HeyReach LinkedIn Outbound Command Center

Critical insight: Connection requests WITHOUT a note often outperform those WITH notes in 2025-2026 testing. The no-note approach gets 5-15% higher acceptance on average. Test both for your ICP.

MCP Tool Reference

All tools use prefix: mcp__760e7cd2-51c1-4b40-803b-a8c1d427bfb6__

Campaign Operations

LinkedIn Account Operations

Lead & List Operations

Lead Intelligence

Conversations

Analytics & Webhooks

Webhook event types: CONNECTION_REQUEST_SENT, CONNECTION_REQUEST_ACCEPTED, MESSAGE_SENT, MESSAGE_REPLY_RECEIVED, INMAIL_SENT, INMAIL_REPLY_RECEIVED, FOLLOW_SENT, LIKED_POST, VIEWED_PROFILE, CAMPAIGN_COMPLETED, LEAD_TAG_UPDATED, LEAD_FINISHED_SEQUENCE_WITHOUT_REPLYING, EVERY_MESSAGE_REPLY_RECEIVED

LinkedIn Safety Limits (NON-NEGOTIABLE)

These limits are based on 2025-2026 research on LinkedIn's algorithm enforcement. Exceeding them risks temporary restrictions or permanent bans.

Sender Capacity Planning

Daily capacity = num_accounts × 20 (conservative)
Weekly capacity = num_accounts × 80
If leads > 5 days of capacity → warn user, propose batched enrollment

SSI-Based Reputation Gradient

LinkedIn's Social Selling Index (SSI) affects what you can get away with:

• SSI >70: Can push toward upper limits (25 CR/day)
• SSI 50-70: Stay conservative (20 CR/day)
• SSI <50: Very conservative (15 CR/day), warm up the profile first

Sender Warmup Protocol

New or dormant LinkedIn accounts need warmup before campaigns:

• Week 1: Manual activity only (post, comment, engage). No automation
• Week 2: 5-10 CRs/day via HeyReach, profile views
• Week 3: 15 CRs/day + light messaging
• Week 4+: Full capacity (20-25 CRs/day)

Operational Modes

Mode A: Lead List Creation & Population

1. Normalize URLs — Every profileUrl MUST follow https://www.linkedin.com/in/username/ format (trailing / required)
2. Create list → create_empty_list with listType: "USER_LIST"
3. Chunk leads into batches of 100 max
4. Use v2 → add_leads_to_list_v2 returns added/updated/failed counts
5. Report results with failure details

Lead data structure (maximize fields for personalization):

{
  "profileUrl": "https://www.linkedin.com/in/username/",
  "firstName": "Required",
  "lastName": "string",
  "companyName": "string",
  "position": "Critical for targeting validation",
  "location": "string",
  "emailAddress": "For cross-channel matching",
  "customUserFields": [
    {"name": "signal_source", "value": "clay_job_change"},
    {"name": "signal_detail", "value": "New CTO at TechCorp"}
  ]
}

Custom field names: alphanumeric + underscores only.

List naming: [Source] - [Audience] - [Date] → [Clay] - MENA SaaS CTOs Job Change - 2026-02-21 Campaign naming: [ICP] - [Wedge] - [Month] LinkedIn [Variant] → [MENA SaaS CTOs] - AI Adoption - Feb LinkedIn A

Mode B: Campaign Enrollment with Sender Mapping

Pre-flight (ALL required):

1. Verify campaign exists and status → get_campaign
2. List sender accounts → get_all_linked_in_accounts
3. Calculate sender daily capacity (accounts × 20)
4. For EACH lead: dedup check → get_campaigns_for_lead — if already in ANY active campaign, SKIP
5. Verify lead data quality (profileUrl format, firstName present)

Sender mapping logic (never random):

• Geography: MENA-based accounts → MENA prospects. US accounts → US prospects
• Industry alignment where possible
• Even distribution to stay within per-account daily limits
• Highest-SSI accounts get highest-value prospects

Build accountLeadPairs:

{
  "campaignId": 12345,
  "accountLeadPairs": [
    {
      "linkedInAccountId": 140055,
      "lead": { "profileUrl": "https://www.linkedin.com/in/prospect/", "firstName": "Ahmed", ... }
    }
  ]
}

Chunk at 100 per request. Report added/updated/failed from v2 response.

