cold-email-list

Cold Email List

Overview

This skill delivers the file a sequencer imports on launch day: one row per prospect, every email valid and validated within four weeks of its send date, duplicates and suppressed contacts removed, no more people per company than the user agreed, personalization columns filled or deliberately blank, headers that the chosen platform accepts, and a row count that the user's mailboxes can send on schedule. A separate catch-all file and a short report come with it.

The mistake it prevents is the half-finished list: a database export pasted into Instantly with unvalidated emails, five people from one company in the same campaign, a {{first_name}} column that is blank on some rows, and 4,000 prospects for three mailboxes. Each of those costs bounces or a week. Alex Berman splits cold email results into 30% targeting, 30% offer and 50% deliverability; this skill covers the targeting share and protects the deliverability share before the first send.

When to Use

• "Build me a cold email list for this campaign."
• "Here are 400 domains, get the right people and make it ready for Smartlead."
• "I need 1,500 verified CFOs at construction companies for Instantly."
• "Take this ICP list and make it send-ready."
• A campaign plan from cold-email-strategy exists and needs its file.

Use a different skill when:

Prerequisites

Authoritative Foundations

Details: references/framework-notes.md.

Workflow

1. Write the campaign brief

One table: offer, segment, sequencer, mailboxes, sends per mailbox per day, steps, launch date, personalization variables the copy will use, countries. Every later number comes from it.

2. Size the list from capacity

daily_sends     = mailboxes x sends_per_mailbox_per_day          (Instantly recommends 30; Berman: under 50)
total_sends     = send_ready_rows x steps
sending_days    = ceil(total_sends / daily_sends)
rows_for_N_days = daily_sends x N / steps
pull_rows       = target_send_ready / yield, rounded up to the next 100

yield is send-ready rows divided by rows pulled on the user's last list for this segment; with no history, pull a first wave and measure it. Six mailboxes at 30 a day send 180 emails a day; with 3 steps that is 60 new prospects a day, or about 1,200 for a 20-day campaign. A list much larger than the campaign window ages past four weeks before it is sent; build it in waves.

3. Pick the sourcing path and count

Show the count, the planned pull and the cost estimate. Call check_wallet. Wait for a go-ahead before any paid step.

4. Pull with cursor paging

Page with after: "0", limit: 100, then pagination.next_after until null or the planned rows. Add excludeDelivered: true when the account has pulled this ICP before and only new people are wanted. For domain lists, rank people per domain and keep the best by role, up to the contacts per company the user agreed (role ranking in decision-maker-finder).

5. Fill gaps

Leave autoValidateEmails off on Website Finder; validation runs once in step 7. Record a miss in the gaps file; no address is built from a naming pattern. Order follows price per usable result: a database email costs $0.00169 / valid share to validate into a usable address (about $0.0021 at the example's 80% valid share) against $0.05 from Email Finder and ($0.0499 + $0.00169) / valid share from Website Finder plus validation.

6. Normalize, dedupe, suppress

Normalize per references/list-quality-standard.md section 3. Dedupe on lowercased email, then LinkedIn URL, then first name + last name + domain. Suppress in this order: unsubscribes, bounces, complaints, current customers, open opportunities, competitors, anyone in another active campaign, anyone contacted inside the user's re-contact window. Count every drop by reason. Drop rows with no first name from a personalized campaign or move them to a non-personalized variant.

7. Validate

Send every email that did not come from Email Finder to validate_emails ($0.00169 each). Route: valid to the send file; valid_catch_all to a catch-all file for a separate campaign at a daily volume the user sets, paused if its bounces pass 2% (Berman), highest-evidence rows first (high email_quality, an address format that matches other known addresses at the domain, no security gateway); unknown held for one retry on a date in the report; invalid and disposable removed and added to the blocklist.

8. Cap and order per company

Keep no more contacts per company_domain than the user agreed, highest buying role first. Write company_contact_order (1, 2, 3...) so the sequencer can stagger them: contact 2 starts only after contact 1's sequence ends without a reply, per Berman's one-person-at-a-time guidance. Instantly and Smartlead both run campaigns per list, so put later contacts in a second wave file.

9. Personalization columns

Add only columns the copy uses, and only facts with a source: industry, city, employee_band, case_study (chosen per segment), trigger (account level, from account-researcher or a signal list). Smartlead's import guide says a blank cell means the value does not appear in the email, which leaves "Hi ," in the inbox; give every variable a fallback in the copy or drop rows where it is blank. Never generate facts about a prospect to fill a column.

10. Format for the sequencer

Map columns to the platform's headers with references/sequencer-import-specs.md: UTF-8 CSV, header row first, unique headers, Instantly headers of 20 characters or fewer starting with a capital letter, no more than 50 variables (Instantly) or 40 custom variables (Smartlead). One file per campaign. Turn on the platform's duplicate check across campaigns and import the blocklist first.

