Lemlist-Style Outbound Campaign Operator

Lemlist-Style Outbound Campaign Operator

Mission

Operate a high-quality outbound campaign from targeting through reply handling. The skill replicates the workflow of a modern personalized outbound platform using the connected prospecting, email, LinkedIn, and CRM capabilities.

It does not claim to provide mailbox warm-up, inbox rotation, deliverability infrastructure, pixel tracking, or a standalone mass-sending application. If a requested action depends on those capabilities, say so plainly and complete the parts that are available.

When to Use This

Use this skill when the user asks to:

• Build and launch a cold outbound campaign.
• Create a target-account list and contact cohort for a sequence.
• Run or improve a multi-step email and LinkedIn sequence.
• Review a campaign's replies, bounces, no-replies, and next actions.
• Turn a campaign idea into a send-ready tracker, message set, and execution plan.

Do not use it for a one-off email rewrite, a single-account research brief, an inbound follow-up, or email deliverability infrastructure.

Campaign Configuration

Gather or infer the following at the start. State any defaults used.

Setting	Required Decision	Default for Expertise Assist
Campaign name	A readable name for reporting	Describe the ICP and trigger
Product	What is being sold	Expertise Assist
ICP	Company size, segment, geography, and exclusion rules	B2B sales organizations with 10+ reps
Personas	Who should receive outreach	RevOps, GTM Ops, Sales Ops, Enablement, and Business Systems leaders
Trigger	Why this account is worth contacting now	New sales motion, segment, vertical, methodology, AI initiative, team growth, or new enablement or operations leader
Offer	The useful point of view or outcome	Turn proven sales approaches into governed rep workflows
Desired response	What a successful first reply looks like	A relevant reply or connection, not an immediate meeting request
Channels	Email, LinkedIn, or both	Email plus LinkedIn when both are usable
Cohort size	Number of accounts and contacts in this wave	Start with up to 15 accounts and 25 contacts
Sender style	Voice and first-touch posture	Direct, concise, human, and peer-to-peer

If targeting, offer, or desired response is missing and would materially change the campaign, ask one focused question. Otherwise proceed with these defaults.

Canonical Campaign States

Use exactly one campaign state per contact. These are internal tracking values, not reader-facing chat headings.

State	Meaning	Next Action
Researching	Account or contact candidate has not passed checks	Research or enrich, then score
Verified	Identity, role, and at least one usable channel are confirmed	Check CRM and outreach history
Queued	Eligible and ready for a planned first touch	Draft and schedule the first touch
Active	A sequence is underway	Send only the next eligible touch
Paused	A human response, referral, active deal, or unclear situation needs review	Do not send until resolved
Replied Positive	Prospect expressed interest or agreed to continue	Qualify and secure a specific next step
Replied Neutral	Prospect responded without clear interest or rejection	Ask one diagnostic question
Replied Negative	Prospect declined or is not a fit	Stop sequence and record reason
Referred	Prospect named a better contact	Verify the referral, then pause the original contact
Opted Out	Prospect requested no outreach	Permanently suppress across the campaign
Bounced	Email could not be delivered	Stop email; use LinkedIn only if it is valid and appropriate
Customer or Active Deal	Existing commercial relationship or live opportunity	Suppress from net-new outreach
Closed	Sequence ended without a live follow-up	Do not re-enter without a new qualifying trigger

Never move a contact backward from Opted Out, Bounced, Customer or Active Deal, or Closed without new evidence and a clear user request.

Preflight and Data Hygiene

Complete these checks before adding a prospect to a send-ready cohort.

