# Win Loss

> Systematic win-loss analysis across all deals. Use when someone says "win-loss analysis", "why did we lose [deal]", "why did we win [deal]", "deal analysis", "analyze our pipeline", "what patterns do you see in our deals", or asks about deal outcomes, pipeline health, or sales patterns.

- Skill: `jp-solumhealth/win-loss` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add jp-solumhealth/win-loss`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jp-solumhealth/win-loss/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: jp-solumhealth (https://skillmd.com/u/jp-solumhealth)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jp-solumhealth/win-loss

---


# Win-Loss Analysis

Perform rigorous, data-driven win-loss analysis by pulling from HubSpot, Fireflies, and Apollo, then cross-referencing everything before drawing conclusions.

## Execution Steps

### Step 1: Determine Scope

Ask the user (or infer from their request):
- **Single deal analysis** — "why did we lose Action Seating" or "analyze the [Company] deal"
- **Portfolio analysis** — "win-loss analysis", "analyze our pipeline", "what patterns do you see"

### Step 2: Pull All Data

Regardless of scope, gather everything first. Do NOT skip any source.

#### 2a. HubSpot — Deal Data

Use the HubSpot MCP tools to pull:

1. **All deals** — search for deals across all stages (won, lost, open)
   - Properties to fetch: `dealname`, `amount`, `dealstage`, `closedate`, `createdate`, `pipeline`, `hs_lastmodifieddate`, `hs_deal_stage_probability`, `closed_lost_reason`, `closed_won_reason`, `notes_last_updated`, `num_notes`, `num_associated_contacts`, `hs_analytics_source`
2. **Associated contacts** for each deal — get contact names, emails, titles, company
3. **Associated companies** for each deal — get company name, domain, industry, size
4. **Deal notes and activity** — get notes, emails, calls, meetings logged against each deal
5. **Stage history** — track how long each deal spent in each stage

#### 2b. Fireflies — Conversation Intelligence

Use the Fireflies MCP tools:

1. **Search transcripts** by contact email addresses from the deals
2. **Get full transcripts** for every meeting associated with deal contacts
3. **Get summaries** for quick scanning
4. Extract from each transcript:
   - Direct quotes about pain points
   - Objections raised (pricing, timing, competition, internal resistance)
   - Competitor names mentioned
   - Pricing discussions (what was quoted, reactions)
   - Buying signals ("when can we start", "who else needs to approve")
   - Decision-maker identification
   - Next steps discussed vs. what actually happened

#### 2c. Apollo — Company Enrichment

Use the Apollo MCP tools:

1. **Enrich each company** associated with deals:
   - Company size (employees)
   - Industry and sub-industry
   - Annual revenue
   - Funding stage and total raised
   - Technologies used
   - Location
2. **Enrich key contacts** — titles, seniority, department
3. Use this to build ICP pattern analysis

### Step 3: Internal Fact-Check Layer

Before any analysis, run these integrity checks. Flag every issue found.

#### Data Quality Checks

| Check | How | Flag If |
|-------|-----|---------|
| Stale close dates | Compare `closedate` to today | Close date is in the past but deal is still open |
| Zombie deals | Check activity on closed-lost deals | Closed-lost deal has emails/calls in last 30 days |
| Ghost deals | Check `num_notes` and last activity | Deal has zero notes or no activity in 60+ days |
| Amount mismatches | Compare HubSpot `amount` to pricing discussed in Fireflies | Amount differs by more than 20% from what was quoted verbally |
| Missing contacts | Check `num_associated_contacts` | Deal has zero associated contacts |
| Stage bottlenecks | Calculate days in current stage | Deal has been in same stage for 2x the average |
| Pipeline fiction | Check deals marked "closing this month" | No activity in last 14 days on a deal closing within 30 days |

#### Cross-Reference Checks

- If HubSpot says "closed lost — went with competitor", search Fireflies for which competitor and why
- If HubSpot says "closed lost — pricing", check Fireflies for what price was quoted and what their reaction was
- If a deal has no close reason in HubSpot, attempt to find the reason in Fireflies transcripts
- Compare deal timeline in HubSpot stages vs. actual meeting cadence in Fireflies

### Step 4A: Single Deal Analysis

When analyzing a specific deal, produce this report:

```
## Win-Loss: [Deal Name]

### Deal Snapshot
| Field | Value |
|-------|-------|
| Company | [name] |
| Amount | $[amount] |
| Stage | [stage] |
| Created | [date] |
| Closed | [date] |
| Days in Pipeline | [n] |
| Outcome | Won / Lost / Open |
| Loss Reason (HubSpot) | [reason or "not recorded"] |

### Company Profile (Apollo)
- Industry: [industry]
- Size: [employees]
- Revenue: [revenue]
- Funding: [stage, amount]
- ICP Fit: Strong / Moderate / Weak — [why]

### Timeline
Map every touchpoint chronologically:
- [Date] — First contact (source: [how they came in])
- [Date] — Discovery call (Fireflies transcript available: yes/no)
- [Date] — Stage moved to [stage]
- [Date] — Demo / proposal sent
- [Date] — Last activity
- [Date] — Outcome

### What They Said (from Fireflies transcripts)
Pull DIRECT QUOTES. Do not paraphrase. Organize by theme:

**Pain Points:**
> "[exact quote]" — [Contact Name], [Date]

**Objections:**
> "[exact quote]" — [Contact Name], [Date]

**Competitor Mentions:**
> "[exact quote]" — [Contact Name], [Date]

**Buying Signals (if any):**
> "[exact quote]" — [Contact Name], [Date]

**Pricing Reactions:**
> "[exact quote]" — [Contact Name], [Date]

### Pricing Analysis
- Amount in HubSpot: $[amount]
- What was discussed in calls: [details from Fireflies]
- Match: Yes / No — [explain discrepancy]
- Price sensitivity level: Low / Medium / High

### Root Cause Analysis
Go beyond the surface. "Went dark" is not a root cause. Dig into:
1. **Primary cause** — the single biggest factor
2. **Contributing factors** — secondary issues that compounded
3. **Process failures** — what we could have done differently
4. **Timing factors** — was the timing right for them?

### Data Quality Issues
Flag anything found in the fact-check layer for this deal.

### Lessons
- **Repeat:** [what worked or would work again]
- **Change:** [what to do differently next time]
- **Watch for:** [early warning signs we missed]
```