11. Report and launch checklist

Before export, read a random sample (size set by the user) against the brief: title in the agreed family, company in the ICP band, person still at the company on their profile or the team page, domain opens the company's own site. Misses that repeat on one check mean the filter or mapping is wrong: fix the query and re-pull instead of patching rows. Record the result as one launch-checklist line ("19 of 20 title fit, 20 of 20 company fit, 18 of 20 still there").

Fill templates/output-template.md, run the launch checklist in the reference, and run scripts/check-output.py.

ScraperCity Implementation

Calls

ICP count:

{"title": "CFO, Chief Financial Officer, Controller, VP Finance, Director of Finance",
 "notTitle": "assistant, intern, former, retired, advisor, board, accounts payable",
 "industry": ["construction"], "minEmployees": 50, "maxEmployees": 500,
 "country": ["United States"], "hasEmail": true, "limit": 1}

Pull: the same body with "limit": 100, "after": "0", then each pagination.next_after. Confirm the industry label with the count; if it returns 0, use keywords instead.

Domain batch:

{"companyDomain": "ridgelinebuilders.example, coppercreekcontracting.example",
 "title": "CFO, Chief Financial Officer, Controller, VP Finance", "notTitle": "assistant, intern, former",
 "limit": 100, "after": "0"}

Website Finder for empty domains: {"domains": ["..."], "jobTitle": "Controller", "autoValidateEmails": false}. Email Finder: {"contacts": [{"first_name": "Lena", "last_name": "Ortiz", "domain": "graniteforge.example"}]}. Validation: {"emails": [...]} for a list that fits in one tool argument; for larger files the CLI's email-validate command with a one-address-per-line file, or POST /api/v1/scrape/email-validator.

Worked cost estimate

1,200 send-ready finance leaders at US construction firms, plus 140 priority domains from the user's CRM:

Ceiling shown before the run: $2.20 (Website Finder, 44 domains) + $0.75 (Email Finder, 15 names) + $2.90 (validation, up to 1,715 emails) = $5.85. Run check_wallet, show it and wait for a yes. Cost per send-ready row in the example: $4.07 / 1,247 = $0.003.

Error handling

Output Format

1. send-ready-[segment]-[date].csv: valid emails only, headers mapped for the chosen sequencer, plus email_status, email_validated_at and company_contact_order kept in a parallel master file.
2. send-ready-[segment]-[date]-wave2.csv for second and third contacts per company.
3. catch-all-[segment]-[date].csv for a separate campaign at the user's daily volume.
4. suppression-additions-[date].csv with invalid and disposable addresses for the blocklist.
5. The report from templates/output-template.md: campaign brief, sourcing plan, count check, pipeline funnel, validation, send capacity, sequencer import map, cost, sample rows, launch checklist.

• What the reader receives is the deliverable only, in product names (Lead Database, Email Validator, Email Finder, Mobile Finder, Website Finder): no tool ids, JSON tool calls, run IDs, skill names, file paths, reference section numbers or runbook IDs; a downstream step is written out as the checks or commands the team runs, never handed to another skill by name; no tables of checks that were not run (one "Not checked:" line instead); and no repeated identical rows where one template plus a list will do. scripts/check-output.py fails a report that carries a tool id, skill name or path.

Quality Check

• Count calls ran and the approved estimate appears before any paid step.
• Funnel adds up stage by stage with a reason on every change.
• Every row in the send file is valid, validated within four weeks of its send date or given a re-validation date.
• 0 duplicate emails, 0 suppression matches, no more contacts per company than agreed, contact 1 only in wave 1.
• Every personalization column used in the copy is filled on every row, or the copy has a fallback.
• The QA sample result appears in the launch checklist, and any repeated miss led to a re-pull.
• Send capacity: mailboxes x sends per mailbox equals daily sends, sends per mailbox is under 50 (Berman), and the sending days fit the campaign window.
• Import headers are unique, under 20 characters and accepted by the platform.
• scripts/check-output.py report.md prints PASS.

Common Pitfalls

Execution Artifacts

Related Skills

• icp-list-builder for the ICP definition and filters behind the pull.
• decision-maker-finder when the user has accounts and needs buying roles.
• email-list-cleaner to re-validate a list that missed its launch window.
• local-business-list for small-business campaigns before import prep.
• cold-email-strategy for the offer, segment and sequence this list serves.
• cold-email-copywriting for the emails that use the personalization columns.
• sending-platforms for upload settings, blocklists and auto-pause.
• email-deliverability for domain and mailbox health before launch.