--- license: MIT name: rlg-signal-to-sent-note skill: 05-05 description: Turn any public buying signal into the right person, a verified address, and a researched note that gets logged. --- # RLG Signal to Sent Note **When to use this skill:** Use this skill whenever a play has surfaced a signal (a job posting, a role change, a lookalike company) and you need to execute: find the right human, get an address you can trust, send a note worth reading, and log it so the system survives delivery week. Every public-signal play runs on this same pipeline. The signal arrives, the junk is stripped out, the search expands, the right person is found, an address is obtained and verified, the note goes out, and the work is logged. By hand this is about 45 minutes per live target plus 90 minutes a week of standing watch. An AI agent compresses the research stages to minutes; the judgement stays yours. **How to activate:** Paste this entire skill into your conversation with Claude, then provide the context requested in the Personalization Required section below. Claude will work through the methodology step by step. --- ## Personalization Required - The signal you are working. The posting, role change, or company, with its URL and where you found it. - The play it belongs to. Which angle produced this signal and what you intend to offer (see the companion RLG skills). - Who you sell to. Function, seniority, company band, so the title ladder lands on the right person. - Your tools. Which of these you hold: LinkedIn or Sales Navigator, an email finder (Hunter, Apollo, Prospeo, Findymail, Dropcontact, LeadMagic), a verifier (NeverBounce, ZeroBounce, MillionVerifier). The skill adapts the waterfall to what you have. - Your sheet. Where you log outreach. If you have none, the agent sets one up from the columns in Step 6. --- ## What This Skill Produces For each signal: a named primary contact with the reasoning, a verified address or an explicit channel decision, a researched note built on one public fact, and a same-day log entry with a follow-up date. Once set up: a monitoring routine that pushes new signals into one place on a fixed cadence. --- ## The Pipeline --- ### Step 1: Set up monitoring so the signal arrives A one-off search is not a system. Create saved searches with alerts, one per play angle: - Title patterns you sell to, or the adjacent titles that trigger your play. Keep each list under eight strings. - Date posted: last 24 hours for the alert, last 7 or 14 days when reviewing backlog. - Company size band you can serve, and geography you will actually cover. - Seniority filters matched to the play: senior for fractional-fit signals, mid for leadership-gap signals, junior for outsource signals. Name each search after the play angle so you know why you saved it. Review in two fixed blocks of forty-five minutes, Monday and Thursday. In the block: open alerts, walk new posts top to bottom, classify or skip, and write keepers into the sheet. You do not write outreach in this block. ### Step 2: Strip out recruiters and agencies You need a named company you can research and a human inside it whose problem you can name. Bin in fifteen seconds when: the posting company is a staffing firm and the client is confidential; the language is candidate-only boilerplate with no product, market, or org detail you could verify; the same advert text appears under three agency names in one week; or the company page has no employees, no product, and no site matching the claim. Try to unmask only when the role is a rare fit and the description has distinctive detail: search a unique sentence from the body in quotation marks. If the true employer surfaces in two minutes, keep it as a company-posted signal. If not, bin and move. Do not spend a research hour on a ghost client. ### Step 3: Expand the search from the posting copy This is the single most valuable move in the whole pipeline, and the one most operators never run. One distinctive sentence turns a single listing into a cohort. Copy one sentence from a keeper posting that is specific enough that a generic template would not contain it: a product claim, an unusual stack, a phrase like "first senior hire in X," a metric, an oddly precise hybrid-location line. Avoid pure cliches. Search it in quotation marks and sort the first page of results into three buckets: - Same role, other boards. Often reveals the employer an agency was hiding, or a richer applicant-tracking page with salary band and team detail. Update your record and prefer the ATS URL as canonical. - Same description at other companies. Whoever wrote the brief copied it, or a recruiter re-used a template across clients who share a problem shape. Each extra company is a ready-made lookalike. Add them as new rows with the same play angle. - The original careers page. Usually more text, sometimes a hiring-manager name. Harvest anything that improves the observation you will send. Stop after one strong sentence and one search page. Expansion is not a research hobby. ### Step 4: Find the person who owns the problem "Usually the CEO" is a starting guess, not a method. Enumerate the org: the company's LinkedIn People tab first, filtered to current employees, scanning for Chief, VP, Head, Director, Founder, Partner. Under about forty people, read the whole list; it is faster than guessing. Cross-check the website team page when LinkedIn is thin. Note who posted the job: that person is often HR or the hiring manager, a participant rather than the buyer. Then walk the title ladder. Start from the function named in the signal and walk up until you hit the most senior person who owns that outcome and still sits close enough to feel the pain. A Head of Marketing hire with no senior marketer above it points at the CEO or founder. A Marketing Manager hire under a vacant Head seat points at the CEO, not the manager candidate. A junior data clerk points at Head of Ops or Finance. A new Head of Sales with no demand counterpart points at that Head of Sales first, the CEO second. Prefer the decision-maker (can approve budget or stop a