Signal-Based Outbound Engine

What it does

The engine runs outbound from start to finish, with very little manual work:

• Scores companies against the client's ideal customer profile (ICP)
• Learns the client's value and best use cases, so it knows what to pitch
• Researches each target company and its recent news
• Spots a likely challenge, like a product launch or a new executive who needs early wins
• Writes a short, personal email: subject under 8 words, body around 88 words
• Saves it to the CRM record, ready for the sales sequence

Problem

Most cold email fails for one of two reasons. It goes to the wrong company, or it says nothing about what's going on for them.

Personal research fixes both, but it takes 15-20 minutes per account. Nobody can do that for hundreds of accounts, so teams fall back to templates and get ignored.

My role

I designed and built the full pipeline:

• Set up ICP scoring, account tiers and "fit gates" so weak-fit accounts drop out early
• Wrote the AI prompts that learn the client's value, read the news and suggest the likely pain point
• Set the email rules: short, specific, one clear ask, no hype
• Connected it to HubSpot so each email lands on the right contact record
• Managed the sales sequences once contacts were loaded

Workflow

1. 01
   Find

   Pull target companies and score them against the ICP.

2. 02
   Filter

   Fit gates remove poor matches before any credits are spent.

3. 03
   Learn

   An AI agent studies the client's value and use cases.

4. 04
   Research

   Read each target's recent news and signals.

5. 05
   Diagnose

   Name the most likely challenge behind that signal.

6. 06
   Write

   Draft a subject line and an 88-word email linking the challenge to the client's help.

7. 07
   Load

   Save the email to the HubSpot contact and enrol it in a sales sequence.

Tools and integrations

• Workflow and enrichment: Clay
• AI research and writing: Claude (Claygent)
• CRM and sequences: HubSpot

Example output

Illustrative only. Made-up companies and people.

Signal: Northwind Robotics hires a new Chief Revenue Officer
Likely challenge: new leader needs pipeline wins in the first 90 days

Subject: New CRO at Northwind, quick idea

Hi Sam, congrats on bringing Priya Shah in as your CRO last month. New revenue leaders usually get 90 days to show a plan, and pipeline coverage is often the first question the board asks. [Client] helps scale-ups like Northwind find in-market accounts and book first meetings without adding headcount. Teams typically start with one segment and a four-week test, so there's something to show early. Worth a 15-minute chat next week to see if it fits Priya's first-quarter plan? Either way, good luck with the new chapter.

Results & status

• Built and used to run live outbound sequences
• Also used to enrich trade show exhibitor lists for event outreach
• Not yet measured publicly: reply rate, meetings booked, pipeline

Limitations

• If the news is thin, the email gets generic, so those accounts need a human touch
• AI can misread a signal, so a person should check emails before they send
• Enrichment credits add up fast at scale, which is why the fit gates come first
• Outreach has to follow local spam and privacy rules

Related projects