Skip to main content

LeenLabs

email outreach - how to qualify sales leads
Screenshot: a walkthrough of the n8n workflow with all nodes set up.

 

Affiliate disclosure: some links in this article may be affiliate links. If you sign up through them, we may earn a commission at no extra cost to you. See our Terms of Service.

Picture a small agency with a list of leads in a table. Every morning someone opens it, picks the people who have not been contacted, copies the same email, types in a first name and a company, hits send, and then types today’s date in a column so nobody gets the same message twice. It takes an hour of dull work, and the day it is skipped, the list goes cold. This free n8n workflow does that routine on a timer. At 10:30 every day it reads the table, keeps the leads that have never been contacted, takes the first 15, asks an AI model to put the right name and company into your email template, sends the message by SMTP, and writes the date of contact and a follow-up date back into the table.

It is a short workflow, and the idea behind it is easy to follow. It is also a workflow that deserves a careful read before you let it near real people. It sends real emails with no pause between them. It sends all of them before it records any of them. It has no way to stop a bad address from blocking the queue. And it writes a follow-up date but never sends a follow-up. In this guide we explain every step in plain English, work through what happens when things go wrong, list what we changed in the file and why, and say what we could not test.

Read this first: this workflow sends real emails

  • Sending comes before saving. As far as we know, n8n runs each step on all the leads before moving to the next step. All the emails of a run go out first and the table is updated afterwards. If one send fails, the run stops and the leads already emailed are not marked, so they can be emailed again at the next run.
  • The AI rewrites your email. It is told to change only the name and the company, but nothing checks what it returns. Whatever text comes back is what gets sent.
  • The file contains placeholders. If you forget to replace the sample text in the template, the subject line or the sender address, that is what your leads will receive or what will fail.
  • There is no unsubscribe link, no bounce handling and no reply handling.
  • No follow-up is sent. Only a date is written in the table.

Start with your own email address as the only lead. This guide is not legal advice.

Template source: this workflow comes from the n8n Library at n8n.io, the public collection of workflow templates. We did not create it. LeenLabs prepared its JSON for publication and documented it. What we changed is listed in the section What we changed in the original template.

At a glance

  • What it does: sends a first-contact email to leads that have not been contacted, personalized with name and company, and records the date.
  • Apps used: n8n, NocoDB (a table database), Groq (an AI service) and any email account that offers SMTP.
  • Trigger: a daily schedule, set to 10:30 in the file.
  • Setup effort: medium. A table with six columns, three accounts, and your own email template.
  • Cost to run: the file is free. The AI service and your email provider have their own terms, so check each provider’s current page.
  • Watch out for: duplicates after a failed run, a template that was never replaced, no unsubscribe link, and a follow-up date that triggers nothing.
  • Test status: the file was checked for structure, and the references between nodes were checked by script. It was not imported into n8n and no service was contacted. Not tested on live APIs.

In this guide

  1. What is email outreach, and what does “personalized” mean here?
  2. A daily routine in four moves: what the workflow does
  3. Who should be on the list: sales leads and qualification
  4. Is the AI step needed? A plain look at the personalization
  5. Words you will see in this guide
  6. Apps and accounts this workflow talks to
  7. What to have ready before you import anything
  8. Where it can go wrong: the order of events, with an example
  9. The big picture: from table to inbox and back
  10. Follow the data: what each of the 9 steps does
  11. Get the workflow JSON
  12. Install it in eleven steps
  13. What it will not do: limits, costs and trade-offs
  14. Access, privacy and responsible use
  15. Making the email read like a person wrote it
  16. Problems you may meet and how to fix them
  17. Too much of a puzzle? Let us build or adapt it for you
  18. What we changed in the original template
  19. What was verified, and what was not
  20. Four ways to adapt the workflow
  21. Frequently asked questions
  22. Tools used in this workflow
  23. Suggested reading
  24. Get more advanced automation and AI resources for business
  25. Keep exploring

What is email outreach, and what does “personalized” mean here?

Email outreach means writing to people who have not asked to hear from you, to start a conversation, offer a service or find out whether there is interest. An outreach email is one of those first messages. The word personalized can mean very different things. At one end, a message that quotes something specific from the reader’s own work. At the other end, a message that only swaps in a first name and a company name, like a form letter.

This workflow sits at the second end. It personalizes two things only: the name and the company. Everything else in your email is the same for every lead. That is not a flaw, but you should know it before you promise yourself better replies. If you want a first line that says something real about each prospect, look at our outreach email openers with Perplexity and Claude or our cold email icebreakers with Apify and GPT. This one is about the routine around the email: who gets it, when, and how the table remembers.

A daily routine in four moves: what the workflow does

  1. Find. It reads every row of your table and keeps the ones with no value in Initial Contact Date. Those are your uncontacted leads.
  2. Cap. It keeps the first 15 of them. The number is a setting you can change.
  3. Personalize and send. For each lead, an AI model fills the name and the company into your template, and the result is sent by SMTP.
  4. Remember. It writes today’s date in Initial Contact Date and a date three days later in Next Follow up/Contact, so the lead is skipped from then on.

It suits a person who keeps leads in a table, has one email to send to all of them, and wants a limited number to go out each day. It does not suit you if you need unsubscribe handling, reply detection, several steps in a sequence, or checks on the email addresses. For the sending side with a follow-up and a reply check, see our outbound outreach workflow.