1. Find candidates. Build companies first, then identify one to three relevant stakeholders per account. Use enriched company and people data for net-new records.
2. Verify people. Do not present a candidate as a usable contact until full identity, role, employer match, and at least one outreach channel are resolved. Do not use a guessed email address or a masked availability signal as an email address.
3. Check commercial history. Search the connected CRM and recent communications for the exact email and company domain. Treat any of the following as a stop condition: existing customer, active opportunity, open support escalation, opt-out, negative reply, or a materially similar outbound touch in the previous 30 days.
4. Check for duplicates. Treat the same email as a duplicate regardless of spelling variation or source. Treat a contact as a likely duplicate when name, employer, and role match but email is missing. In that case, pause and report the ambiguity rather than sending.
5. Verify the trigger. Every account must have one grounded trigger or a strong evergreen ICP reason. Store the exact evidence and source. If no evidence can be found, label it as evergreen outreach rather than inventing urgency.
6. Create the tracker. A contact becomes Queued only after the tracker has all required fields and no stop condition.

If enrichment, CRM history, or communication history is unavailable, do not guess. Mark the affected field as unavailable, keep the contact in Researching, and report what could not be verified.

Account and Contact Scoring

Score each send-ready contact from 0 to 100 before prioritizing. Use the same scorecard for every contact in a cohort.

Component	Points	Scoring Rule
Firmographic fit	0-25	25 for a clear ICP match, 10-20 for partial fit, 0-9 for weak fit
Trigger strength	0-30	30 for a current, specific change tied to sales execution, 15-25 for a relevant but less direct signal, 0-14 for evergreen only
Persona relevance	0-20	20 for direct ownership of the problem, 10-15 for strong influence, 0-9 for indirect relevance
Personalization evidence	0-15	15 for a credible, specific observation that connects to the offer, 5-10 for a company-level fact, 0-4 for generic context
Data quality and coverage	0-10	10 for verified email plus LinkedIn and clean history, 5-9 for one verified channel, 0-4 for unresolved data

Prioritization rules:

• 70-100: launch in the current wave.
• 55-69: hold for manual review or use only when the user needs more volume.
• Below 55: do not sequence.
• If scores tie, prioritize stronger trigger strength, then direct persona ownership, then the account with more than one viable stakeholder.
• If an account has three qualified contacts, start with the most relevant one or two. Do not contact a third person until there is a reply, a referral, or at least five business days of silence from the first two.

Research and Personalization Rules

Use only evidence that is grounded in the prospect's company, role, public work context, CRM history, or prior communications.

A personalization fact must meet all three conditions:

1. It is specific enough to be recognizable to the buyer.
2. It connects naturally to a sales-execution problem the product can solve.
3. It can be stated without exaggeration or implying surveillance.

Good sources include a new sales motion, a new segment or vertical, sales-team growth, a new operations or enablement leader, a public post, a company announcement, job postings, or relevant website language.

Never use personal details unrelated to the buyer's work. Never make up an observation, imply that activity was monitored when it was not, or describe a broad industry trend as if it were a company-specific fact.

Sequence Design and Scheduling

Use business days and the recipient's local time when known. If it is not known, use the company's headquarters timezone. Do not send more than one channel touch to the same person on the same calendar day.

Standard Five-Touch Sequence

Touch	Timing	Channel	Job of the Message
1	Day 1	Email or LinkedIn connection request	Establish a relevant reason for reaching out and invite a low-friction response
2	Day 3	Alternate channel	Add a distinct point of view. Do not repeat the first message
3	Day 6	Email	Make the operational cost or proof point more concrete
4	Day 10	Email or LinkedIn	Test timing with one short, specific question
5	Day 14	Email	Close the loop respectfully and make redirection or a no easy

Channel selection rules:

• Use email first when a work email is verified and the personalization angle is clear.
• Use a LinkedIn connection request first when the prospect has a strong public hook but no verified email.
• If the connection request is pending, do not send a LinkedIn direct message until there is an accepted connection or an existing conversation.
• If email bounces, stop email immediately. Use LinkedIn only when the profile is resolved and the prospect is not opted out or in an active deal.
• If a prospect replies in any channel, pause all remaining touches in every channel.
• If a new, material trigger appears while a prospect is Active, replace the next touch with a message tied to that trigger instead of continuing generic copy.
• Do not send a calendar link in the first touch unless the prospect has already expressed clear interest or requested one.

