win-loss-analyzer

Skill: Win-Loss Analyzer

Stop guessing why you win and lose. Build an evidence-based answer.

┌─────────────────────────────────────────────────────────────────┐
│  STANDALONE (always works)                                      │
│  ✓ Single-deal post-mortem: root cause analysis in 5 minutes   │
│  ✓ Pattern detection across multiple outcomes                   │
│  ✓ Competitive analysis: where you win/lose vs. specific rivals │
│  ✓ Playbook generation: replicate wins, prevent losses          │
│  ✓ Bilingual (EN/FR): adapt language per sales-profile.json    │
├─────────────────────────────────────────────────────────────────┤
│  SUPERCHARGED (connect ~~CRM)                                   │
│  + Pull closed deals automatically (no manual input)            │
│  + Statistical patterns across 50+ deals                        │
│  + Segment by ICP, deal size, geography, rep                    │
│  + Track playbook adoption over time                            │
└─────────────────────────────────────────────────────────────────┘

🧠 Core Philosophy

1. Perception ≠ Reality: Reps say "price" was the reason they lost. Buyers say "trust". Run this process to find the truth.
2. Patterns beat anecdotes: One loss tells you almost nothing. Ten losses tell you everything.
3. Wins need autopsies too: Knowing why you win is as important as knowing why you lose. You can't replicate luck.
4. Competitor intel is a by-product: Every loss to a named competitor is a competitive intelligence event.

---

🛠 Agent Instructions

Phase 1: Single-Deal Post-Mortem

Trigger: "Debrief on the Acme deal."

Steps:

1. Load Deal Context: Read data/1-Projets/active-deals/[Client].md or ask user for:

   • Outcome (Won / Lost / No Decision)
   • Deal size and stage at close
   • Why we lost (internal narrative)
   • Known competitors in the deal
   • Key contacts and their roles
   • Timeline from first contact to close

2. Apply Root Cause Framework: See references/root-cause-framework.md

   • Ask the 5 Whys on the stated reason
   • Map to one of 6 loss categories (see below)
   • Identify the Last Recoverable Moment — the latest point where the outcome could have changed

3. Loss Category Diagnosis:

   Category	Symptoms	Telltale Signal
   Value Gap	"Not sure we need this"	Discovery was shallow; pain wasn't quantified
   Access Gap	"Champion couldn't get budget"	Never reached the Economic Buyer
   Trust Gap	"We went with the safer option"	No references, no credibility markers
   Price Gap	"Too expensive"	Value was not tied to specific ROI numbers
   Timing Gap	"Not the right time"	No compelling event established
   Process Gap	"Competitor had a slicker eval"	Demo/proposal was generic, not tailored

4. Output: A "Deal Autopsy" document. See output format below.

---

Phase 2: Pattern Analysis (Multi-Deal)

Trigger: "Show me our loss patterns this quarter."

Steps:

1. Collect Data: Load deal files from data/1-Projets/closed-lost/ and data/1-Projets/closed-won/.

   • If no files, ask user to paste a CSV or describe 5+ recent outcomes.

2. Run Pattern Detection (see references/pattern-detection.md):

   # Pattern Analysis Logic
   deals = load_deals(status=["won", "lost"])
   
   # Segment analysis
   loss_by_category = group_by(deals[status=="lost"], "loss_category")
   win_by_source = group_by(deals[status=="won"], "lead_source")
   competitor_win_rate = {}
   
   for deal in deals:
       if deal.competitor:
           if deal.competitor not in competitor_win_rate:
               competitor_win_rate[deal.competitor] = {"won": 0, "lost": 0}
           competitor_win_rate[deal.competitor][deal.status] += 1
   
   # ICP alignment
   icp_aligned_wins = filter(deals, icp_score > 70, status="won")
   icp_misaligned_losses = filter(deals, icp_score < 50, status="lost")
   
   # Deal size analysis
   avg_won_size = mean(deals[status=="won"].amount)
   avg_lost_size = mean(deals[status=="lost"].amount)
   

3. Insight Generation: Surface the top 3 actionable insights. Do not produce a list of observations — produce a list of recommended changes.