Who should be on the list: sales leads and qualification

The workflow contacts everyone in the table who has no contact date. It does not judge whether a person is a fit. That judgment is yours, and it matters more than any setting in this file. A list of B2B sales leads bought or exported in bulk contains many people who will never buy, and every wrong email costs a little of your sender reputation. If you are not sure how to qualify sales leads before you send, our B2B lead qualification framework for solo sellers gives five questions you can apply to a list in an afternoon. If you do not have a list at all, our workflow on generating sales leads and outreach emails starts from companies and finds a contact at each.

A simple way to use the two together: add a column of your own, such as Qualified, fill it by hand for a small batch, and change the Filter node so that it also requires that column to say yes. We did not test that change, but it uses the same kind of condition the Filter already has.

Is the AI step needed? A plain look at the personalization

The AI in this workflow is asked to replace a name and a company in a text. A simple text replacement can do the same, with no AI, no cost per lead and no risk that the model changes a sentence. In n8n, a node that sets fields (called Edit Fields, or Set) can build the email text with two expressions, one for the first name and one for the company name. We say this plainly because the template presents AI personalization as a feature, and for this particular job it is a choice, not a necessity.

Why would you keep the AI anyway? There are reasons. It can cope with a company name that needs a small grammatical change, and it can adapt when your template does not have clean placeholders. The cost of that flexibility is that the text that leaves your mailbox is produced by a model, not by you. The prompt asks the model to change nothing else, but models do not always obey, and the Send email node does not check the result. If you do keep the AI, read the first emails of every new batch. If you do not need the flexibility, our adaptation section shows how to replace the step.

Words you will see in this guide

Word What it means here
Workflow A chain of steps in n8n. Each step is a box called a node.
Node One step of the workflow, such as “read the table” or “send an email”.
Item One piece of data traveling through the nodes. Here, one item is one lead.
Trigger What starts the workflow. Here, a daily schedule.
NocoDB A tool that turns a database into a table you can edit like a spreadsheet. It can be hosted or run on your own server.
SMTP The standard way for a program to send email through an email account. You need the server address, a login and a password from your email provider.
Groq A company that runs AI models and lets programs call them with an API key.
Model The AI program that reads the prompt and writes the answer.
Prompt The written instructions you give the model.
Placeholder A marker in a text that is replaced later, such as ((Name)) in the email template, or YOUR_SENDER_EMAIL in the file.
Expression A small formula between double curly brackets, such as {{ $json.email }}, that reads a value from an earlier node.
Credential The saved login n8n uses to reach an app. Credentials are not included in the JSON file.
Sender reputation How email providers judge your address and domain from past sending. A bad one sends your messages to spam.

Apps and accounts this workflow talks to

App Its role What you need
n8n Runs the workflow every day. An n8n account or your own installation. It must stay on for the schedule to fire.
NocoDB Holds the leads and the contact dates. A table with the right columns, an API token, and the project and table identifiers.
Groq Runs the model that fills in the name and company. A Groq account and an API key.
Your email account Sends the messages by SMTP. The SMTP server, port, login and password from your provider. Many providers want a special app password, not your normal one.

What to have ready before you import anything

The table. Create a table in NocoDB with these columns. The names must be written exactly like this, with the same capitals and the same spaces, because the workflow refers to them by name:

  • first_name: the first name used in the greeting.
  • email: the address the message is sent to.
  • organization_name: the company name. The original description calls it optional, but the prompt uses it, so an empty value means the model receives an empty company.
  • Initial Contact Date: leave it empty for leads you want to contact. The workflow fills it.
  • Next Follow up/Contact: the workflow fills it with a date three days after the send.
  • Id: the row identifier. NocoDB normally creates it. The update step uses it to find the row.

The email. Write the email you want to send once. Decide where the first name and the company go and mark them with ((Name)) and ((Company)). Decide the subject line. Add whatever your situation requires at the bottom of the message, such as who you are and how to opt out. The workflow adds nothing to the message, and the attribution line that n8n can append is switched off.

The accounts. A NocoDB API token, a Groq API key, and the SMTP details of the account that will send. If you can, use a separate address for your first tests. Do not test with a list of strangers.

A few leads you control. Add two or three rows with your own addresses, or addresses of friends who agreed, and nothing in Initial Contact Date.

Where it can go wrong: the order of events, with an example

The most important thing to understand about this workflow is the order in which things happen. As far as we know, n8n runs a node on all the items it receives before it moves to the next node. That is the default behavior, and nothing in this file changes it. Here is a made-up example. Imagine four leads in the table, none contacted yet:

Row First name Email Company
1 Maya maya@… Lantern Studio
2 Omar omar@… Oakfield
3 Priya priya@… (empty)
4 Leo (empty) Pinewood

At 10:30 the workflow runs. The Filter keeps all four, since none has a contact date, and the Limit keeps all four, since the limit is 15. The AI step then runs four times, giving four email texts. Row 3 has no company, so the model has an empty value to put where the company goes, and we cannot tell what it will write there. Then the Send email step runs on the four items one after another. Rows 1, 2 and 3 are sent. Row 4 has no address, so the send fails.

When a node fails and nothing tells n8n to carry on, the run stops. The nodes after it, which are the two that record the date, never run for any of the leads. So the table still shows rows 1, 2 and 3 as uncontacted. Tomorrow at 10:30 the Filter keeps them again, and because the Limit takes the first leads in table order, they come first. Maya, Omar and Priya receive the same email a second time. Row 4 fails again. The same thing happens the day after, every day, until you fix row 4. That is the sequence we would worry about most in this workflow, and it follows directly from how the nodes are connected.

