Outcome Playbook · Mission 3 · Cold Outreach
Build a cold outreach agent that books 2-5 calls per week while you stay out of the inbox.
I build a cold outreach agent. A web dashboard where I press one button to scrape leads, one to load them into Instantly, one to launch. A 5-step sequence runs for 14 days with A/B testing. I read the AI insights once a week and adjust.
Your outreach dashboard lives at /admin/outreach. Your Instantly campaign is running. Your scoreboard shows 5 closed rows with real revenue.
One day to build the dashboard with Claude Code. Then 2-6 weeks of weekly batches until you hit 5 clients.
Instant implementation package: downloadable Claude Code skill, dashboard build map, 5-step sequence templates, and launch checklist. Free OpenClaw skills used inside. 30-day refund — see /refund.
The one number
2 to 5 booked calls per week.
That's the metric. Not opens, not reply rate, not list size. Those numbers exist to feed this one. At 1-2 closes per week, you hit 5 paying clients in 3-5 weeks.
Everything upstream (subject lines, sequence copy, targeting, deliverability) is a lever you pull to move that number. The AI Insight panel tells you which lever.
Why 5 clients, not 1
One client is a fluke. Two could be friends doing favors. 5 is the floor where "I might have something" becomes "this works." 5 unlocks three things:
- →Case study material. 5 different stories instead of one.
- →Real pricing data. What got accepted, what got pushback, what made them ghost.
- →A pattern. Same objections, same words that made them lean in. That pattern becomes your script, your landing page, your YouTube hooks.
The cold outreach agent
One skill. /cold-outreach-agent
Drop the skill file into your Claude Code ~/.claude/skills/ folder, then type /cold-outreach-agent. Claude reads your business.md and scaffolds the whole operation for you.
What gets built
- • Web dashboard at
/admin/outreach - • 3 API routes: scrape · load · activate
- • 2 Supabase tables (campaign config + batch records)
- • 5-step email sequence customized to your niche
- • AI Insight panel that diagnoses each batch
What you already have
- ✓ Apify account + API token (free tier is enough to start)
- ✓ Instantly account (or equivalent SMTP fleet)
- ✓ Supabase project (free tier works)
- ✓ One warmed-up sending domain with SPF/DKIM/DMARC
- ✓ A
business.mddescribing your offer in your voice
Download the skill file (ships with the playbook)
cold-outreach-agent.md — drop into ~/.claude/skills/, run /cold-outreach-agent.
What the dashboard does
Three buttons. One weekly review. That's the whole operation.
Scrape section
"Scrape 100 Leads" button. Calls Apify Apollo actor (300M+ B2B database). Filters by job title, company size, country. Personal emails prioritized for deliverability. Auto-saves to Supabase. 30 seconds per batch.
The 5-step email sequence
Day 0 (A/B test pain-point hook + CTA to free resource) → Day 3 (social proof with specific numbers) → Day 6 (honest question) → Day 9 (the simplest 4-step version) → Day 13 (breakup + free resource + YouTube link). Every email under 100 words. Subject lines under 6 words. Humanizer-checked. No em-dashes. No AI slop.
Launch control
Select daily send limit (10 / 20 / 30 / 40 / 50 per account per day) → "Load to Instantly" button (transfers Supabase → Instantly via API) → "Launch Campaign" button. Scrape. Load. Launch. Three clicks. Live.
14-day timeline
Visual of which days emails go out. Days 0, 3, 6, 9, 13. Once you press launch, the agent sends for 14 days without you touching it.
AI Insight panel
After each batch's first week, 4 diagnostic cards: (1) Subject line analysis + 3 alternatives if open rate <40%. (2) Email body analysis — flags too-salesy / too-long / too-generic. (3) Target audience fit — were replies from the right titles? (4) List quality score + bounce check. Ends with next-batch recommendation: "Keep subject A. Rewrite body paragraph 2. Filter to <50 employees. Increase daily to 30."
Performance gauges
Three live bars. Open rate: >40% green, 30-40 amber, <30 red. Reply rate: >3% green, 2-3 amber, <2 red. Bounce rate: <2% green, 2-3 amber, >3 red. One glance, one decision.
Campaign history
Every batch recorded: date, search query, lead count, open/reply/bounce, what you changed, status. Look back at 10 batches and see exactly what improved and why.
Lead list
Airtable-style table: email, name, company, type (personal vs company), current step, step status, send dates. Sortable and filterable.
The reframe
The real skill is not sending.
It's tracking.
Sending emails is easy. AI writes them for you. That's not the hard part.
The hard part is tracking what works and making decisions from the data. Which subject line got opens? Which body copy got replies? Which list produced the right people? The AI Insight panel gives you the answers. Your job is reading them and deciding what to change next batch.
A template that got replies last week might stop working next week. People get used to patterns. A "quick question" subject line used to get opens — now everyone knows it's an ad and skips it. That's the outreach game. It keeps changing. The founders who win are the ones who notice the shift and adjust before their numbers tank.