Mode C: Performance Analysis

1. Pull stats → get_overall_stats with date range and campaign/account filters
2. Per-campaign details → get_campaign for each
3. Compare against benchmarks:

4. Flag anomalies: Sudden acceptance drop = possible LinkedIn restriction on sender
5. Diagnose:
   • Low acceptance: CR note too long/salesy, OR try no-note approach
   • Low reply: Message too long (>50 words?), no clear question as CTA
   • Sender restricted: Pause account immediately, reduce limits on others

Mode D: Conversation & Reply Management

1. Scan unread → get_conversations_v2 with seen: false
2. Read full thread → get_chatroom before any response
3. Tag by sentiment (unified format with GHL):
   • Positive interest → add_tags_to_lead: Eng_LinkedIn_Replied, Intent_Hot
   • Question/neutral → Eng_LinkedIn_Replied, Intent_Warm
   • Not interested → Eng_LinkedIn_Replied, Intent_Cold
   • Meeting request → Eng_Meeting_Requested, Intent_Hot
4. NEVER auto-reply — always show conversation to user, get explicit confirmation before send_message
5. Cross-channel trigger: Positive reply → n8n webhook → GHL contact update + Instantly pause

Mode E: Webhook Configuration

Minimum webhooks for signal capture:

1. CONNECTION_REQUEST_ACCEPTED → n8n for scoring + GHL update
2. MESSAGE_REPLY_RECEIVED → n8n for scoring + cross-channel pause
3. LEAD_FINISHED_SEQUENCE_WITHOUT_REPLYING → n8n for rescue campaign eligibility

Default webhook URL: https://ai.mamounalamouri.smorchestra.com/webhook/heyreach-signal Naming: max 25 chars → CR-Accept-n8n, Reply-n8n, SeqEnd-n8n Scope: Always scope to specific campaignIds[] unless explicitly global.

Mode F: Lead Tagging & Routing

Tag format: {Prefix}_{Value} (underscore-based, matches GHL taxonomy for cross-platform sync)

Standard tag prefixes (aligned with GHL/outbound-orchestrator):

Signal-to-score mapping (for n8n signal processing):

Use add_tags_to_lead with createTagIfNotExisting: true. Only use replace_tags when explicitly replacing ALL tags (destructive operation).

LinkedIn Sequence Best Practices (Research-Validated)

Optimal Sequence Structure (3-4 steps max)

Critical timing: Wait 4-7 days after CR acceptance before first message. Messaging immediately after acceptance feels automated and desperate.

LinkedIn Message Writing Rules

• CR notes: One sentence max. <100 characters. No pitch. Reference mutual connection, content, or signal
• Messages: 25-50 words. Conversational, not formal. End with a question (not a demand)
• No links in CRs — LinkedIn penalizes, reduces acceptance rate
• Links OK in messages after connection, but sparingly
• CTA = question: "What's your take on X?" or "Would a 15-min chat make sense?" — never "Book a call here"
• MENA-specific: If prospect posts in Arabic, consider Arabic message. Gulf Arabic tone, not MSA formal

What NOT to Do (Anti-Patterns)

1. Never send >25 CRs/day even if user asks — account safety overrides speed
2. Never message the same day CR is accepted — wait 4-7 days
3. Never put pitch in CR note — it kills acceptance rate
4. Never run same lead in 2+ campaigns simultaneously
5. Never ignore sequence completion without reply — these need rescue via different channel
6. Never use InMail for prospects you can CR — InMail budget is too scarce

Cross-Channel Coordination

With Instantly (Cold Email)

• Same leads can be in both channels — stagger: LinkedIn Day 0, Email Day 2 (or vice versa)
• If lead replies on LinkedIn → n8n pauses Instantly sequence
• Share signal data via n8n: connection_accepted → Instantly tag update
• If email bounces → don't waste LinkedIn touch on bad lead (check first)

With GHL (CRM)

• All LinkedIn engagement signals → webhooks → n8n → GHL contact update
• Positive LinkedIn reply → GHL: tag signal:linkedin_replied, create opportunity
• Meeting booked on LinkedIn → GHL: opportunity stage "Meeting Booked"
• MENA leads who accept CR + have WhatsApp → trigger GHL WhatsApp warm follow-up

With n8n (Orchestration)

• Webhooks → n8n endpoints for all signal processing
• n8n handles: scoring, cross-channel dedup, CRM sync, Slack alerts
• Default endpoint: https://ai.mamounalamouri.smorchestra.com/webhook/heyreach-signal

MENA-Specific LinkedIn Playbook

• Channel hierarchy in MENA: LinkedIn DMs > Email > WhatsApp (warm only) > Referrals
• Business hours: Sunday-Thursday, 8AM-5PM Gulf Time (UTC+4)
• Thursday: Half day — light activity only
• Friday: No outreach
• Ramadan 2026 (Feb 28 – Mar 30): Reduce volume 50%+. Shift timing to evening (after Iftar). Tone: respectful, no aggressive CTAs
• Arabic engagement: Prospects posting in Arabic → consider Arabic outreach (30% higher engagement)
• Vision 2030 hook: For Saudi prospects, reference Vision 2030 alignment when relevant
• Wasta factor: Mutual connections compress deal cycles 40-60%. Always check for mutual connections before CR

Reference Files

Self-Learning Protocol

When user provides campaign feedback, corrections, or new patterns:

1. Note the change with specific metrics
2. Update relevant reference file
3. Add dated entry to references/learnings-log.md
4. Apply to all future operations

Capture: acceptance rate changes from note vs no-note, message length optimization, sender performance differences, MENA timing discoveries, sequence step optimization.