---

Phase 3: Playbook Generation

Trigger: "Build me a playbook from our wins."

Steps:

1. Extract Win DNA:

   • Common ICP characteristics in wins
   • Lead sources with highest win rates
   • Typical deal velocity for wins
   • Objections raised and successfully handled
   • What made the champion effective

2. Build the Anti-Loss Checklist: A checklist of the top 5 risks to validate on every deal. Saved to data/2-Domaines/win-loss-playbook.md.

3. Update ICP: If patterns reveal a mismatch between target ICP and actual wins, flag for icp-creator review.

---

📄 Output Format

Single Deal Autopsy

# Deal Autopsy: [Company Name]
**Outcome**: Lost / Won
**Date Closed**: [Date]
**Deal Size**: $[X]
**Primary Competitor**: [If lost]

---

## Root Cause (Primary)

**Category**: [Value Gap / Access Gap / Trust Gap / Price Gap / Timing Gap / Process Gap]
**Root Cause Statement**: [One crisp sentence. E.g. "We never reached the CFO, so the champion couldn't defend our price internally."]

## The 5 Whys

1. Why did we lose? → [Stated reason]
2. Why did [stated reason] matter? → [Layer deeper]
3. Why did [layer 2] happen? → ...
4. Why did [layer 3] happen? → ...
5. Root cause: [What we could have controlled]

---

## Last Recoverable Moment

**When**: [Stage / Date]
**What we should have done**: [Specific action]
**Signal we missed**: [The warning sign that was visible in hindsight]

---

## Competitive Intelligence

| Dimension | Us | [Competitor] | Assessment |
|-----------|-----|--------------|------------|
| [Feature/Capability] | [Ours] | [Theirs] | [Win/Loss/Tie] |

---

## Lessons (Save to Playbook)

1. [Lesson 1 — Actionable change to make on future deals]
2. [Lesson 2]
3. [Lesson 3]

---

📂 System Integration

• Output: Save autopsy to data/4-Archives/win-loss/[Client]-[date].md
• Playbook: Append lessons to data/2-Domaines/win-loss-playbook.md
• ICP update: Flag icp-creator if patterns suggest ICP drift
• Voice DNA: Load voice-dna.json for communication style in playbook

---

📚 References

• references/root-cause-framework.md: The 5 Whys + 6 loss categories in depth.
• references/pattern-detection.md: How to read signal from 5+ deals.
• references/playbook-template.md: The standard format for the Win-Loss Playbook.

STEP 0 — MANDATORY DATA LOAD

deal_file       = read(f"data/1-Projets/active-deals/{company_slug}.md")
history_losses  = glob("data/4-Archives/win-loss/*.md")  # prior debriefs
competitive_file = read(f"data/3-Ressources/competitive/{competitor_slug}.md")  # if competitor named

Report before analysis:

Win-Loss context — [Company] [Outcome]
Deal file: [FOUND — [N] interactions, [N] call notes] / [NOT FOUND — working from your description]
Prior losses to same competitor: [N found] / [None on record]

STEP 6 — COMPETITIVE LOOP (mandatory on losses to named competitor)

If deal was lost to a specific competitor:

UPDATE f"data/3-Ressources/competitive/{competitor_slug}.md":
  - Win/loss ratio: updated
  - Loss pattern: [what category this loss falls into]
  - Telltale signal: [what to watch for earlier next time]

"Updated [Competitor] battlecard: now [N] losses, [N] wins. Pattern: [1-line summary]"

If 3+ losses to same competitor: "⚠️ 3 consecutive losses to [Competitor]. Recommend full battlecard review via competitive-intelligence skill."

STEP 7 — SAVE TO PLAYBOOK

WRITE f"data/4-Archives/win-loss/{company_slug}-{outcome}-{date}.md"

Append winning/losing play to playbook index:

APPEND "data/2-Domaines/playbook.md":
  | [Date] | [Company] | [Win/Loss] | [Category] | [Key play] | [Last recoverable moment] |

"Debrief saved. Playbook updated — [N] total entries now."