Templates are the floor, not the ceiling
We ship winning templates. They work. But they're your starting point. After 2-3 batches, you start writing variations based on what the data tells you. Your version outperforms the original because it's built from your replies, yourniche, your voice. The template got you moving. The iteration gets you clients.
Your weekly workflow
15 minutes a week. That's the whole operation.
After the dashboard is built.
Open /admin/outreach. Check for new replies. If someone wants a call, respond. Close the tab.
Read the AI Insight from last batch. Check gauges. Scrape next 100 with adjusted filters. Load. Launch.
Review all batches. Identify winning subject + sequence. Update config. Scale daily send limit if metrics are green.
The numbers to hit
Every metric has a floor. Below the floor means something is broken.
Week 1 — validation sprint
Week 2+ — scale
Most common bottleneck: copy. Your cold email is too long, too generic, or too polished. The AI Insight panel catches this. Rewrite it shorter and more human. Run the next batch. Check the delta.
The free OpenClaw skills in the stack
The skills are free. The roadmap is the product.
This playbook uses 5 free OpenClaw skills. All are Fleet Shield scored. You can inspect each one, download it, and run it yourself — free. The paid playbook is the sequence + templates + dashboard spec + the weekly decision loop that turns those skills into 2-5 booked calls per week.
From where you are — to where you want to be
The playbook is the bridge. Here is each side of it, in the buyer's own words.
- →A TAM list of 2,000+ accounts and a calendar week where you sent 40 personalized emails, maybe.
- →Sequences that die at email 1 because writing emails 2-5 feels like homework.
- →Replies sitting in a crowded inbox — meeting-intent looks the same as spam.
- →No idea which subject / offer / persona is actually producing booked calls.
- 5-email sequences generated per lead at the rate of the full TAM, not 40/week.
- Reply classifier separating hot / auto / unsubscribe — you only touch the hot ones.
- Booked-calls counter moving daily by Day 7, not "maybe next quarter."
- A per-vertical control panel showing which segment is actually converting.
Florida Digital Marketing Experts (FDME)
We had the TAM (thousands of FL contractors) and a proven offer, but writing personalized sequences at scale was the bottleneck. We were booking 2-4 meetings per week instead of the 10-15 the list could support. Discovery calls felt random because targeting was "whoever replied first."
Same scored Cold Outreach Pack + the sequence templates in this playbook + the /admin/outreach dashboard we built with Claude Code. AI writes the sequences per lead. Hot replies ping Telegram — we only touch those. Weekly cadence moved from "wrote emails on Sunday" to "reviewed booked meetings on Monday."
Full disclosure: We run our own agency (FDME) on this stack. This is our operating playbook, not a paid customer case study. Customer case studies will publish as they come in — the mechanism and templates do not change.
What's included — and what it would cost to build alone
Every line below is part of the implementation package. The prices on the right are what this planning, testing, and setup work normally costs when you hire it out or buy each tool separately.
First-meeting guarantee
Build the dashboard inside Week 1 with deliverability set up and a list of 500+ verified contacts. If you have not booked your first meeting by Day 21 of active sending, email us — we refund the $29 and walk through your targeting + sequence one-on-one until we find the bottleneck. The one thing we won't refund: not actually running the sprint.
Full terms at /refund. Email hello@openclawskillpacks.com if a step breaks and we'll unblock you before you ask for a refund.
The two traps that kill this playbook
Both are procrastination in disguise.
Trap 1 — Tool shopping
Your first batch gets 1% reply rate. You start googling "best cold email tool 2026." You sign up for 2 new services. Stop. The tool is not the problem. Your copy or your targeting is. The AI Insight panel already told you what to fix. Fix that. Not the tool.
Trap 2 — Fallback to "content marketing"
Two batches underperform. You decide "cold email is dead" and start posting on LinkedIn. Three months later: 40 followers, zero clients. Content marketing before you have 5 closes is procrastination dressed as strategy. Fix the cold email. Run another batch. Graduate when you have 5 paying clients.
The 60-second self-test
Before you call this playbook complete, check all 5.
Hit all 5 and this pack is done. Move to Content Engine or Appointment Booking.
About Stripe (optional for now)
The /cold-outreach-agent skill does NOT install a Stripe MCP server. That's on purpose — before you have revenue, connecting Stripe is one more key to manage for zero value. After your first close, come back to the Claude Code panel and type "install the Stripe MCP server and save my keys to .env." Until then, accept payment however works (invoice, bank transfer, Stripe dashboard by hand) and record the dollar amount in the scoreboard manually.
Get the playbook. Build the agent. Book the calls.
The OpenClaw skills are free. The /cold-outreach-agent skill file, the 5-step sequence, the AI Insight prompts, the dashboard spec, and the weekly decision loop — that's the $29.
Instant implementation package. Preview access. 30-day refund details at /refund.
Shut up and send.