### Step 4B: Portfolio Analysis

When analyzing all deals, produce this report:

```
## Win-Loss Portfolio Analysis
*Analysis Date: [today's date]*
*Deals Analyzed: [n won] + [n lost] + [n open] = [total]*

### Pipeline Metrics
| Metric | Value |
|--------|-------|
| Total Deals | [n] |
| Open Deals | [n] |
| Won | [n] |
| Lost | [n] |
| Win Rate | [%] |
| Avg Deal Size (Won) | $[amount] |
| Avg Deal Size (Lost) | $[amount] |
| Avg Sales Cycle (Won) | [days] |
| Avg Sales Cycle (Lost) | [days] |
| Pipeline Value (Open) | $[amount] |
| Weighted Pipeline | $[amount] |

### Stage Funnel
Show deals by stage with average days in stage and conversion rate to next stage:
| Stage | Deals | Avg Days | Conversion to Next |
|-------|-------|----------|-------------------|
| [stage] | [n] | [days] | [%] |

### Patterns in Wins
Analyze won deals for common traits:

**ICP Profile of Winning Deals:**
- Company size range: [range]
- Industries: [list]
- Common pain points: [list with frequency]
- Decision-maker titles: [list]
- Lead source: [breakdown]

**What Worked:**
- Messaging that resonated (with transcript evidence)
- Objections we overcame and how
- Average touchpoints to close

### Patterns in Losses
Analyze lost deals for common traits:

**Why We Lose:**
| Reason | Count | % of Losses |
|--------|-------|-------------|
| [reason] | [n] | [%] |

**Where Deals Stall:**
- Stage with highest drop-off: [stage]
- Average days before going dark: [days]

**Competitor Displacement:**
| Competitor | Times Mentioned | Deals Lost To | Key Differentiator They Cited |
|-----------|----------------|---------------|-------------------------------|
| [name] | [n] | [n] | [what prospects said] |

### Zombie Deals (Immediate Action Required)
Deals that need attention right now:

| Deal | Issue | Amount | Last Activity | Recommended Action |
|------|-------|--------|---------------|--------------------|
| [name] | Stale close date | $[x] | [date] | [action] |
| [name] | Closed lost but still active | $[x] | [date] | [action] |
| [name] | No activity 60+ days | $[x] | [date] | [action] |

### Data Quality Report
| Issue | Count | Deals Affected |
|-------|-------|----------------|
| Missing close reason | [n] | [list] |
| No associated contacts | [n] | [list] |
| No notes or activity | [n] | [list] |
| Amount = $0 or blank | [n] | [list] |

### Recommendations

**ICP Refinement:**
Based on win patterns, the ideal customer profile should be refined to:
- [specific, data-backed recommendations]

**Process Improvements:**
- [specific changes to sales process with evidence]

**Pricing Adjustments:**
- [data-backed pricing observations]

**Pipeline Hygiene:**
- [specific cleanup actions needed]

**Follow-Up List:**
Deals worth re-engaging and why:
| Deal | Why Re-engage | Suggested Approach |
|------|---------------|--------------------|
| [name] | [reason from transcript/activity] | [approach] |
```

## Rules

1. **Never guess.** If data is missing, say so. Do not fill in blanks with assumptions.
2. **Use direct quotes.** When citing what prospects said, pull exact quotes from Fireflies transcripts. Paraphrasing loses the signal.
3. **Cross-reference everything.** HubSpot data alone is unreliable. Always verify against Fireflies and Apollo.
4. **Flag data quality issues prominently.** Bad data leads to bad decisions. Surface every inconsistency.
5. **"Went dark" is not a root cause.** Dig deeper. Why did they go dark? What was the last thing discussed? Was there a trigger?
6. **Separate facts from interpretation.** Present the data first, then your analysis. Make it clear which is which.
7. **Be blunt.** This is an internal tool. Do not soften bad news. If the pipeline is fiction, say so.
8. **Include the "so what."** Every insight should connect to a specific action.
9. **Time-box the data pull.** If a data source is unavailable or returns errors, note it and continue with what you have. Do not block the entire analysis on one failed API call.
10. **Save the output.** Write the final report to `~/Documents/Claude/Agents/reports/win-loss-[date].md` so it is preserved.

## Related Skills

- founder-sales
- sales-qualification
- enterprise-sales
- pricing-strategy
- pricing-coach