The $337K/quarter case at a glance
- Quarterly revenue: $337K (~$112K/mo)
- Outreach volume: 750 messages/day, ~45,000/quarter
- Reply-to-request rate: under 0.3%
- Job-site crew: 0 (100% subcontracted)
- Fixed monthly overhead: ~$6K (BD contractor + tooling + insurance base)
- Ticket size: $20K–$500K per project, brokered at 15-25% margin
The stack. Python + Playwright scraper against municipal open-data portals, Instantly or Smartlead for sending, Clay for enrichment, Claude/GPT-5 to classify filings as funded-but-not-yet-staffed, Google Sheets as CRM. The operator himself is a regional GC on paper. In practice he's a two-sided broker matching property managers with unspent budget to subs with open calendar. No dashboard, no SaaS layer, no app. One question on repeat: "Do you have any upcoming works?"
What this construction broker is actually selling
The product isn't the cold email. The product isn't even the construction. The product is a filtered list — a scored slice of public municipal data telling him which project managers just got funded for work they haven't hired for yet.
That's why the copy is deliberately dumb. When your list is pre-scored on loans, permits, and bond issuances, the message doesn't have to be clever. It has to be short and land in an inbox belonging to someone who genuinely does have upcoming works. The same 750 messages against a random purchased list is spam, and Exchange kills you by Thursday. Against a filings-scored list, it's a pipeline.
Volume gets the credit. The filter does the work.
The economics stack neatly on top. Broker margin on tickets that average $20K–$500K means his conversion can be dogshit — under 0.3% — and he still prints. Fixed costs are $6K/mo. No WIP capital. No payroll. No trucks. Geography is uncapped because he doesn't own the crews.
Every boring business you envy is quietly a data business wearing a costume. His costume is a construction license. The asset is the CSV.