multi-channel-outreach

Multi-Channel Outreach

Overview

This skill produces a day-by-day cadence where each channel has a job: email carries the offer and the proof, the phone call turns interest into a meeting, LinkedIn makes the sender's name familiar, and a postcard or letter reaches a physical desk. Every touch has a day offset and weekday, a clock time in the recipient's zone, copy written out in full, a skip branch and a cost; the standing checks are stated once. Every reply type and call outcome has a row that says what stops, who answers, how fast and the words to send. The sequencer setup is written as settings a team can turn on today.

The mistakes it prevents: buying and calling every mobile number Mobile Finder returns before tiering; calling someone who already said no on another channel; dialing from a caller ID that carriers have labeled spam; LinkedIn automation or invitation volume that gets the seat restricted; handing reps touches that say "see the copy doc"; and spending $0.25 per mobile number on contacts who will never be called.

When to Use

• "Add calls and LinkedIn to this cold email sequence."
• "Build a multichannel cadence for [segment]."
• "Find mobile numbers for these leads so my SDRs can call them."
• "Should we send postcards to these [dentists, roofers, property owners]?"
• "Which of these numbers should we call, and when?" (run the phone check, then hand off to cold-calling).

Use something else for nearby jobs:

Prerequisites

1. The email campaign plan (cells, offer, touch map) or enough to write one.
2. Who will call: in-house reps, hours available per day, dialer, and the caller ID numbers each rep dials from.
3. LinkedIn: which seats send, Sales Navigator or not, and a commitment to manual sending.
4. Direct mail: budget per piece, a print and mail vendor, and whether the address is a business or a residence.
5. The sequencer and CRM the team uses, so the stop rule and LinkedIn and call steps are written as that tool's settings.
6. Recipient time zones, so every call lands in Berman's call windows in the recipient's local time.
7. Deal value, close rate from a meeting, and the meeting rate per contact the user expects for each tier, so the break-even per contact decides which channels each tier gets.
8. The offer's public price or price range and what moves it, and one true proof point, so every touch can be written in full. With no price, the price answer uses {{price_range}} and the top note asks for it; with no proof point, the copy goes without one; never invent a result.
9. The user's own settings: list size and start date (without them, the dated launch is a labelled worked example), the email send window and the re-contact date for "not now" replies that give no date.

Authoritative Foundations

Workflow

1. Assign channel roles

One rep owns every channel for an account, so the stop rule works: a reply on any channel reaches the person who stops the other channels. Rules: references/channel-operating-rules.md section 3c.

Tools that run several channels, rated against LinkedIn's terms (vendor pages and help centers, 2026-10-13; detail in references/channel-operating-rules.md section 3a):

Plans from this skill use Low-risk LinkedIn touches only. The deliverable names only the tools the plan uses; this comparison goes into a "Tool Choice" section only when the user asked which tool to pick.

2. Decide which contacts get which channels, and thread the account

Spend on the expensive channels only where the deal value pays for it. Break-even per contact = deal value x close rate from a meeting x meeting rate per contact, from the user's numbers; a tier gets a channel only when its cost per contact (references/channel-cost-model.md section 5) stays under that break-even.

Promote to tier A on a dated trigger. For buyers around public money (contractors, engineering firms, suppliers), free non-ScraperCity sources supply the trigger: new federal awards on USAspending.gov, state DOT bid tabulations, and job postings for the role your product supports; for any segment, Sales Navigator's "Changed jobs in past 90 days" on the buyer title. Carry the trigger's own fields onto the row (award_agency, award_project, posted_role) so the copy names the award or the role, and write one copy code per trigger type. Recipes: references/channel-scripts-and-cadences.md section 10.

Every contact with a LinkedIn URL gets the connection request. Sales Navigator signals (posted in the past 30 days, changed jobs in the past 90 days, follows your company) only set the order: those contacts go first in the day's 10 to 25 requests (section 6c of the scripts reference).

