Building LeadScrapper Pro.
One pipeline for discovery, audit, and outreach, built so the signal from the first scrape survives to the last draft.
Problem
Freelancers and agencies prospecting for local clients were spending hours finding businesses, manually checking their websites, then writing outreach from scratch: three disconnected steps that most tools only handle one of.
Research
Talking to freelance web designers and SEO consultants surfaced the same pattern: the audit findings were the actual pitch. A cold message with no specifics gets ignored; a message that names the exact broken thing on a prospect's site gets a reply.
Design
The product had to make the audit findings the center of the experience, not a footnote: a health score and specific flagged issues, positioned so a user could pitch with confidence from the first screen.
Engineering
The hard part was keeping signal attributable across three stages: discovery finds a business, an audit runs against its site, and outreach has to reference the specific finding without the pipeline losing track of which finding belongs to which business at scale.
Tradeoffs
The audit only checks what maps to an outreach line a user can actually send (speed, mobile, SSL, SEO, dead CTAs), not a full Lighthouse-style report. Coverage was traded for a latency budget under 30 seconds; see the Labs article on that decision.
Outcome
Shipped as a working pipeline in five days. Freelance and agency users report landing client work directly from audit-backed pitches within the first week of use.
Lessons
The first version ran every audit check sequentially before the latency budget forced a rework: a mistake documented plainly in the Journal rather than quietly fixed and forgotten.