Problem
SDRs needed 500+ qualified clinic leads a week across three brands. Purchased lists were expensive, stale, and nobody knew which ones worked.
Constraints
Healthcare targeting rules: an ICP drawn from closed-won deals, never touch a company with an open deal, no geographic filtering. On top of that, HubSpot’s 10K search cap, vendor rate limits, and no dedicated engineering help.
What I designed
A pipeline, not a list. It finds underpaid clinics in reimbursement data, scores them against each brand’s ICP, checks that the practice still exists and isn’t owned by a health system, enriches decision makers through an email waterfall, and dedupes against the CRM. Then it builds a branded report for each clinic showing what it’s leaving on the table, pushes everything to HubSpot, and enrolls each lead into the right rep’s sequence every weekday at 8am.
How it works
A Node service on Railway with no runtime dependencies, built with Docker and deployed on every merge. Separate modules handle pulls, enrichment, SDR enrollment (with a kill switch and no double enrollment), pool performance and proformas. A Claude skill wraps the whole flow, so someone who isn’t an engineer can run any brand’s pull end to end.
I took over the first version in August and rebuilt it. I’ve written about 290 of the roughly 300 commits since.
Result
- Loaded cost per lead under $20. The compute bill was a rounding error.
- 100+ booked meetings in the first seven weeks.
- A re-engagement screen of historical leads produced five live lists. Verification disqualified 30% of leads before a rep ever dialed.
- Enrollment went from “someone remembers to run it” to a daily job. Missed days went to zero.