Thread the account the way Berman does: start with the senior person (CFO, owner) and go one level down (controller, ops manager) only when the senior person has not replied in one to two weeks; he warns that emailing several people at one company at once can get you blocked company wide. In the default cadence the second person starts on day 16, the Tuesday after the senior person's last touch, so the two threads never overlap, and a reply from either stops both. Rules: references/channel-scripts-and-cadences.md section 11.

3. Run the phone check before buying or dialing numbers

For every number, in order. Any "no" removes the phone touch for that contact. Full checklist: references/channel-operating-rules.md section 1.

1. Is the number on your internal do-not-call list, or has the person opted out on any channel? Drop it. An email unsubscribe also stops calls and mail, the same day.
2. Has the person had 3 calls this cadence (the three calls in Berman's sequence)? Stop. Past five attempts with no engagement, Berman moves on to fresh names.
3. Is it inside Berman's call windows in the recipient's local time: 08:00 to 10:00 or 16:00 to 17:00, Tuesday to Thursday for the heavy blocks, never 12:00 to 14:00? Otherwise reschedule.
4. Is the caller ID the rep dials from labeled spam on carrier apps? Swap or fix the number first; pickups collapse once it is labeled.

4. Build the day-by-day cadence

Default 14 business day cadence for tier A, adapted from the six-touch sequence on Berman's cold calling page with a mail piece and more LinkedIn touches added. Day 1 is a Tuesday so every call lands Tuesday to Thursday; times are the recipient's local time. The email send window is the user's setting; test it against reply rates. The cadence is day offsets; dates belong to the launch calendar.

Checks, stated once above the table: one-click unsubscribe on every email; the phone check (step 3) before every call block; every LinkedIn touch done by hand in LinkedIn.

LinkedIn volume per seat stays under the community estimate of about 100 invitations a week on a rolling 7 days, sent 10 to 25 a day in small batches; invitations pending 3 to 4 weeks are withdrawn (Overloop, July 2026). One InMail per person: I1 is a single touch.

Stop rules: any reply on any channel stops every prospecting touch for that person and their company (Berman: stop the sequence on a reply; confirmation touches for a booked meeting still run); an opt-out on one channel is an opt-out on all, the same day. A contact who finishes with no reply re-enters only with a new dated trigger and new copy, never the same sequence again. Every touch has a skip branch so a contact missing a LinkedIn URL, a callable number or an address moves forward instead of stalling. SMS enters only after a reply that asks to text, as a confirmation touch; plans do not use WhatsApp.

4b. Write the sequencer setup

Write the stop rule, the LinkedIn rows and the call rows as settings in the user's sequencer. With no tool named, use HubSpot sequences, whose settings are documented in references/channel-operating-rules.md section 8: reply and meeting unenrollment set to "All contacts at the company" plus "Unenroll contacts from the same company"; "Sales Navigator - Connection request task" and "InMail task" steps for LinkedIn; "Call task" steps for calls; the rep unenrolls the contact when a call reaches a conversation (a workflow can do it on Enterprise).

5. Write every touch in full

A rep should be able to send every row without opening another document.

• Email: subject and full body for every email, under 75 words (Berman's templates page), one question, merge fields for name and company with a fallback ({{first_name|there}}) plus the trigger's own fields on trigger variants ("congratulations on the {{award_project}} award from {{award_agency}}", never "the new public work"). An email mentions another channel only when that touch happened (section 13 of the scripts reference).
• One copy code per trigger and persona (E1p and M1p for a public award, E1h and M1h for a hiring post). Never one body that says "X, or Y" to cover two cases.
• Calls and voicemails: opener, voicemail per call and front-desk line, from references/channel-scripts-and-cadences.md; hand dialer setup and coaching to cold-calling.
• Openers: never "Did I catch you at a bad time?" (Gong: about 40% less likely to book). Name and company, then "The reason I'm calling is...".
• LinkedIn: L1 under 200 characters so it fits every seat, L2 one line and one question, the InMail and the voice note written out word for word. Free seats get a few notes a month, so plan them with no note; notes at volume need Premium or Sales Navigator (Core $119.99 a month, 50 InMail credits).
• Pricing questions in a reply: one line with the user's public price or range and what moves it (or {{price_range}} when the user has not given one), then a 20 minute call with two times. No quote, proposal or price sheet by email (Berman's templates page).
• Mail: the full letter or postcard text, one offer, one URL or phone number and the rep's signature.

