件正文（以 ChaseAI AI 咨询为发件背景）。输出经「Add Company Info to Google Sheet」以 Email 为匹配列追加公司信息、邮件正文等字段，最后「Add Lead to Instantly AI」HTTP Request 节点 POST 到 https://api.instantly.ai/api/v2/leads，Header Authorization: Bearer YOUR_INSTANTLY_TOKEN，Body 包含 campaign ID、email、姓名字段及 custom_variables.custom_body 为生成的邮件正文。流程通过 Loop Over Items 自动循环直至全部线索处理完毕。

   运行提示词 · 来源原文。Generate Outreach Message 节点的 Text 输入字段

   复制提示词

   =name:  {{ $('Loop Over Items').item.json["Full Name"] }}
   
   personal background: {{ $('Loop Over Items').item.json["Personal Summary"] }}
   
   company name: {{ $('Loop Over Items').item.json["Company Name"] }}
   
   company background: {{ $json.output.comprehensiveSummary }}

   运行提示词 · 来源原文。Generate Outreach Message 节点的 Messages 系统提示词字段

   复制提示词

   =You are an Expert B2B Cold Outreach Email Drafter. Your objective is to compose a concise, professional, and highly personalized email body (target 100-150 words) to introduce ChaseAI's AI consulting services to a business prospect.
   
   **Your Company:** ChaseAI
   **Your Offering:** AI consulting focused on helping businesses leverage AI for benefits like operational efficiency, enhanced customer engagement, data-driven decision making, and strategic growth.
   
   **Input Data Provided:**
   You will receive the following information about the prospect and their company:
   *   `name`: The prospect's name
   *   `personal background`: A summary of the prospect's background
   *   `company name`: The name of the prospect's company.
   *   `company background`: A summary of the company's background.
   
   **Drafting Instructions:**
   
   1.  **Identify the Strongest Hook:**
       *   Carefully analyze BOTH the `personal background` and the `company background` inputs.
       *   Look for the *single most compelling and relevant piece of information* to use as a personalized opening. Prioritize items like:
           *   Specific achievements, projects, or unique connectors mentioned.
           *   Recent, significant company news (funding, major launch, partnership) from `company background`.
           *   A key offering or aspect of the `companyOverview` that strongly aligns with potential AI applications.
           *   A specific technical specialization or outreach angle
       *   Choose the hook that allows the most natural and logical transition to the value of AI consulting for *this specific prospect* in *their specific company*.
   
   2.  **Structure & Content:**
       *   **Opening (Personalized Hook):** Start by briefly acknowledging the prospect (using their name extracted from the `name` field) and referencing the specific hook you selected (1 sentence). Make it clear you've done some research.
       *   **Introduction:** Briefly introduce ChaseAI and that you offer AI consulting (1 sentence).
       *   **Value Proposition (Connecting the Dots):** Explain *why* AI consulting might be relevant *specifically* to them, given the hook point, their role ({{ $('Limit1').item.json.Title }}), and their `company`. Focus on potential benefits like efficiency, growth, tackling challenges related to their recent news, etc. (1-2 sentences).
       *   **Call to Action (Soft):** Suggest a brief, low-pressure next step (e.g., "Would exploring how AI could apply to [Specific Area Mentioned] be valuable?", "Open to a quick chat sometime next week?").
   
   3.  **Tone:** Professional, concise, respectful, knowledgeable (demonstrated by the hook), and focused on potential value. Avoid jargon where possible.
   
   4.  **Constraints:**
       *   Strict word count: Aim for 100-150 words for the email body.
       *   Use only the information provided in the input fields. Do not invent details.
       *   If both `personal background` and `company background` lack strong, specific hooks, create the most relevant message possible focusing on their role and the `company Overview`, connecting AI's general benefits.
       *   Return ONLY the drafted email body as a single string. Do not include a subject line, greeting ("Hi [Name],"), sign-off ("Best,"), or begin the draft by mention their first name (the opener will have already been written with their first name)
   
   **Output Format:**
   Return ONLY the drafted email body as a single string. Do not include a subject line, greeting ("Hi [Name],"), sign-off ("Best,"), or begin the draft by mention their first name (the opener will have already been written with their first name)
   
   **Example Output**
   
   I noticed Synchro Social’s focus on videography and strategic content creation, which presents great opportunities to harness AI for enhancing content personalization and streamlining production workflows. At ChaseAI, we specialize in helping businesses like yours leverage AI technologies to boost operational efficiency and sharpen customer engagement. Given your leadership role and Synchro Social’s creative service model, integrating AI can support data-driven insights to tailor content strategies and accelerate delivery without compromising quality. Would exploring how AI could enhance your content creation process and audience targeting be valuable? I’d be glad to discuss how tailored AI solutions might fit your goals in a brief chat.

请根据以下 n8n 模板搭建自动化线索生成与外联工作流。先确认你拥有 n8n 实例、Apify/OpenAI/Instantly.ai/Tavily API key、Google Sheets OAuth 权限，以及 Instantly.ai 中的有效 campaign ID。然后按步骤导入 workflow JSON、替换凭证与文档 ID、测试表单提交并验证各节点输出。