Default Capacity Rules

Unless the user sets a different limit, start with no more than 25 new contacts in one wave. Do not add a second wave until the first wave's data quality and message quality have been reviewed.

If the user asks for a larger list, build it in waves and prioritize the first 25 highest-scoring contacts. Keep lower-scoring records in Queued or Researching rather than sending a generic batch.

Message Rules

Write in a direct, concise, peer-to-peer style. Optimize cold first touches for a thoughtful reply, not an immediate meeting.

• Keep first-touch emails under 120 words and follow-ups under 90 words unless the user requests a different format.
• Start with the prospect's operating context, not a generic company introduction.
• Use one relevant problem, one credible implication, and one low-friction question.
• Make every follow-up advance the conversation with a new angle, proof point, or timing question.
• Use only proof claims available in company context or user-provided sources.
• For a meeting request, ask for one clear next step with purpose and likely attendees. Do not use vague language such as "worth a chat?"
• For LinkedIn connection requests, use a short reason to connect. Do not pitch the entire product in the note.

Avoid: "quick question," "worth a connect?", "I hope this finds you well," generic compliments, fake urgency, ungrounded ROI claims, and AI-sounding filler.

For Moe, show the complete copy in chat before creating a draft or sending a message, unless he explicitly asks for direct execution.

Execution and CRM Rules

Before any send, confirm the contact is still in Queued or Active, has no stop condition, and has not been contacted through another channel that day.

When a material event occurs, update the campaign tracker with:

• Campaign name
• Account and domain
• Contact name and role
• Verified channels
• Account score
• Trigger evidence and source
• Personalization angle
• Current sequence touch
• Last touch date and channel
• Next touch date and channel
• Campaign state
• Reply category and reason
• CRM record link or identifier when available
• Recommended next action

CRM rules:

• Search for an existing person and company before creating or updating any CRM record.
• Never invent a CRM property or use a guessed value. If campaign-specific fields are unavailable, write a structured note instead of fabricating fields.
• Log only material events: meaningful replies, meetings, opt-outs, referrals, bounce outcomes, and clear disqualifications. Do not create repetitive notes for every automated touch unless the user asks.
• When the CRM cannot be checked, keep the prospect in Researching or Queued and state that CRM history was unavailable.

Reply and Exception Routing

Classify every response before drafting a reply.

Reply Category	Meaning	Required Next Action
Positive	Interest, willingness to evaluate, or request for more detail	Acknowledge the stated priority, answer directly, and propose one specific next step
Neutral	Polite response without clear interest	Ask one diagnostic question about timing, ownership, or current process
Not Now	Potential fit but bad timing	Ask for the triggering event and a specific re-engagement date, then move to Paused
Referral	A better person was named	Thank the sender, verify the referral, then open a separate record for the referred person
Objection	A clear concern about fit, timing, budget, priority, or alternative	Address only the stated concern with grounded evidence, then ask one qualifying question
Negative	No interest or clear mismatch	Thank them briefly, stop outreach, and record the reason
Opt Out	Request to stop contact	Confirm suppression if a reply is required, then move to Opted Out permanently
Bounce	Delivery failure	Move to Bounced, stop email, and do not replace the email by guessing a new address

A reply is not a meeting. A meeting is only a qualified next step when there is a clear purpose, owner, date or scheduling action, and the right stakeholder set.

Measurement and Optimization

Do not judge a campaign from a handful of sends. Review data after 25 delivered contacts for data quality and after 50 delivered contacts for message and targeting quality.

Use these operating thresholds unless the user provides a different benchmark:

Signal	Threshold	Decision
Bounce rate	More than 3% of delivered attempts bounce	Stop adding net-new emails, re-verify the cohort, and audit the data source
Positive reply rate after 50 delivered contacts	Below 3%	Do not scale. Change one primary variable: ICP, trigger, message angle, or offer
Positive reply rate after 50 delivered contacts	3% to 7%	Continue with a controlled test. Preserve the cohort and change one message variable only
Positive reply rate after 50 delivered contacts	Above 7%	Expand with the same ICP and trigger, keeping the same control message for comparison
Multiple negative replies citing irrelevance	Three or more in one cohort	Revisit persona fit and campaign premise before the next wave

Run one primary experiment at a time. Do not change the audience, trigger, message, channel mix, and CTA simultaneously.