Two small points also follow from this example. First, the Limit of 15 counts leads, not successes, so a few bad rows at the top of the table can use up the day’s quota. Second, the date columns are filled only for leads whose send worked and only when the whole run reaches that point. The table is a record of what the run finished, not of what was sent. If you want a safer order, the adaptation section describes a version that saves each lead right after its own send.

The big picture: from table to inbox and back

The workflow has nine working nodes in one straight line, with the AI model attached to one of them from below:

  1. Start. Schedule Trigger.
  2. Read and choose. Get many rows, Filter and Limit.
  3. Write. Basic LLM Chain, with Groq Chat Model as its model.
  4. Send. Send email.
  5. Record. Date & Time and Update a row.

There are no branches. No node handles an error, no node waits, and there is no loop. That simplicity is what makes it easy to understand, and it is also the reason one problem can stop everything.

Follow the data: what each of the 9 steps does

The names below are the exact names in the JSON file. The file also holds three notes on the canvas, which are reminders and not steps.

Schedule Trigger

The start of the workflow. It runs every day at 10:30, using the timezone of your n8n. The file sets the hour and the minute and nothing else, so the interval is the default one, which we understand to be every day. Check the timezone in your n8n settings. The schedule only works while the workflow is switched on and n8n is running. To test by hand, use the run button on the canvas instead.

Get many rows

A NocoDB node that reads the table. It is set to get all rows, with no limit, so it reads the whole table at every run, finished leads included. It needs a NocoDB API token credential, and you replace YOUR_NOCODB_PROJECT_ID and YOUR_NOCODB_TABLE_ID. With a very large table this daily read grows with it, because filtering happens in the next node, not in NocoDB.

Filter

Keeps only the rows whose Initial Contact Date does not exist. In plain words: a lead with a contact date is skipped, a lead without one passes. The check is set to expect a date and to be strict about types. We could not confirm how an empty cell comes out of NocoDB, as no value or as empty text, or how a column of plain text is treated, so test it on one row with a date and one without. The Filter does not look at email, first_name or organization_name.

Limit

Keeps the first 15 items and drops the others. It keeps the first in table order, not the best or the newest, so the same leads come first on every run until they are marked as contacted. The number 15 is a plain setting. The original description presents it as protection for your sender reputation, which we cannot confirm for your situation. Change it to match what your provider allows.

Basic LLM Chain

The node that holds the prompt. It is set to use the text written in the node, with the first name and the company inserted from the lead by two expressions, {{ $json.first_name }} and {{ $json.organization_name }}. The prompt tells the model to change only the name and the company, to keep the formatting, and to return only the email. It contains a block for your template between START OF TEMPLATE and END OF TEMPLATE, which you must replace. The prompt also tells the model to guess where the name and company go when it finds no marker, which can lead to a rewritten text. The email address is not part of the prompt. The answer is in a field called text.

Groq Chat Model

The model behind the chain, reached through Groq under the name openai/gpt-oss-120b. Groq’s documentation lists this model, described as an open-weight model from OpenAI. It needs a Groq API key credential. Model names change, so check that it is still offered to your account. The node has no special settings in the file.

Send email

Sends the message by SMTP in plain text. The body is the text that the chain returned. The recipient comes from the lead, using an expression that reads email from the Limit node. The file sets the n8n attribution line to off. The original file had no sender address and a subject about lead qualification, so LeenLabs added YOUR_SENDER_EMAIL and YOUR_SUBJECT_LINE. All leads receive the same subject. This node has no setting to continue after an error: one failed send stops the run, and the nodes after it do not run.

Date & Time

A node with no setting written in the file. n8n leaves out values that are on their default, so we read this as the default operation, which we understand to be getting the current date into a field called currentDate. The documentation page describes the operation but does not state the default, so open the node and check that it says Get Current Date with that output field. Without a timezone chosen, it uses your instance’s timezone. Update a row reads this field.

Update a row

The NocoDB node that records what happened. It finds the row by Id, taken from Get many rows with an expression, and writes two cells. Initial Contact Date gets today’s date written as month, day and year, with the pattern M/dd/yy. Next Follow up/Contact gets the date three days later in the same pattern. The month has no leading zero and the day has one, so a date can read 2/02/27. The name of the first field starts with an equals sign in the file, which we think is a leftover and probably harmless. Nothing reads the follow-up date afterwards.

Get the workflow JSON

Open the box below to copy the JSON, or download the file. It contains the 9 nodes above, connected as described, with placeholders for your table, your sender address and your subject line, and 3 notes on the canvas, including one called Before you turn it on. It contains none of your logins.

Please read this first: this JSON is free and provided as is. n8n versions and installed nodes differ from one person to another, so it may not run correctly on your side without adjustments. We cannot test it on every setup. That is why this article includes a long section, Problems you may meet and how to fix them. And if you would rather not deal with any of it, you can request a custom workflow and let us handle it.

Click to show or hide the workflow JSON

Click inside the box, select the whole text, copy it, then paste it onto an empty n8n canvas. If selecting is awkward on your device, use the download link above.

