Personalised sales emails sent from a lead list

What it is

This is an n8n workflow that turns a spreadsheet of prospects into a set of ready-to-send sales emails. It reads each row, asks a language model to write a message for that specific person, and hands the result to your email tool. You keep control of the sending step, so nothing leaves your outbox until you decide it should.

The value is not the writing itself. It is the fact that you stop copying details from one tab to another, and you stop rewriting the same introduction fifty times with a different name at the top.

When to use it

Use it when you have a defined list of people worth contacting and a clear offer. A list of fifty to a few hundred leads is a good fit. It suits founders doing outbound for the first time, small sales teams without a dedicated ops person, and anyone running a single campaign rather than an always-on sequence.

Skip it if your leads have no useful context attached. A model given only a name and a domain will produce something generic, and generic outreach is worse than none.

How to use it

1. Prepare your lead sheet. One row per person, with columns for first name, company, role, and a short note on why you are reaching out. That note matters most.
2. Import the workflow into n8n and open the first node. Point it at your sheet, either through a spreadsheet node or a webhook if your data lives elsewhere.
3. Set up the model node. Add your API credentials, then write the instruction. Tell it who you are, what you sell, the tone you want, and a hard limit of around 120 words. Ask for a subject line and a body as separate fields.
4. Add a filter or approval step before sending. A simple review queue, a draft folder, or a Slack message per email all work. Read a sample of ten before you release the rest.
5. Connect your email node. Map the generated subject and body to the right fields, set the sender, and switch the workflow to active.
6. Log every send back to the sheet with a timestamp and status, so you never contact the same person twice.

What a good result looks like

Each email should read as if a person wrote it in five minutes after looking at a profile. It names the company, references something specific, and makes one clear ask. For example: a note to a head of operations at a logistics firm that mentions their recent warehouse opening and asks whether route planning is still manual.

Measure replies, not volume. If fewer than a handful of people answer out of fifty, the problem is usually the lead list or the offer, not the wording. Fix the input before you touch the prompt. Keep a copy of every generated email so you can compare what worked across campaigns.