This playbook walks you through setting up a comprehensive system for LinkedIn outreach using n8n, Apollo, Apify, OpenAI (GPT-4), Google Sheets, and PhantomBuster.
1. Define Your Audience and Generate an Apollo Search URL
Start with an n8n Form Input: Your system kicks off with an n8n form where you describe your target audience in plain English. For example, you might type something like, “I want all creative agencies in the United States with company sizes between 1 to 1,000”.
AI Generates Apollo Search URL: This natural language description is then fed into an AI model, specifically GPT-4, which autonomously generates a specific Apollo search URL. This URL is crucial because it allows the system to define the exact search parameters within Apollo’s database.
2. Scrape Leads using Apify
Utilise Apify as a Scraper: Instead of directly exporting leads from Apollo (which can be very expensive), you’ll use Apify, a tool that acts as a scraper. Apify allows you to plug in the generated Apollo URL, and it then scrapes the HTML of the page to find and extract leads. This is a common “hack” currently used because Apollo itself scrapes LinkedIn Sales Navigator.
Extract Detailed Profile Data: Apify identifies individuals from your defined audience and extracts valuable information such as their email addresses (work and personal) and other detailed profile data completely autonomously. This also includes crucial information like their LinkedIn profile URLs.
Cost Efficiency: Scraping via Apify is cost-effective, with an example cost of about $1.20 per 1,000 leads, providing enough leads for a month of LinkedIn outreach activities.
Manual Testing First: It’s highly recommended to manually verify that the scraping process works as expected before building out the full automated system in n8n. This ensures the “use case” works before automating it.
3. Enrich Leads with Personalisation using AI
Generate Customised Icebreakers: The scraped lead data is then passed to an AI model (GPT-4) to create customised “icebreakers” for your connection requests. The AI’s task is to take LinkedIn profile information (like name, job title, company, and past employment history) and generate a very short, punchy snippet.
Adhering to LinkedIn Limits: It’s vital to note that LinkedIn connection requests have a character limit of approximately 300 characters (around 42-75 words), so the personalisation snippet needs to be concise.
Human-Like Touch: The AI is specifically prompted to paraphrase information from LinkedIn fields rather than using exact variables. This makes the message appear human-written, avoiding the robotic “Hello dear person I would like to sell you stuff” tone. The aim is to create messages that feel customised, like “Hey Peter, saw that you went to the U of A, that’s super cool, I love that you did X Y and Z and this may be totally out of left field but I thought we should connect”. This significantly increases the likelihood of a higher conversion rate for connection requests.
4. Dump Data to a Google Sheet
Centralised Data Storage: All the aggregated data, including the unique LinkedIn ID, first name, last name, full name, LinkedIn URL, title, email status, photo URL, and the newly generated Icebreaker, is then logged into a Google Sheet database.
Mapping Columns: You’ll append rows to your Google Sheet, manually mapping the data from the previous steps to specific columns you’ve set up, such as ‘LinkedIn URL’ and ‘Icebreaker’.
5. Trigger LinkedIn DMs via PhantomBuster
Automated Outreach Tool: PhantomBuster is the tool used to physically produce and send the LinkedIn connection requests and messages. It’s designed to simulate real human activity to avoid detection as a bot.
Input from Google Sheet: PhantomBuster takes your Google Sheet URL as input, allowing it to pull the necessary data for sending messages.
Message Customisation: Within PhantomBuster, you define the message template, pulling in variables like first name or, more importantly, the Icebreaker generated by AI.
Connecting Your LinkedIn Account: You’ll connect your LinkedIn account using a PhantomBuster Chrome extension. Be aware that sending customised connection requests might require a LinkedIn Sales Navigator subscription for connections beyond your second-degree network.
API Trigger: The PhantomBuster agent is triggered via an API call (HTTP POST request) from n8n. This allows the entire outreach process to be automated from your n8n workflow.
Sending Limits: LinkedIn imposes a message limit of about 100 outbound connection requests per week, so the system should be configured to send small batches (e.g., 5-10 requests per launch).
6. Tracking and Monitoring
Google Sheet Status Tracking: While the direct web hook for detailed completion status from PhantomBuster can be tricky to implement, the Google Sheet serves as a primary record of individuals processed by the system.
PhantomBuster Deduplication: PhantomBuster automatically handles deduplication, meaning that if a person is already on your sheet and has been processed, it won’t send them another request. You can also check the results directly within PhantomBuster to see a list of everyone you’ve sent messages to.
Continuous Improvement
- Start Small: Begin with small batches (5 to 10 requests per day) and gradually increase as you get more comfortable.
- Respond Quickly: Regularly check your LinkedIn account and respond to replies promptly. Responding within a minute can significantly boost your conversion rate (by around 400%).
- Refresh Templates: Update your message templates every couple of weeks to keep them fresh and effective.
- Monitor Metrics: Continuously monitor your campaign metrics to understand what’s working and iterate your approach, as cold outreach is largely about iteration.
This complete AI-powered LinkedIn Outreach system, built entirely in n8n, is designed to save you hours of manual work, improve response rates, and offer a level of personalisation that often surpasses other automated tools. Remember, always test each manual step thoroughly before automating it to avoid issues down the line.