{
  "name": "Email Outreach with n8n: AI, NocoDB and SMTP (free JSON)",
  "nodes": [
    {
      "id": "d5cc818c-806e-4b8a-86a3-6eb090fbf10b",
      "name": "Schedule Trigger",
      "type": "n8n-nodes-base.scheduleTrigger",
      "position": [
        0,
        0
      ],
      "parameters": {
        "rule": {
          "interval": [
            {
              "triggerAtHour": 10,
              "triggerAtMinute": 30
            }
          ]
        }
      },
      "typeVersion": 1.2
    },
    {
      "id": "cc7c393b-1edb-496f-9739-71ba7dcfe692",
      "name": "Get many rows",
      "type": "n8n-nodes-base.nocoDb",
      "position": [
        176,
        0
      ],
      "parameters": {
        "table": "YOUR_NOCODB_TABLE_ID",
        "options": {},
        "operation": "getAll",
        "projectId": "YOUR_NOCODB_PROJECT_ID",
        "returnAll": true,
        "authentication": "nocoDbApiToken"
      },
      "typeVersion": 3
    },
    {
      "id": "2b19fa40-85c2-4d12-810c-04bcffabb877",
      "name": "Filter",
      "type": "n8n-nodes-base.filter",
      "position": [
        352,
        0
      ],
      "parameters": {
        "options": {},
        "conditions": {
          "options": {
            "version": 2,
            "leftValue": "",
            "caseSensitive": true,
            "typeValidation": "strict"
          },
          "combinator": "and",
          "conditions": [
            {
              "id": "0fac1715-b303-4697-a3cf-1cb8c47ab439",
              "operator": {
                "type": "dateTime",
                "operation": "notExists",
                "singleValue": true
              },
              "leftValue": "={{ $json['Initial Contact Date'] }}",
              "rightValue": ""
            }
          ]
        }
      },
      "typeVersion": 2.2
    },
    {
      "id": "46c0f6e6-64c8-4f46-b322-16d20e711aa5",
      "name": "Limit",
      "type": "n8n-nodes-base.limit",
      "position": [
        560,
        0
      ],
      "parameters": {
        "maxItems": 15
      },
      "typeVersion": 1
    },
    {
      "id": "5f2d6070-ac2a-4e75-a12d-0dae1cda1abb",
      "name": "Basic LLM Chain",
      "type": "@n8n/n8n-nodes-langchain.chainLlm",
      "position": [
        752,
        0
      ],
      "parameters": {
        "text": "=You are an email personalization assistant. Your job is to modify an email template by replacing ONLY two specific elements:\n\n1. Replace the recipient's name\n2. Replace the company name\n\nCRITICAL INSTRUCTIONS:\n- Change ONLY the name and company name\n- Keep ALL other content exactly the same\n- Maintain the exact formatting, spacing, and structure\n- Do not modify subject lines, signatures, or any other text\n- Do not add extra content or make style changes\n\nINPUT FORMAT:\nYou will receive:\n- Original Email: \n---START OF TEMPLATE---\nHey ((Name)),\n\nPASTE YOUR EMAIL TEMPLATE HERE. Write ((Name)) where the first name goes and ((Company)) where the company name goes.\n---END OF TEMPLATE---\n\n- New Name: {{ $json.first_name }}\n- New Company: {{ $json.organization_name }}\n\nREPLACEMENT RULES:\n- Look for the placeholder ((Name)), in any capitalization, or obvious name placeholders\n- Look for the placeholder ((Company)), in any capitalization, or obvious company placeholders\n- If no placeholders exist, intelligently identify where names/companies should go\n- Use the exact capitalization provided in the input\n\nOUTPUT FORMAT:\nReturn only the modified email with no additional commentary, explanations, or metadata.\n\nEXAMPLE:\nInput:\nOriginal Email: \"Hi ((name)), Welcome to ((company))! We're excited to work with you.\"\nNew Name: \"Sarah Johnson\"\nNew Company: \"TechCorp Solutions\"\n\nOutput:\nHi Sarah Johnson, Welcome to TechCorp Solutions! We're excited to work with you.",
        "batching": {},
        "promptType": "define"
      },
      "typeVersion": 1.7
    },
    {
      "id": "7cac1b8d-5521-42ac-b33d-3dbc148b6e6d",
      "name": "Groq Chat Model",
      "type": "@n8n/n8n-nodes-langchain.lmChatGroq",
      "position": [
        752,
        208
      ],
      "parameters": {
        "model": "openai/gpt-oss-120b",
        "options": {}
      },
      "typeVersion": 1
    },
    {
      "id": "11d4208c-d028-4502-89a7-516e3afa3c6e",
      "name": "Send email",
      "type": "n8n-nodes-base.emailSend",
      "position": [
        1104,
        0
      ],
      "parameters": {
        "text": "={{ $json.text }}",
        "options": {
          "appendAttribution": false
        },
        "subject": "YOUR_SUBJECT_LINE",
        "toEmail": "={{ $('Limit').item.json.email }}",
        "emailFormat": "text",
        "fromEmail": "YOUR_SENDER_EMAIL"
      },
      "typeVersion": 2.1
    },
    {
      "id": "917b005c-8f57-4692-bb07-08d1f7bcdd82",
      "name": "Update a row",
      "type": "n8n-nodes-base.nocoDb",
      "position": [
        1520,
        0
      ],
      "parameters": {
        "table": "YOUR_NOCODB_TABLE_ID",
        "fieldsUi": {
          "fieldValues": [
            {
              "fieldName": "Id",
              "fieldValue": "={{ $('Get many rows').item.json.Id }}"
            },
            {
              "fieldName": "=Initial Contact Date",
              "fieldValue": "={{ $json.currentDate.toDateTime().format('M/dd/yy') }}"
            },
            {
              "fieldName": "Next Follow up/Contact",
              "fieldValue": "={{ $json.currentDate.toDateTime().plus(3, 'Days').format('M/dd/yy') }}"
            }
          ]
        },
        "operation": "update",
        "projectId": "YOUR_NOCODB_PROJECT_ID",
        "authentication": "nocoDbApiToken"
      },
      "typeVersion": 3
    },
    {
      "id": "024256ad-46f8-4eab-82fa-e27aa8f58cd3",
      "name": "Date & Time",
      "type": "n8n-nodes-base.dateTime",
      "position": [
        1312,
        0
      ],
      "parameters": {
        "options": {}
      },
      "typeVersion": 2
    },
    {
      "id": "7b67d2c4-327d-4f7d-ae58-5bf3f04eb2b3",
      "name": "Before you turn it on",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        -60,
        -640
      ],
      "typeVersion": 1,
      "parameters": {
        "color": 4,
        "width": 640,
        "height": 520,
        "content": "## Before you turn it on\n\nThis file contains placeholders. Replace each one:\n\n- `YOUR_NOCODB_PROJECT_ID` and `YOUR_NOCODB_TABLE_ID`, in both NocoDB nodes\n- `YOUR_SENDER_EMAIL` and `YOUR_SUBJECT_LINE`, in Send email\n- the sample text between START OF TEMPLATE and END OF TEMPLATE in Basic LLM Chain\n\nThen add your own credentials: NocoDB API token, Groq API key, SMTP account.\n\nFirst run: put only your own email address in the table, run the workflow by hand, and read the email before anything else.\n\nThe file has been checked for structure only. It has not been run against live accounts."
      }
    },
    {
      "id": "28ecbb9a-a783-442c-b8f1-4b875390649e",
      "name": "Table columns",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        620,
        -640
      ],
      "typeVersion": 1,
      "parameters": {
        "color": 5,
        "width": 520,
        "height": 520,
        "content": "## Table columns\n\nThe names must match exactly, including capitals and spaces:\n\n- `Id` (the table's own row id)\n- `first_name`\n- `email`\n- `organization_name`\n- `Initial Contact Date`\n- `Next Follow up/Contact`\n\nA lead is picked up only when `Initial Contact Date` has no value. After a send, the workflow writes the date in the form month/day/year, for example 11/14/26. Check that your column accepts that format."
      }
    },
    {
      "id": "ae1a5399-e4f7-4813-8612-3307a99f48e2",
      "name": "What it does not do",
      "type": "n8n-nodes-base.stickyNote",
      "position": [
        1180,
        -640
      ],
      "typeVersion": 1,
      "parameters": {
        "color": 6,
        "width": 560,
        "height": 520,
        "content": "## What it does not do\n\n- It does not send follow-ups. It only writes a follow-up date.\n- It has no unsubscribe link and does not read replies or bounces.\n- All emails of a run are sent first and the table is updated afterwards. If one send fails, the run stops before any update, and the leads already emailed can be emailed again next time.\n- The limit of 15 is a setting in the Limit node, not a safe level.\n- Check the rules that apply to cold email where you and your leads are."
      }
    }
  ],
  "connections": {
    "Limit": {
      "main": [
        [
          {
            "node": "Basic LLM Chain",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Filter": {
      "main": [
        [
          {
            "node": "Limit",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Send email": {
      "main": [
        [
          {
            "node": "Date & Time",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Date & Time": {
      "main": [
        [
          {
            "node": "Update a row",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Get many rows": {
      "main": [
        [
          {
            "node": "Filter",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Basic LLM Chain": {
      "main": [
        [
          {
            "node": "Send email",
            "type": "main",
            "index": 0
          }
        ]
      ]
    },
    "Groq Chat Model": {
      "ai_languageModel": [
        [
          {
            "node": "Basic LLM Chain",
            "type": "ai_languageModel",
            "index": 0
          }
        ]
      ]
    },
    "Schedule Trigger": {
      "main": [
        [
          {
            "node": "Get many rows",
            "type": "main",
            "index": 0
          }
        ]
      ]
    }
  },
  "settings": {
    "executionOrder": "v1"
  }
}

