Operating Limits
- Connection requests: warm to 15–25/day; stay human-like (randomized intervals)
- DMs: 30–60/day per seat with throttling; sequence length ≤ 3 messages
- Profile views: 100–200/day staggered
- Always include manual review steps for new campaigns
Data Model
lead(id, first_name, last_name, headline, company, title, email, li_url, tags[], owner)
activity(id, lead_id, type, direction, channel, body, ts, outcome)
sequence(id, name, steps[])
step(id, seq_id, day_offset, channel, template_id)
Enrichment & Scoring
- Company size, tech signals (e.g., HubSpot, Salesforce), recent hiring
- Score: role fit, industry match, ICP keywords in headline/about
- Route: A (SDR), B (nurture content), C (exclude)
Templates (short, value-first)
Connection: "Hey {{first}}, loved your post on {{topic}} — would you connect?"
DM1 (after accept): "Thanks {{first}} — we automate {{function}} for {{peer_industry}}. 1 line overview if helpful?"
DM2: "Can share a 30s Loom on how we booked {{metric}} for {{peer}} — want it?"
DM3: "Happy to park if not relevant. Quick 15-min slot?"
Automation Orchestration
n8n workflow (high level)
cron -> fetch new ICP leads -> enrich -> score -> assign ->
create LI actions queue -> human approve -> execute (connection/DM) ->
track outcomes -> update CRM -> if positive -> create booking link -> notify Slack
Safety & QA
- Rate limit per seat and across org; backoff on failures
- Message review queue with side-by-side preview
- Exclude lists: competitors, customers, do-not-contact
- A/B 2 variants max; stop on negative signal
Content Engine
- 3 pillars: Proof (case metrics), Process (how-to threads), POV (takes)
- Cadence: 2–3 posts/week; 1 short video; 1 carousel/month
- Repurpose: post → email → LI DM follow-up hooks
