Case Study
From lead search to
sent, tracked outreach.
A full-cycle outreach workspace that finds prospects, enriches each lead, scores fit by lane, drafts three email variants, audits tone, gates every send, delivers through Gmail under hard limits, and tracks replies to outcomes.
See OutreachIQ in motion.
The full pipeline in about a minute: search, enrich, score, draft, audit, gate, send, track.
What is OutreachIQ?
OutreachIQ is the internal system I use to run real outreach end to end: ICP recommendations, Google Maps prospecting, Firecrawl and Apollo enrichment, Claude lane scoring, three-variant email drafting, a local tone audit with deterministic repair, a human review queue, and a send gate in front of throttled Gmail delivery with follow-ups, reply and bounce detection, and outcome tracking. It has processed 1,600+ real leads and sent 320 emails through the gate. This page is the build story. If you want to run it yourself, it is open source: see the OutreachIQ product site.
What was broken.
The first version of OutreachIQ solved lead collection: search, enrich, score, draft. The moment it started working, the real problem moved. Quality control at scale is hard: choosing the right market, avoiding repeated contacts, keeping drafts specific, and catching risky tone before anything is sent. A good lead is easy to lose in a long table. A scraped inbox can appear on two different companies. A draft can sound generic, ask for too much too early, or open with a greeting that clearly came from a company name instead of a person.
Sending is where outreach systems usually get dangerous. Without hard limits, one bug or one over-eager afternoon can burn a sender domain. Without an audit trail, nobody knows which variant actually went out. Without systematic follow-up, warm threads go quiet, and replies get lost in a busy inbox. The system needed the same discipline after the draft as before it.
What was built.
OutreachIQ treats outreach as a controlled workflow rather than a pile of prospects. It starts with ICP recommendations and market presets, including a partner-lane mode for finding agencies that may need overflow brand identity support. Each new lead carries its search context forward, so Claude scores direct client prospects and agency partners differently.
Enrichment runs before scoring. Firecrawl extracts website summaries, brand signals, and candidate contact emails. Apollo adds company size, industry, founding year, revenue band, keywords, and company LinkedIn context. The dashboard stores enrichment snapshots, shows missing-field audits, and lets a single lead be re-enriched without restarting the search.
Email writing produces three short variants instead of one long draft: observation, peer, and opportunity. A local tone audit checks each variant for risky greetings, hard CTAs, repeated phrases, portfolio links in the body, and weak evidence. Flagged leads move into a review queue, and a deterministic repair flow rewrites weak drafts without extra AI calls, so fixing copy costs nothing and stays predictable.
The send layer is where the system earns trust. A send gate checks blockers and warnings before anything leaves: duplicate emails already contacted, missing contacts, unresolved audit flags. Approved emails go out through the Gmail API with 20 seconds between sends, a hard cap of 30 per day, a kill switch, and a do-not-contact list. After sending, the system bumps non-replies with threaded follow-ups, polls the inbox to detect replies and bounces automatically, and records every lead through prepared, sent, replied, meeting, and converted.
I still decide what goes out. The system does the repetitive search, enrichment, scoring, drafting, and checking, and the gate enforces the limits, so my job at review time is judgement, not bookkeeping.
Architecture in plain English.
Learn the complete OutreachIQ workflow.
Open only the lesson you need. Videos stay unloaded until you choose to play them.
Lesson 01
Search and enrich
0:58
Read transcript
How OutreachIQ finds and enriches your leads. Open Search, set your service priority and market, choose an industry niche or a custom query, then run it. Companies come back ranked into your pipeline, and enrichment runs automatically: Firecrawl reads each company website and Apollo adds size, revenue, and context, so every lead arrives ready to score.
Lesson 02
Score and draft variants
0:48
Read transcript
How OutreachIQ scores and drafts. Every enriched lead gets a score from zero to one hundred by lane, so the hottest leads rise to the top. For each one, Claude drafts three short variants, not one generic email: Observation opens with something specific, Peer speaks studio to studio, and Opportunity leads with a clear offer. Each draft is short, on brand, and ready to edit or refine in one click.
Lesson 03
Audit, review and send
0:44
Read transcript
How OutreachIQ protects every send. A local tone audit reads each draft and flags generic greetings, tired phrases, and risky lines. Weak drafts repair themselves in one click, with no extra AI calls. Flagged leads wait in a review queue, and nothing sends until you approve it. Then the send gate checks every blocker before Gmail delivers, twenty seconds apart, thirty a day, with a kill switch. The system drafts and checks. You decide what goes out.
Lesson 04
Replies and follow-ups
0:39
Read transcript
How OutreachIQ handles what happens after the send. Every message is tracked straight from Gmail: sent, replied, or bounced, the status updates itself. Replies rise to the top of your pipeline so a warm lead never gets buried. Leads that stay quiet get a follow-up, spaced on your schedule and never pushy. A reply pauses the sequence, and you can stop any thread at any time.
Lesson 05
Read your results
0:33
Read transcript
How to read your results. Every step adds up to one honest funnel: companies searched, leads enriched, drafts written, and the emails actually sent through the gate. Every number is a real total counted from your own data, with no vanity metrics, so you always know the system is working and where to tune it next. Search, enrich, score, draft, audit, and send, all in one calm dashboard.
See it in action.
A guided tour of the current system: search and enrichment, lane scoring, three variants, the tone audit, the review queue, and a simulated gated send.
Guided walkthrough with sample data only. The demo does not include real prospects or private contact details.
The useful pattern is not blind autonomy. OutreachIQ uses AI for the repetitive search, enrichment, scoring, and drafting layer, while the human keeps the judgement layer before anything goes out.
Quality controls wrap the whole path now: from preset to enrichment to scoring, and from review gate to throttled sending and tracked outcomes.
Built with.
What changed.
The numbers above are live pipeline totals from the real system, not projections. What changed is not volume for its own sake: every one of those sends passed enrichment, lane scoring, a tone audit, and a gate before Gmail was allowed to deliver it, and every reply and bounce is detected automatically and tracked to an outcome. OutreachIQ turned outreach from a willpower problem into a reviewable pipeline that keeps its reasoning trail.
Want a system
like this one?
Book a free 30-minute call. We map your situation, scope a similar build, and agree on a fixed quote before anything starts.
Book Free 30-min Call