Install it in eleven steps

  1. Create the table in NocoDB with the six columns from the preparation section, and add one test lead with your own address and nothing in Initial Contact Date.
  2. Get the JSON from the box above or the download link.
  3. Import it. Open a new empty workflow in n8n, then paste the JSON onto the canvas or import it from a file.
  4. Check that the 9 nodes appear, with 3 notes and no missing-node warning.
  5. Connect NocoDB in Get many rows and Update a row. Replace YOUR_NOCODB_PROJECT_ID and YOUR_NOCODB_TABLE_ID in both nodes.
  6. Connect Groq in Groq Chat Model, and check that the model name is still offered to your account.
  7. Write your template. In Basic LLM Chain, replace the sample text between START OF TEMPLATE and END OF TEMPLATE with your own email, keeping ((Name)) and ((Company)) where they belong.
  8. Connect your SMTP account in Send email. Replace YOUR_SENDER_EMAIL and YOUR_SUBJECT_LINE.
  9. Open Date & Time and check that it is set to get the current date and writes it to a field called currentDate.
  10. Run it by hand from the canvas with only your own address in the table. Read the email you receive, then open the table and check both date columns.
  11. Only then add a few real leads, lower the Limit if you want to start small, and switch on the schedule. In recent n8n versions the switch may be labeled Publish instead of Active.

