Book Call
⚙️ How I Built a Lead Scoring System in 3 Hours and 40 Minutes (Condensed Into 100 Seconds)

When a client asks, "How do we prioritize the right accounts in our CRM?" — this is the answer.

In this post, I'll walk you through the exact process I followed (in just under 4 hours) to build a custom lead scoring model using Clay, AI enrichment, and historical customer insights. The final result? A reliable, repeatable way to surface the most qualified accounts — and cut out the noise.

💡 Note: The entire project was captured in a 100x time-lapse video. Scroll down to watch it play out in under two minutes.

Start With What You Already Know (Good vs. Bad Fit) Before diving into tech, I pulled together a list of ~100 companies the client had already interacted with. Some were clearly great fits. Others… not so much.

From there, I built an Ideal Customer Profile (ICP) using the patterns we could already see:

  • Industry type (e.g. food distributor)
  • Business model (e.g. direct store delivery)
  • Size, age, ownership structure
  • Ethnic food focus, product lines, and delivery capabilities
Practical Applications of Lead Scoring
🗂️ Scoring your entire CRM database

Automatically enrich and re-rank every lead and account already sitting in your CRM. Weed out stale, low-fit records — and surface hidden gems.

📥 Scoring inbound leads as they arrive

Run this scoring model in real-time (or daily batches) on new form fills and demo requests to route high-fit leads faster.

📤 Scoring outbound lists before assigning to sellers

Use this to prioritize cold outbound lists (whether from Apollo, Clay, or a trade show list) so your reps focus only on the best matches. Great for territory planning and SDR focus.

📹 Watch the Build in Action (100x Speed)

If you'd like this set up for your business — or want help defining your ICP from scratch — [reach out here] or drop me a message.

Building a Quantifiable Lead Scoring Framework
20
Ethnic Food Focus

Imports or distributes ethnic food

20
Client Base

Serves foodservice/retail clients

20
Delivery Network

Has its own fleet or delivery network

15
Sales Structure

Operates with sales reps or field delivery

15
Company Size

11–200 employees

5
Business Maturity

Family-owned or in business 5+ years

Map Each Trait to a Clear Scoring Rule Each ICP trait got assigned a point value based on importance. This framework let me turn subjective opinions into quantifiable signals.

Use Clay + AI to Enrich Every Company

Clay is my go-to for scalable research. For each lead, I used Clay AI Agents to answer questions like:

  • "Does this company import or distribute ethnic foods?"
  • "Do they serve restaurants, retailers, or foodservice?"
  • "Do they operate their own fleet?"
  • "Is it a family business? When was it founded?"

Each question returned a Yes/No/Partial + AI reasoning. Then I used formulas to assign points accordingly.

Rescale Scores and Tier Every Lead

Once all companies were enriched, I totaled their scores (out of 100) and bucketed them into three tiers:

80%
High Fit

80-100 points

60%
Medium Fit

60-79 points

59%
Low Fit

Less than 60 points

This gave the client a simple, sortable view of their entire lead universe.

Debug What's Dragging Scores Down

Not all leads scored well — and that's the point. I audited which ICP traits were missing or underperforming and discovered:

  • Some companies didn't mention fleet or delivery
  • Others were too large or too vague online
  • A few fields needed better AI prompts or secondary lookups

These insights fed back into improving the Clay logic for future batches.

Contact Information