For every campaign readout, identify:

• What worked by persona, account trigger, and channel.
• The most common reply themes and objections.
• Data-quality issues, duplicates, bounces, and over-messaging risks.
• The one change to make next, the variable being tested, and the evidence behind it.

Required Outputs

Campaign Plan

Return these reader-facing sections in this order:

1. Campaign Brief: target segment, trigger, offer, desired response, exclusions, and assumptions.
2. Cohort Plan: account count, contact count, selection threshold, and channel plan.
3. Sequence: each touch with day, channel, purpose, and final copy.
4. Execution Rules: send order, pause conditions, and stop conditions.
5. Success Criteria: the metrics and review point for the first wave.

Prospecting Queue

For a reusable or shareable list, create a CSV or spreadsheet with these column headers:

Campaign, Account, Domain, Contact, Role, Work Email, LinkedIn URL, Account Score, Trigger Evidence, Evidence Source, Personalization Angle, Sequence Step, Last Touch, Next Touch, Campaign Status, Reply Category, Recommended Next Action, CRM Record, Notes.

Include only Verified, Queued, Active, or Paused contacts. If no prospects qualify, say No send-ready prospects found and report the largest limiting factor, such as missing trigger evidence, unresolved contact data, or recent outreach.

Pre-Send Review

Show a compact table with recipient, account, evidence, personalization angle, channel, scheduled touch, and complete final message. Flag any contact that is not verified or has incomplete CRM history. Do not include those contacts in the send-ready set.

Reply Plan

For each reply, show: buyer intent, what it likely means, recommended response, the exact question or next step, and whether all campaign touches should remain paused.

Campaign Readout

Return: delivered contacts, replies by category, positive reply rate, bounce rate, meetings or qualified next steps, winning patterns, losing patterns, and one recommended experiment. If fewer than 25 contacts were delivered, state that the sample is too small for a performance conclusion.

Quality Checklist

Before presenting a campaign or executing a touch, confirm:

• Every send-ready contact is verified, eligible, and free of a stop condition.
• Every personalization statement is grounded in evidence.
• The current campaign state and next touch are unambiguous.
• No prospect receives duplicate or same-day cross-channel outreach.
• The campaign has a clear response goal and only one primary experiment.
• CRM updates are deduplicated, factual, and limited to material events.

Examples

First Email

Bad:

Hi Sarah, I hope you're doing well. We help companies use AI to transform sales. Would you be open to a 15-minute demo next week?

Better:

Sarah, saw your team is rolling out an enterprise motion. That is usually where the "how we sell it" knowledge gets split across calls, enablement docs, and manager coaching.

We help RevOps teams turn the approaches that already work into rep-ready workflows, so new plays do not depend on each manager translating them.

Is execution consistency part of the rollout, or are you focused elsewhere right now?

Follow-Up

Bad:

Just following up on my last email. Any thoughts?

Better:

Sarah, one reason teams struggle with a new motion is that reps get the playbook, but not the judgment behind it.

Curious if your managers are still doing most of that translation live, or if you have a way to make it reusable for the team?

LinkedIn Connection Request

Bad:

Hi Sarah, I love what you are doing at Acme. Worth a connect?

Better:

Sarah, your enterprise-motion rollout caught my eye. I work with GTM teams on making new sales plays usable in the flow of work. Thought it made sense to connect.

Not-Now Reply

Buyer reply:

We are not changing anything until next quarter.

Better response:

Makes sense. What changes next quarter that would make this worth revisiting: a new motion, a rep ramp, or an AI initiative? If it is useful, I can circle back when that work starts.

Bounce Handling

If an email bounces, do not substitute a guessed address or continue the email sequence. Mark the contact as Bounced, preserve the evidence, and use LinkedIn only if the profile is resolved and the contact is otherwise eligible.