What it will not do: limits, costs and trade-offs

  • It does not send follow-ups. It writes a date three days ahead and nothing reads it. You need a second workflow, or a person, to act on that date.
  • It does not handle replies, bounces or opt-outs. A lead who answers “stop” is still in the table with a contact date, which keeps them from being emailed again by this workflow only. But nothing records the request, and any other tool you use will not know.
  • It does not check email addresses. A bad address makes a send fail and stops the run.
  • It does not pause between emails. They go out one after another, which can look like bulk sending to your provider.
  • It does not save as it goes. The table is updated after all the sends.
  • The limit of 15 is only a setting. The original description says it helps protect your reputation. We cannot say that 15 is safe for your domain, your provider or your audience. Check your provider’s published limits.
  • Costs depend on your providers. The original description quotes a cost per personalization. We did not verify it and we do not repeat it. Check the current pages of the AI service, your email provider and your NocoDB hosting.
  • The email is one text for everyone. Only the name and the company change.

Access, privacy and responsible use

This workflow sends messages in your name to people who did not ask for them. A few things are worth deciding before you switch it on.

  • What goes to the AI service. The prompt contains your template, the first name and the company name. The email address is not placed in the prompt. Even so, a name and a company are personal data in many places, so check what your AI provider says about how it handles what you send.
  • What goes to your email provider. The whole message and the address of each lead. Read the provider’s rules on bulk and unsolicited email. Some providers restrict accounts that send cold email.
  • Who you email. Rules about unsolicited email differ from one country to another, and may differ for business and personal addresses. We do not give legal advice and we do not claim that this workflow meets any law. Our guide to cold email compliance for small teams is a good place to start, and a qualified person should confirm what applies to you.
  • How people can say no. The workflow has no unsubscribe step. If your situation requires an opt-out, put the instructions in your template and handle each request yourself or in another workflow.
  • Who can open the workflow. Anyone with access to your n8n can read the template, run the workflow and see the table. Limit access to the people who need it.

Making the email read like a person wrote it

Because this workflow only swaps two words, the quality of the message is the quality of your template. A few habits help. Write it the way you would write to one person, in short sentences. Say who you are in the first lines. Ask one clear question, not three. Do not claim results you cannot show. Leave out words that make a message read like a mass mailing. Our article on AI cold emails that do not sound like AI goes through the common mistakes, with examples.

Test the greeting too. If a first name is empty in your table, the email will start with “Hey .” and the model may invent something to fill the gap. Fill every first_name before the lead enters the queue, or add a condition that skips rows with no name.

Problems you may meet and how to fix them

Every n8n setup is different, so this workflow can behave differently for you than in our checks. Here are the most likely problems. For each one: what you see, why it happens, and what to do. They come from reading how the nodes are set up, not from a live run with real accounts.

1. The workflow imports, but nodes show as unknown or with warnings

What you see: red or grey nodes after import, especially Basic LLM Chain and Groq Chat Model.

Why: your n8n version may not include the same AI nodes, or not at the same version, as the one the template was made on.

Fix: update n8n, or add the same kind of node from your own menu and copy the settings across.

2. NocoDB does not connect or cannot find the table

What you see: an error in Get many rows or Update a row.

Why: YOUR_NOCODB_PROJECT_ID or YOUR_NOCODB_TABLE_ID was not replaced, the API token is wrong, or n8n cannot reach your NocoDB address.

Fix: replace both identifiers in both nodes, create the credential again, and check that your NocoDB server is reachable from where n8n runs.

3. The workflow stops right after Get many rows

What you see: nothing after the first node, or an empty result.

Why: the table is empty, or the node points to another table.

Fix: open the node, run it alone and read its output. It should list your rows.

4. The Filter lets nobody through, or lets everybody through

What you see: no emails sent although leads exist, or leads that were already contacted get emailed again.

Why: the Filter keeps rows where Initial Contact Date does not exist. How an empty cell is stored depends on the column type. We could not confirm whether your table returns an empty cell as no value or as an empty text, and the two may be treated differently.

Fix: run Get many rows, look at what an empty Initial Contact Date looks like in the output, then test the Filter on a row with and a row without a date.

5. The Filter complains about a wrong type

What you see: an error saying a value is not a date.

Why: the Filter is strict about types and expects a date. If the column is plain text, or holds text such as 11/14/26, it may not be read as a date.

Fix: use a date column type, or relax the type check in the Filter and test it.

6. Groq refuses the call or the model is missing

What you see: an error in Groq Chat Model or Basic LLM Chain, perhaps about the model or your quota.

Why: the key is wrong, the model name openai/gpt-oss-120b is no longer offered to your account, or you hit a limit.

Fix: create the credential again, pick a model from the list in the node, and check your account’s limits on the provider’s page.

7. The email says “PASTE YOUR EMAIL TEMPLATE HERE”

What you see: that text, or the words ((Name)) and ((Company)), inside the message you receive.

Why: the sample text in Basic LLM Chain was not replaced, or your template used different markers from the ones the prompt tells the model to look for.

Fix: replace the sample text with your email, use ((Name)) and ((Company)) as the markers, and run a test before every change of template.

8. The email contains extra words or a different text

What you see: an introduction such as “Here is the email”, a changed sentence, or a different signature.

Why: the model is asked to change only the name and company, but it is not forced to. The prompt even tells it to find a place for the name if it sees no marker.

Fix: use clear markers, read the first emails of each batch, or replace the AI step with a simple text node, as described in the adaptation section.

9. The company part reads badly or is invented

What you see: a greeting like “Love .” or a company name you never gave.

Why: organization_name is empty for that lead, so the model gets an empty value and has to decide what to write.