hire) over the participant (feels the pain or screens CVs). One primary contact is the default. Add a second only when the play explicitly needs two seats, and then two notes with two different angles, never the same paragraph twice. If you cannot name a human in twelve minutes, bin the row and protect the week. ### Step 5: Get to an address, using the waterfall A work email is the cleanest channel for a specific observation. Each finder only holds part of the world and the parts do not overlap neatly, so a single lookup that returns nothing means you asked one source, not that the address does not exist. Run cheapest and most certain first, stop at one verified address, and never run every tier on every name for sport. The cascade, in order: 1. Already known (free). A prior thread, calendar invite, proposal, or newsletter reply. Sixty seconds. If it is there, stop. 2. Pattern guess plus verify (free or low-credit). If you hold one good address at that domain, build the candidate from the name. Never send an unverified guess. 3. Public pages (free). Team page, press contact, author byline, conference bio. Rare for seniors. Still verify. 4. First paid finder (Hunter or equivalent). Strong on standard patterns, weak on rare formats and catch-alls. 5. Second finder with a different coverage map (Apollo or equivalent), often already open from Step 4. 6. Profile-based finder (Prospeo or equivalent). Takes the LinkedIn URL rather than name plus domain, so it resolves people the first tiers cannot see. 7. Verified-on-find tier (Findymail or equivalent). Returns fewer addresses and bounces less. Use when earlier tiers gave you something you do not trust. 8. Regional finder (Dropcontact or LeadMagic). For misses that are geographic rather than obscure. 9. Verify every candidate before send (NeverBounce, ZeroBounce, MillionVerifier, or your finder's verifier). Invalid means do not send. Catch-all or risky means the server accepts anything, so "valid" is not proof a human is there: prefer LinkedIn for first touch, or send only when the observation is strong and you accept a silent fail. Never blast several guessed names into a catch-all domain. An unverified guess can burn domain reputation you cannot un-burn. 10. LinkedIn as fallback, not default. A connection note or InMail when email fails or stays risky. Better for short observations, worse for logging and longer threads. Cap the waterfall at fifteen minutes. Then LinkedIn or bin. A compliance note for UK and EU senders: when you cold-email a named individual at their work address about a matter relevant to their job, the practical basis most operators rely on is legitimate interest under UK GDPR, with PECR in view. Keep the note relevant to their role, identify yourself clearly, offer an easy opt-out, and do not use scraped consumer addresses. If they opt out, stop. This is an operating standard, not legal advice. ### Step 6: Send, log, and follow up Channel order: work email first when you have a clean address; LinkedIn second when email is missing, risky, or the person lives on the platform. Do not double-send the same text on both channels the same day. If you use both, wait for silence, then send a shorter bump on the other channel that references the first. Follow-up rule: one follow-up only, four to six business days later, shorter than the first note, adding one new fact or offering to close the thread. Then stop unless they reply. A third chase is volume posture. The ceiling is two outbound touches per person per signal. Log the same day you send, in one shared sheet across every play: - Date, person, company, title - Signal source and play angle - Channel used, address or profile URL - The one-line observation you sent - Status (new, expanded, contacts found, address ready, drafted, sent, replied, closed) and next date - Outcome when it lands (replied, meeting, no, later) A sheet updated weekly is a diary. A sheet updated at send time is a system. --- ## Quick Reference: The Arithmetic | Stage | Time by Hand | |---|---| | Monitoring setup | ~60 minutes, once | | Standing watch | ~90 minutes a week (two 45-minute blocks) | | Agency strip | ~15 seconds per post, inside the watch block | | Expand | ~7 minutes per keeper | | Find the person | ~12 minutes per new company | | Address waterfall | ~8 minutes typical, 15-minute cap | | Draft | ~10 minutes | | Send and log | ~3 minutes | | One follow-up later | ~5 minutes | | Total per live target | ~45 minutes | Five researched sends in a week is roughly 3 hours 45 minutes of pipeline work plus the standing watch, about 5 hours 15 minutes in total. Across a year, the hiring plays alone come to about 270 hours for one client type. The tool inventory a working manual stack usually holds at the same time: Sales Navigator, one or two databases with finder tiers, a profile-based finder, a verified-on-find tier, a regional finder, and a verifier. Indicatively 330 to 540 pounds a month in 2026, several logins, and credits that expire whether you used them or not. There is a product category (FullEnrich, BetterContact) whose only job is to chain the finders for you, which exists precisely because a single source is never enough. --- ## Relationship to Other Skills rlg-network-sprint → produces its own list; uses Steps 5 and 6 for sending rlg-won-deal-lookalikes → hands its 18 companies to Steps 4 to 6 rlg-job-posting-four-plays → hands every keeper posting to this pipeline rlg-new-in-role → hands every verified role change to Steps 5 and 6 gtm-prospecting (Swan) → builds net-new account lists this pipeline can execute against gtm-humanizing-outreach (Swan) → the quality gate on every draft before Step 6 --- ## About This Skill This skill is the manual version of the tradecraft chapter from The 7 Network Plays to Fill Your Pipeline, the relationship-led growth playbook by Nynch. Nynch (https://nynch.com) runs this entire pipeline in the background, from signal to verified address to drafted note, so your hours go on the conversations rather than the plumbing. This skill is the same machinery, run by hand with your AI agent.