As being a start-up with limited resources, we want to automate our Sales activities as much as possible. There forth I started an Automation workflow which includes our ICP (Ideal Customer Profile), a bunch of LinkedIn profiles of possible leads and an AI Agent to determine the score per lead.
See below the workflow and which steps it follows
- Read out LinkedIn profile url and enriches them with all their hard- and soft-skills using the SourceGeek node (SourceGeek for LinkedIn integrations | Workflow automation with n8n)
- Use an AI Agent which includes our ICP
- The AI Agent comes up with a score per lead
- Also it creates 3 messages for the outreach, an intro, a follow-up and a final message
After it’s finished you can check out the Google Sheet and determine which lead you want to reach out to.
Next up is the executing of the outreach and the messages. Please let me know what you think and how you want to do your automated LinkedIn outreach.
5 Likes
This is a neat workflow. Using AI for lead scoring usually works much better than simplistic rule-based systems. I also work at a small startup where time is precious. We use AI a lot to help us move faster.
LinkedIn of course has a ton of useful data, but it’s not always easy for AIs to access. So your node offers a lot of value. I’m curious to hear more about how you’re using an agent to score each lead. What criteria are you using?
One issue I’ve struggled with is making the scores consistent across different leads. When evaluating a lead in isolation, it’s difficult to know where it ranks relative to others. My team has built a tool called everyrow-sdk which can research leads and rank them according to natural language criteria. I’ve found that it’s very important to be explicit about your scoring rubric so that the AI can apply it consistently.
We don’t have an official n8n integration for the SDK yet. But it shouldn’t be too hard to integrate with n8n as-is. You could use a Code node to run the Python code directly, or even an HTTP Request node to call the API directly. If there’s interest, we could work on an official integration.
4 Likes
Thanks for your message. What you can do in your AI Agent is to very clearly write our your Ideal Customer Profile (ICP) and tel it match with that. Also on top of that ask the AI Agent to write a reasoning so you can see the details of the scoren.
So you can include the following in your AI Agent:
- What is the role of your ICP?
- What is the company size?
- Industry, Geography and possible budget?
And write down a few persona’s so AI does have some examples.
Last but not least, describe your ‘anti-ICP’. What kind of people or companies do not fall in your target audience.
As an extra bonus you can let AI also write you highly personalized messages which you then can use to reach out to them, through LinkedIn for example.
I hope this helps.
1 Like