Fix: fill the column for every lead, or add a condition in the Filter that skips leads with no company.

10. Send email fails on the sender or the login

What you see: an error in Send email, perhaps about authentication or the sender address.

Why: YOUR_SENDER_EMAIL was not replaced, the SMTP login is wrong, or your provider needs an app password and not your normal one.

Fix: replace the sender address with the address of the SMTP account, and check the host, port and password with your provider’s instructions.

11. One bad address stops the whole run

What you see: some leads got the email, the run shows an error in Send email, and the table was not updated for anyone.

Why: the node has no setting that lets the run continue after an error, so the first failed send stops it. See the example in this guide.

Fix: clean the addresses before each run, delete or correct the failing row, and consider the safer order described in the adaptation section.

12. The same leads are emailed again the next day

What you see: people tell you they received the message twice, or a lead you already contacted is in the new batch.

Why: the update step did not run, because the run stopped earlier, or it ran and wrote nothing because of a column problem.

Fix: after every run, open the table and check that each contacted lead has a date. If not, mark them by hand before the next run, and fix the cause first.

13. The wrong lead gets the date, or the update fails

What you see: a date on a different row, or an error in Update a row.

Why: the update takes the row Id from Get many rows with a formula that pairs each lead with its earlier row. We could not confirm that this pairing always holds across all the nodes in between. The error may also come from a table that has no Id column.

Fix: test with two leads and compare the dates to the rows. If they do not match, tell us in a custom request or rebuild the link with the Id carried along in each item.

14. The dates look odd, or the table refuses them

What you see: dates such as 2/02/27, with a month without a leading zero and a day with one, or a refusal by the column.

Why: the format is written as month, then day, then year, in the pattern M/dd/yy. A date column may expect another form, and the American order may confuse readers elsewhere.

Fix: change the pattern in Update a row to the form your column accepts, such as year, month and day.

15. The current date is missing or has the wrong day

What you see: an empty date, a message about currentDate, or a date one day off.

Why: the Date & Time node has no setting in the file, because n8n leaves out values that are on their default. We believe the default is to get the current date into currentDate, but the documentation page does not state it. The node uses your instance’s timezone unless you choose another.

Fix: open the node and make sure it reads Get Current Date with the output field currentDate. Set the timezone you want in its options.

16. The workflow does not start at 10:30

What you see: no run at the expected time.

Why: the workflow is not switched on, n8n was not running, or the timezone is not the one you think.

Fix: switch the workflow on, check the timezone in your n8n settings, and look at the list of executions the next day.

17. Emails land in spam, or your provider slows you down

What you see: no replies, bounces, warnings from your provider, or messages in spam folders.

Why: the messages are identical apart from two words and are sent in a quick series from one address. The setting of 15 does not prevent this.

Fix: lower the number, send from an address with history, and read your provider’s guidance on sending limits and on authentication of your domain.

18. Nobody follows up

What you see: a follow-up date in the table, and nothing on that date.

Why: no node reads Next Follow up/Contact. The workflow only writes it.

Fix: follow up by hand, or build a second workflow. Our missed lead follow-up system explains how to keep follow-ups from slipping.

19. Leads receive a subject line that says “YOUR_SUBJECT_LINE”

What you see: that text in the subject of a real message.

Why: the placeholder in Send email was not replaced.

Fix: replace it, then check the subject in your first test email. All leads get the same subject, because it is a fixed text.

20. A lead with no email or no name is still processed

What you see: a failed send, or a greeting with no name.

Why: the Filter checks only the contact date. It does not check that the other columns have values.

Fix: add conditions to the Filter for email, first_name and organization_name, or fill them before the lead enters the table.

Too much of a puzzle? Let us build or adapt it for you

Some people enjoy tinkering with n8n. Others want an outreach routine that behaves, with checks they can trust. If you are in the second group, LeenLabs builds and adapts n8n workflows for people like you. We can:

  • Save each lead right after its own send, so one failure never causes duplicates.
  • Check addresses and skip rows with missing fields before anything is sent.
  • Add pauses and a daily ceiling that follows your provider’s limits.
  • Add follow-up steps that read the follow-up date and send the next message.
  • Record replies, bounces and opt-outs in your table.
  • Replace the AI step with a plain text step, or keep it with stricter checks.
  • Connect your own CRM instead of NocoDB.
  • Build a new workflow from scratch for another task you want to automate.

We build the automation. You stay in charge of who is contacted and what is said.

Request a Custom Workflow, describe what you want to automate, and we will come back to you with the next steps. Prefer to ask a question first? Write to support@leenlabs.site.

What we changed in the original template

We want you to know exactly how this file differs from the template it comes from. Compared with the file we received, LeenLabs made these changes:

  1. Removed internal data. We removed the template’s instance identifier, the credential references, the webhook identifier of the email node and the empty saved test data, renamed the workflow to this article’s title and kept only the execution-order setting.
  2. Replaced the author’s table. The identifiers of the author’s NocoDB project and table became YOUR_NOCODB_PROJECT_ID and YOUR_NOCODB_TABLE_ID.
  3. Added a sender placeholder. The file had no sender address in the email node. We added YOUR_SENDER_EMAIL.
  4. Replaced the subject line. The original subject, which was about lead qualification, was specific to the author’s offer. It is now YOUR_SUBJECT_LINE.
  5. Replaced the sample template. The sample email text and the unfinished marker [[Insert Email Template Here]] became a clearly marked block between START OF TEMPLATE and END OF TEMPLATE. The original had an opening bracket that was never closed.
  6. Made the markers case-insensitive. The prompt asked for ((name)) in one place and showed ((Name)) in its own example. We now ask the model to look for the markers in any capitalization.
  7. Rewrote the notes. The four large notes mentioned a price per personalization, a free tier and minute estimates for setup. We replaced them with three short notes: what to replace, what the columns are, and what the workflow does not do.
  8. Left unchanged. The schedule, the filter, the limit of 15, the model name, the date formulas, the follow-up of three days and the structure.

