scientiacapital Specific deal: "should I kill the Acme deal?" → analyzes that deal specifically → recommendation
Recovery campaign: "run recovery on stalled deals" → generates re-engagement emails for recoverable deals
Trigger phrases:
- "dead deals" / "stalled deals" / "zombie deals"
- "clean pipeline" / "pipeline cleanup" / "pipeline hygiene"
- "disqualify deals" / "which deals should I kill"
- "deal recovery" / "win back" / "lost deal review"
Architecture
TRIGGER DIAGNOSIS TRIAGE ACTION
──────────────────────────────────────────────────────────────────────────────────────
"clean pipeline" → Pull all open deals → Score deal health → HEALTHY: monitor
→ Activity history → Classify H/R/D → RECOVERABLE: re-engage
→ Contact engagement → Identify root cause → DEAD: disqualify
→ MEDDIC completeness → Predict recovery % → Gmail drafts
→ Close date integrity → → HubSpot updates
Stage 1: Data Collection
1a. Pull All Open Deals
FilterGroup 1 — Tim is COLLABORATOR:
hs_all_collaborator_owner_idsCONTAINS_TOKEN87486452hs_is_closedEQfalse
FilterGroup 2 — Tim is OWNER:
hubspot_owner_idEQ87486452hs_is_closedEQfalse
Properties: dealname, dealstage, amount, hubspot_owner_id, hs_lastmodifieddate, closedate, createdate, num_associated_contacts, hs_deal_stage_probability
1b. Activity History per Deal
| Tool | Purpose |
|---|---|
ask_agent |
"Show all activity for [company] deals in last 90 days" |
hubspot_get_deal |
Deal notes, stage change history |
1c. Contact Engagement + Clari Check
| Tool | Purpose |
|---|---|
hubspot_search_contacts |
Email opens, clicks, last engagement date |
clari_search_calls |
Calls in last 60 days |
clari_get_call_summary |
Last call sentiment |
Stage 2: Deal Health Diagnosis
Score each deal on 5 death signals (0-20 pts each, 100 total).
Signal 1: Activity Recency (0-20)
| Last activity | Score |
|---|---|
| < 7 days | 20 |
| 7-14 days | 15 |
| 14-30 days | 8 |
| 30-60 days | 3 |
| 60+ days | 0 — DEAD SIGNAL |
Signal 2: Champion Engagement (0-20)
| Status | Score |
|---|---|
| Active champion responding | 20 |
| Responded in last 14 days | 15 |
| Last response 14-30 days | 8 |
| Ghosting 30+ days | 2 |
| No champion ever identified | 0 |
Signal 3: Close Date Integrity (0-20)
| Status | Score |
|---|---|
| Future, never pushed | 20 |
| Pushed once, still future | 12 |
| Pushed 2+ times | 5 |
| Past (overdue) | 0 — OVERDUE |
Signal 4: MEDDIC Completeness (0-20)
| Status | Score |
|---|---|
| 5-6 dimensions identified | 20 |
| 3-4 dimensions | 12 |
| 1-2 dimensions | 5 |
| 0 dimensions | 0 — UNQUALIFIED |
Signal 5: Stakeholder Breadth (0-20)
| Contacts | Score |
|---|---|
| 3+ contacts including ATL (economic buyer) | 20 |
| 2+ contacts with ≥1 ATL | 15 |
| 2+ contacts but all BTL | 8 |
| Single-threaded (1 contact only) | 3 |
| 0 active contacts | 0 — ABANDONED |
Classification
| Total | Class | Action |
|---|---|---|
| 60-100 | HEALTHY | Monitor via deal-momentum-analyzer |
| 30-59 | RECOVERABLE | Run recovery campaign |
| 0-29 | DEAD | Recover in 14 days or disqualify |
Root Cause Mapping
| Root Cause | Recovery Strategy |
|---|---|
| GHOST | Multi-channel blitz: email + call + LinkedIn |
| NO_CHAMPION | Find new contact via Apollo |
| SINGLE_THREADED | Research org chart, find peer or manager |
| BUDGET_STALL | Re-engage at next budget cycle |
| COMPETITOR_WIN | Log intel. Re-engage in 12 months. |
| TIMING | Set calendar reminder for stated timeline |
| UNQUALIFIED | Disqualify immediately |
Stage 3: Recovery Campaign (3-touch, 10 days)
Email 1 — Check-In (Day 0): Tactical empathy. "It's been a while since we connected about [topic]..."