6. Map every reply and call outcome

Build the branch table from references/channel-scripts-and-cadences.md section 8: interested, accepts a time, price question, "send me something", forwarded internally, referral, wrong person or left, not now, uses a competitor, not interested, opt-out, angry, out of office, hard bounce, LinkedIn accepted and LinkedIn reply, and each call outcome (meeting agreed, prospect returns a call or voicemail, callback request, front desk, voicemail, wrong number), plus a request to text. Each row names the action on every channel, the owner and response time (5 minutes for hot replies and 2 to 4 business hours for the rest, from Berman's follow-up guide and templates page), and the text to send. Branch on replies, acceptances and call outcomes; never on email opens, which Apple Mail Privacy Protection inflates.

7. Confirm every booked meeting and set the daily run order

A booked meeting gets three confirmation touches from the same rep (Reisert): a call about a week out, an email 24 hours before, and a call 2 to 3 hours before when the email got no reply. The week-out call asks the buyer to bring the operational owner, the person who runs the process day to day; Gong found closed-won deals include 67% more contacts than closed-lost deals. Each rep works the day in this order: confirmations, replies and callbacks, cadence calls in the two call windows, LinkedIn tasks, then the phone check on tomorrow's contacts. Scripts: references/channel-scripts-and-cadences.md sections 9 and 12.

8. Price every channel, show the estimate and track

Use the cost model below. Get approval before any pull over about $5, and run check_wallet first. Record which channel produced each first positive response; report meetings, cost per meeting and connect or acceptance rate per channel, with a decision rule for each, and feed the numbers to outbound-unit-economics.

ScraperCity Implementation

Order of operations to avoid wasted spend: pull the list, validate emails, tier the contacts, run the phone check on tier A and B, then call find_mobiles only for contacts who passed it and have no contact_mobile already.

{
  "tool": "find_mobiles",
  "inputs": [
    "[email protected]",
    "[email protected]"
  ]
}

CLI: npx scrapercity mobile-find. REST: POST /api/v1/scrape/mobile-finder, poll /api/v1/scrape/status/{runId}, download the CSV.

Worked cost estimate (400 tier A local business owners)

Mobile Finder bills only on a hit, so the table shows the ceiling (every lookup found). Replace it with the expected cost after a 50-input Mobile Finder test (at most $12.50) measures the hit rate on this list.

Callable mobiles after the check = 120 + numbers found - matches on the internal do-not-call list and opt-outs. Owners without a callable mobile get business-line calls asking for them by name. Show this table and get approval before find_mobiles (it passes $5) and before the mail order.

Output Format

One Markdown plan built from templates/output-template.md, with these ## sections in order and no others: Summary, Channel Roles, Contact Tiers, Cadence, Tool Setup, Reply Branches, Channel Copy, Launch and Cost, Tracking. Add "Tool Choice" only when the user asked which multichannel tool to use.