Where the description and the file differ. The description says the email is sent by SMTP or the Gmail node. The file contains only the SMTP node. It says the workflow sets up follow-up routines. The file only writes a date. It says the daily limit protects your reputation. The file only keeps the first 15 leads. We report these gaps so that you do not expect more than the nodes do.

What was verified, and what was not

Here is exactly what we did. This template has no Code node, so there was no code to run. We checked the structure of the file: 9 working nodes, 3 notes, connections that match the description, no logins, and none of the author’s identifiers or text. With a script, we listed every value that nodes read from earlier nodes by name and checked that those nodes exist. There are two: Send email reads the address from Limit, and Update a row reads the row identifier from Get many rows. We also compared the claims in the original description with the contents of the nodes, and read the documentation pages of the Send Email, Date & Time and NocoDB nodes and the page of the Groq model.

We did not import this exact file into n8n and run it. We did not connect NocoDB, Groq or an SMTP account, and we did not send any email. So we cannot confirm how the Filter treats an empty date, whether the pairing used by Update a row holds, what the Date & Time node does with no setting, what the model returns, how your provider reacts to a series of identical messages, or what the date format does in your table. Not tested on live APIs. That is why we recommend a full test with your own address before you use any real lead.

Four ways to adapt the workflow

  1. Replace the AI with a text node. Add an Edit Fields node that builds the message from your template with the first name and company taken from the lead. It costs nothing per lead and cannot change a sentence.
  2. Save each lead after its own send. Put a Loop Over Items node in front of Send email with a batch size of one, so that each lead is emailed and then recorded before the next one starts. A failure then affects one lead instead of the whole batch.
  3. Skip incomplete rows. Add conditions to the Filter so that only rows with an email, a first name and a company pass.
  4. Send the follow-up. Build a second workflow that looks for rows whose follow-up date is today and whose lead has not replied, and send the second message from there.

These variations are suggestions based on how n8n works, not changes we tested on this template. If you want one built for you, send us a request.

Frequently asked questions

What is email outreach?

Writing first messages to people who have not asked to hear from you, to start a conversation or offer a service. This workflow sends one such message a day to a limited number of leads.

What is an outreach email?

One of those first messages. Here it is a single template where only the name and the company change.

Does this workflow personalize the whole email?

No. It replaces two things, the first name and the company name. The rest of the email is the same for every lead.

Do I need the AI step?

Not for this job. A plain text step can swap the name and the company without cost and without risk of changes. The AI step is a choice that adds flexibility and some risk.

How many emails per day per domain is safe for cold outreach?

There is no single number. It depends on your provider’s published limits, the age and reputation of your domain, and how recipients react. The 15 in this workflow is only a setting. Check your provider’s guidance and start with a small number.

Does it send follow-up emails?

No. It records a follow-up date three days after the first email. Nothing in the workflow reads that date.

What happens if one email fails?

The run stops. The leads already emailed are not recorded, so they can be emailed again at the next run. See the example in this guide.

Can I use Gmail instead of SMTP?

The original description says yes. The file contains only the SMTP node. If you replace it with a Gmail node, set the address and the message fields again and test.

Can I use a database other than NocoDB?

The original description says any database can replace it. You would swap the two NocoDB nodes for the nodes of your database and keep the same column names or adapt the formulas.

Which leads should I put in the table?

Only leads that you have reason to contact. Qualify B2B sales leads before they enter the queue, so that you do not email people who will never be a fit.

The JSON does not work on my setup. What do I do?

Read the problems section first. Most issues come from placeholders that were not replaced, credentials, column names and the node versions of your n8n. If you are still stuck, request a custom workflow.

Who created the original template?

It comes from the n8n Library at n8n.io. LeenLabs prepared and documented the file, and did not create it.

Tools used in this workflow

The links below go to each tool’s home page, and some may be affiliate links, as explained in our Terms of Service. We do not link to NocoDB because we could not verify its home page.

  • n8n (may be an affiliate link): the automation platform that runs the workflow, available hosted or self-hosted.
  • Groq (may be an affiliate link): the AI service that runs the model which fills in the name and the company.

Suggested reading

If cold email, leads and follow-up are your focus, these LeenLabs articles go well with this one.

Workflows

Guides

Get more advanced automation and AI resources for business

A workflow article covers one template at a time. For people who want to go further, we will send more advanced resources on automation and AI in business, beyond what fits in a single workflow guide. Leave your email below and you will receive them as they are released.

Email
Accept the terms

Keep exploring

Browse more free n8n workflows from LeenLabs.

Prefer a ready-made workflow?

If you run your own outreach, this is the kind of work that a ready-made workflow could take off your plate. The LeenLabs Marketplace is the home of paid n8n workflows for Solopreneurs & Freelancers. Each workflow is tested on an n8n instance before it is listed. Leave your email on the category page to be notified when new workflows are listed.

See Solopreneurs & Freelancers workflows   Browse the whole marketplace

Need this adapted to your own setup?

Tell us what you want to automate and we will come back to you with the next steps.

Request a Custom Workflow