Case Study CRE Lead Generation BOFU 2026-04-27 11 min read

AI Lead Generation System:
A Practical Case Study

This is the breakdown of a production system — not a concept, not a template, but a live 5-tool pipeline built for a US Commercial Real Estate broker. ICP precision targeting, a relational lead database, 4-scenario Make automation, and AI-personalized outreach. 28 real screenshots. The full architecture documented end-to-end.

The Problem: A Pipeline That Required Constant Manual Attention

The context: a top-ranked Commercial Real Estate broker in Florida, specializing in industrial and logistics properties. Strong deal flow, strong close rate — but a prospecting operation that depended entirely on personal outreach and referrals. No systematic pipeline. No consistent top-of-funnel activity. Revenue was strong when deals were in progress and flat when they weren't.

The objective was to build a lead generation system that could run continuously in the background — identifying and engaging qualified decision-makers without requiring the broker to do the prospecting work personally. The system needed to operate at scale without sacrificing the high-touch feel that CRE relationships require.

The constraint: The broker's time is the scarcest resource. Any system built had to minimize the number of decisions required from the broker — and maximize the quality of the meetings it produced. The system should do the work. The broker should show up to the conversation.

System Design: Five Tools, One Pipeline

// CRE Lead Factory — Production Architecture
Apollo.io
ICP targeting + AI outreach
Airtable
Relational DB + scoring
Make.com
4-scenario automation
Pipedrive
Pipeline intelligence
ClickUp
Delivery ops
01

ICP Precision Targeting — Apollo.io

The ICP was defined with precision: VP Real Estate, Director of Real Estate, CFO, and COO at companies with 500–5,000 employees in logistics, 3PL, e-commerce fulfillment, and light manufacturing — located in Florida and Texas, with demonstrated real estate activity signals. 15 stacked Apollo filters produced a working list of 160 net-new contacts from an initial pool of 1,343. The saved search refreshes weekly, automatically surfacing new contacts who meet the ICP but weren't in last week's batch.

02

Relational Lead Database — Airtable

Four linked tables manage the full lead lifecycle: Contacts, Companies, Deals, and Activities. An ICP score formula auto-calculates from company size, job title seniority, geographic match, and industry fit. Eight operational views give the broker instant visibility: Hot Queue (score ≥ 8), New This Week, Follow-Up Today, Active Conversations, Closed Won, and three reporting views. Airtable's native automation handles duplicate detection and status updates without requiring Make.com for in-system triggers.

03

4-Scenario Make Automation

Four scenarios handle the cross-tool handoffs: Scenario 01 ingests new Apollo contacts into Airtable with deduplication. Scenario 02 escalates hot leads — when an ICP score crosses the threshold, the contact moves to Hot Queue and the broker gets a Slack notification. Scenario 03 creates a Pipedrive deal automatically when a contact replies to the outreach sequence. Scenario 04 creates a ClickUp delivery task when a Pipedrive deal moves to Won. The broker touches none of these — they happen the moment the trigger fires.

04

AI Personalization Layer — Claude Haiku

For each contact, Apollo passes the LinkedIn headline and company description to a Claude Haiku API call. Claude generates a contextual opener — referencing the contact's actual role and likely real estate concerns — which is injected as the first line of the outreach sequence. 160 different opening lines, generated in seconds, each unique to the contact. Reply rates on this system in the CRE market: 6–12%, compared to the 0.5–2% typical of generic mass outreach.

160
ICP-targeted contacts from 1,343 total — precision filtered
4
Make.com scenarios — zero manual handoffs
28
real screenshots documenting the live system
6–12%
reply rate vs 0.5–2% for generic outreach

What the System Produced

A self-sustaining top-of-funnel pipeline. The broker's weekly time commitment dropped from 3–4 hours of manual prospecting to 30 minutes of reviewing the Hot Queue and responding to interested contacts. The pipeline fills continuously because the Apollo saved search runs on schedule. The broker never starts a week with an empty calendar — the system ensures there are always qualified prospects in the engagement phase.

The broader outcome: predictable pipeline. Instead of deal flow that depends on personal outreach volume, the system produces a consistent inflow of first conversations — which the broker converts at the same rate as before, with significantly less operational overhead.

What Makes This System Transferable

Every component adapts. The ICP filters change for each market. The Airtable schema reflects the specific deal structure. The Make scenarios are built around the client's existing tool stack. The Claude prompts are trained on the actual value proposition. What doesn't change is the architecture: a precision-targeted sourcing layer, a structured relational database, multi-scenario automation connecting the tools, and an AI personalization engine that makes every outreach feel specific.

The full system — with 28 real screenshots, all four Make scenarios, the complete Airtable schema, and the Claude prompt structure — is documented at the Work Sample page.

The takeaway: The system does not close deals. The broker closes deals. What the system does is ensure that the broker is always talking to the right people — so that closing rate, whatever it is, gets applied to a consistently full pipeline rather than a sporadically filled one.

// Frequently Asked Questions

Common Questions

The hardest part was not the technical build — it was the ICP definition. Without precision targeting, the automation produces volume without quality. Defining the right ICP (VP Real Estate, Director RE, CFO at logistics/industrial companies in FL and TX with 500–5,000 employees) required understanding the broker's actual deal patterns, not just applying generic B2B filters. Getting the ICP right took one week of research before a single scenario was built.

The full system — Apollo ICP configuration, Airtable 4-table schema with scoring formula, 4 Make.com scenarios, Pipedrive integration, ClickUp task automation, and Claude personalization — took approximately 3 weeks from first discovery call to live pipeline. Week 1 was ICP research and schema design. Week 2 was tool configuration and testing. Week 3 was scenario automation and QA with live data.

Minimal. Once the system is live, the primary maintenance is: (1) reviewing the Apollo saved search monthly to refine ICP filters based on actual reply quality, (2) monitoring Make.com scenario error logs weekly, and (3) updating the Claude prompt if the outreach angle needs adjustment. The Airtable database and Pipedrive pipeline maintain themselves through the automation. Monthly maintenance time: approximately 2 hours.

Yes. The architecture (Apollo → Airtable → Make → Pipedrive → ClickUp) is tool-agnostic and the logic adapts to any B2B market where decision-maker targeting, data enrichment, and multi-touch outreach are relevant. The ICP definition, Airtable schema fields, and Claude prompt context change for each market — the underlying automation structure does not.

Want this system built for your pipeline?

Discovery call is free. 15 minutes to scope your automation needs.

Book a Call →
← AI vs Manual Lead Generation
← Back to Insights