• Top: the title, one sentence naming stand-ins to replace (product, reps, numbers, {{price_range}}), and one "Not checked:" line with everything assumed or not verified. Say it there once; nowhere else repeats "assumed".
• Summary: four sentences at most. The sequence first (tiers, channels, length, threading), then the launch it was sized for, the cost, and the spend to approve.
• Channel Roles: one row per channel used. Claims about how the segment behaves (who opens mail, who answers the phone) appear only as what a Tracking metric tests.
• Cadence: day offsets, not dates. One "Checks:" line above the table holds the standing checks once; the table has Tier, Day, Time, Channel, Touch (with its copy code), Owner and Skip when, plus the three confirmation rows. Under it: account owner per tier, stop rule, LinkedIn branch, a "LinkedIn volume:" line, and the daily run order ending with one line on the caller ID spam-label check.
• Tool Setup: three to five lines of settings in the named sequencer.
• Reply Branches: a row for every reply type and call outcome with the action on every channel, owner and response time, and the text to send ("None" when nothing is sent). The price row answers with a figure or {{price_range}}.
• Channel Copy: every code used in the cadence, written word for word, one code per trigger or persona. Emails as "Subject: ... Body: ..." under 75 words; no brackets or blanks.
• Launch and Cost: the dated cohort calendar, the data pull in two or three sentences, and the cost table with product names (Mobile Finder, Email Validator, B2B Lead Database), counts, unit prices and a total; rep-time lines name the user's inputs. Without a user list size or start date, the first line labels it a worked launch for the stated test size.
• Tracking: each metric with the decision it triggers; bounce and complaint limits live here only.

The reader receives the deliverable only, with each fact or check stated once. Leave out internal file paths, reference section numbers, tool ids, skill names, "Next:" handoff lines, checklists, self-scores, quality check sections, tools the plan does not use, tool risk ratings, and any law, regulator, registration or consent requirement.

Quality Check

python3 scripts/check-output.py plan.md enforces:

• The nine sections in order and no others (Tool Choice optional); one "Not checked:" line above Summary; "assumed" at most four times; no table cell of 40 or more characters repeated more than three times.
• A "Checks:" line above the cadence table naming the unsubscribe and the do-not-call list and opt-outs, and no Check column; no calendar dates inside Cadence; a caller ID line.
• Every cadence row has a day, a clock time, a copy code found in Channel Copy (profile views excepted), an owner and a skip branch; no connection request skipped for lack of a LinkedIn signal.
• Every email copy row has a subject and a body of 75 words or fewer (Berman's cap); no bracket placeholders; no copy that covers two cases with "X, or Y"; trigger tiers have copy with trigger merge fields.
• Reply Branches has at least 12 rows covering interested, price, referral, forwarded, wrong person, not now, competitor, not interested, opt-out, angry, out of office and bounce; with phone rows, also voicemail, front desk, wrong number and a returned call; the price row carries a figure or {{price_range}}.
• SMS rows run only after the prospect asks to text; LinkedIn touches are manual; a plan with connection requests states the accept-skips-email branch and a LinkedIn volume of 100 or fewer per seat per week.
• Tool Setup names the company-level stop on reply and, with LinkedIn rows, the task steps.
• One owner per tier across every channel; at least three channels; a stop rule; confirmation rows at a week, 24 hours and 2 to 3 hours, one asking for the operational owner.
• Launch and Cost has cohort dates, per-channel lines, a total, Mobile Finder at $0.25 per number found and Sales Navigator Core at $119.99 a month.
• No "bad time" opener, legal or registration term, internal path, section number, tool id, skill name, "Next:" line, LinkedIn automation tool, risk rating, checkbox list or quality check section.

Also confirm by reading:

• Calls fall in Berman's windows, 08:00 to 10:00 and 16:00 to 17:00 recipient local time, Tuesday to Thursday where the cohort calendar allows.
• No more than 3 calls per cadence to one person.
• The second person at an account starts only after the first person's touches end, and only if the first has not replied.
• Every number and rate is from the catalog, a dated source or the user's own inputs; rates not yet measured are stated as the user's hypotheses.
• Mobile Finder runs only after tiering and the phone check.

Common Pitfalls

Execution Artifacts

Related Skills

• cold-email-strategy and cold-email-copywriting set the email strategy and copy rules this skill writes to.
• cold-calling owns scripts, dialer setup and call coaching.
• reply-handling owns reply classification, response times and the full response library once someone engages.
• mobile-appender appends mobiles at CRM scale with the same phone check.
• local-business-list and scrapercity-real-estate-data supply addresses for mail.
• signal-triggered-outbound decides which contacts deserve tier A.
• outbound-unit-economics compares cost per meeting by channel.