LinkedIn Lead Generation with GPT-4o, Apify, Google Sheets, and Phantombuster

Audience description → GPT-4o generates Apollo search URL → Apify scrapes leads → GPT-4o writes icebreakers → Write to Google Sheets → Optionally trigger Phantombuster connection requests

Who

Sales/SDR teams, indie hackers, or anyone automating LinkedIn lead sourcing and initial outreach

You get

Input audience description via form, auto-generate Apollo search URL, scrape lead data, write personalized icebreakers, log to Google Sheets, and optionally auto-trigger Phantombuster to send LinkedIn connection requests

You need

n8n · openai · http-request · google-sheets

Steps

1. Collect audience description via form

   Configure the Form Trigger node in n8n. Form title is 'Audience', with one required field 'descritpion of the company ' (source spelling), placeholder prompting for company location, size, industry, etc. Publish the form URL; users or SDRs submit target audience descriptions through it.

2. Generate Apollo search URL with GPT-4o

   Configure an OpenAI node, select GPT-4o model, enable JSON output. System prompt sets a sales assistant role; user prompt includes an Apollo URL template example, explains modifiable fields (location, titles, keywords), and requests output in {"SearchUrl":"..."} format. Input variable is the form-submitted 'descritpion of the company '.

   Execution prompt · Source text。OpenAI node system message (node: Genrating appolo Url for apify to scrap)

   Copy prompt

   you are a helpful intelligent sales assitant

   Execution prompt · Source text。OpenAI node user message template (node: Genrating appolo Url for apify to scrap)

   Copy prompt

   your task is to take as input of a prospect audeince and turn that into a apollo serach URL here is an example of the URL 
   https://app.apollo.io/#/people?personLocations[]=United%20States&includedOrganizationKeywordFields[]=tags&includedOrganizationKeywordFields[]=name&sortAscending=false&sortByField=recommendations_score&qKeywords=marketing%20agency%20&page=1&personTitles[]=ceo%20and%20founder&personTitles[]=co-founder
   This url describes a search for peolpe that are 
   -located in the united state 
   - Holds the titles: CEO,founder,co founder 
   - have a keyword assosiated with "marketing "agency"
   you can only make changes to these fields only 
   return your response as using this format  
   {"SearchUrl":"SearchUrlgoeshere"}

   Execution prompt · Source text。OpenAI node dynamic input expression

   Copy prompt

   ={{ $json['descritpion of the company '] }}

3. Scrape leads via Apify Actor

   Configure an HTTP Request node, method POST, URL fixed at https://api.apify.com/v2/acts/jljBwyyQakqrL1wae/run-sync-get-dataset-items. JSON body includes cleanOutput: true and a url field referencing the SearchUrl from the previous step. Headers must include Accept: application/json and Authorization: Bearer YOUR_TOKEN_HERE key (replace YOUR_TOKEN_HERE key with actual Apify API token).

   Execution prompt · Source text。HTTP Request node JSON body (node: Run apify actor to scrap the proscpect)

   Copy prompt

   ={
       "cleanOutput": true,
       "url": "{{ $json.message.content.SearchUrl }}"
   }

   Execution prompt · Source text。HTTP Request node Authorization header value (placeholder to be replaced)

   Copy prompt

   Bearer YOUR_TOKEN_HERE key

4. Generate personalized icebreakers with GPT-4o

   Configure a second OpenAI node, same GPT-4o model. System prompt is 'you are a helpful '; user prompt requests an ultra-short, punchy icebreaker based on LinkedIn profile data, specifying template format 'Hey {{name}}, loved seeing {{thingaboutThem}}, Im also into {{plausibleTieIN}}, thought Id connect.' Emphasizes paraphrasing (not direct quoting from LinkedIn fields), avoiding clichés, and maintaining a laconic, spartan style. Output format: {"Icebreaker":"..."}. Input is lead data returned by Apify.

   Execution prompt · Source text。OpenAI node system message (node: Genrate ice breaker by scraping linkedin data)

   Copy prompt

   you are a helpful

   Execution prompt · Source text。OpenAI node user message template (node: Genrate ice breaker by scraping linkedin data)

   Copy prompt

   your task is to input a bunch of linkedin profile information of a user, and then geenrate a very short,very punchy incebreaker that i can use as a variable in my connection request 
   return your output in this format 
   {"Icebreaker":"Icebreakergoeshere"}
   in order to ensure icebreakers are punchy and high quality, make them follow this template 
   Hey {{name}}, loved seeing {{thingaboutThem}}, Im also into {{plausibleTieIN}}, thought Id connect.
   For {{thingaboutThem}}and {{plausibleTiein}}, never use the exact information provided in a linked in field. instead,always parapharse. this makes it seems human written instead of just an automated message, Also, make it super short. Dont say stuff like ""passionate about truning vision into reality " or anything like that be extremely Laconic and spartan.

   Execution prompt · Source text。OpenAI node dynamic input expression

   Copy prompt

   map the prospect's LinkedIn or details from the previous node.

5. Append icebreakers to Google Sheets

   Configure a Google Sheets node, operation appendOrUpdate. Pre-create a Google Sheet with columns: name, linkedin url, company website, description, and other columns generated by the Apify node. Map corresponding fields in the node; documentId and sheetName must be filled with actual sheet details (left empty in source, user-configured). Ensure Google Service Account credentials are shared with the sheet.

   Setup guidance · Source text。Sticky Note guidance for Google Sheets column setup

   Copy prompt

   Create a google sheet add relevant colums name,linkedin url,company website,description and oter coloums that are genrated by the apify node and map them accordingly

6. Aggregate data batch

   Configure an Aggregate node, select aggregateAllItemData. This node aggregates multi-row Google Sheets output into batched format, preparing for the Phantombuster trigger. No additional prompts needed.

7. Trigger Phantombuster to send connection requests (optional)

   Configure an HTTP Request node, method POST, URL https://api.phantombuster.com/api/v2/(your agent id goes here)/launch, replacing (your agent id goes here) with actual Agent ID. Header: X-Phantombuster-Key: api key goes here (replace with actual API key). No request body. Disable this node if auto-sending connection requests is not needed.

   Setup guidance · Source text。Sticky Note guidance for Phantombuster HTTP Request configuration

   Copy prompt

   Triggers the phantom buster a platform that sends linkedin connection requests. to your leads, you have to go to the api credentials to get the relevant url but i hvae harcoded one for the connection request and add your api key to the header, and the agent ID of the agent you are running will be available in the url of the agent in phantombuster thats it no body  is needed.

You are an n8n workflow setup assistant. The user wants to reproduce a LinkedIn lead generation automation. Help them: 1) Confirm all required credentials (OpenAI, Apify, Google Sheets, Phantombuster); 2) Import workflow.json and configure each node step-by-step; 3) Replace all placeholders (YOUR_TOKEN_HERE key, your agent id goes here, documentId/sheetName); 4) Run one end-to-end test with sample data. Do not execute actual API calls or send real connection requests on behalf of the user.