Email 2 — Value Bomb (Day 4): Challenger reframe with new case study or vertical insight.
Email 3 — Break-Up (Day 10):
Subject: Should I close your file?
Hi [First Name], I haven't heard back — one of three things:
1. Timing isn't right (totally understand)
2. You've gone a different direction (helpful to know)
3. You're buried and this fell off the radar
Would it be a terrible idea to reconnect next quarter instead?
Create Gmail drafts via gmail_create_draft for all 3 touches.
Stage 4: Disqualification Protocol
Before closing any deal as Lost:
- ☐ Final recovery email sent (Email 3: Break-Up)
- ☐ At least one call attempt made
- ☐ 14-day waiting period after last recovery touch
- ☐ Root cause identified and documented
- ☐ Lost reason selected in HubSpot
HubSpot actions on disqualification:
- Move deal to "Closed Lost" with reason + notes
- Create follow-up reminder at re-engagement date (if applicable)
Stage 5: Output Format
╔══════════════════════════════════════════════════════════════╗
║ PIPELINE CLEANUP REPORT — [Date] ║
║ Open Deals: [X] | Pipeline: $[total] ║
╠══════════════════════════════════════════════════════════════╣
🟢 HEALTHY: [X] deals ($[amount]) — no action
🟡 RECOVERABLE: [X] deals ($[amount]) — recovery campaign
🔴 DEAD: [X] deals ($[amount]) — disqualify within 14 days
💀 PIPELINE DEAD WEIGHT: [X]% of deals
🟡 RECOVERABLE — TAKE ACTION THIS WEEK:
1. [Deal] — $[amt] | Root cause: [X] | Recovery plan: [X]
→ Gmail draft created: Yes/No
🔴 DEAD — DISQUALIFY:
2. [Deal] — $[amt] | Evidence: [X] | Learning: [X]
PIPELINE IMPACT:
Before: $[current] | After cleanup: $[cleaned]
╚══════════════════════════════════════════════════════════════╝
Stage 6: Lost Deal Review (Monthly)
Pull all deals closed as Lost in last 30 days and analyze:
- Win rate by vertical
- Loss by root cause
- Loss by pipeline stage
- MEDDIC correlation with win rate
- Recovery success rate
| Frequency | Action |
|---|---|
| Weekly (Friday) | Full pipeline cleanup scan |
| Daily | Check for deals crossing 60-day stall threshold (via morning-brief) |
| Monthly | Lost deal review + pattern analysis |
| On-demand | "should I kill [deal]" |
Sibling Skills Referenced
deal-momentum-analyzer— Shares deal health scoring; dead-deal-recovery goes deeper on RED dealsmorning-brief— Daily feed of deals approaching death thresholdsmeddic-call-prep-auto— If recovery call needed, generates MEDDIC brief
Emit Outcome Sidecar
As the final step, write to ~/.claude/skill-analytics/last-outcome-dead-deal-recovery.json:
{"ts":"[UTC ISO8601]","skill":"dead-deal-recovery","version":"1.0.0","variant":"default",
"status":"[success|partial|error]","runtime_ms":[estimated ms from start],
"metrics":{"deals_scanned":[n],"healthy":[n],"recoverable":[n],"dead":[n],"recovery_drafts_created":[n]},
"error":null,"session_id":"[YYYY-MM-DD]"}
Use status "partial" if some stages failed but results were produced. Use "error" only if no output was generated.