win-loss-analysis — Day 4 research skill
The Example 1 Day 4 skill. Reads your won + lost sales-call transcripts and writes a pattern-level analysis to marketing/win-loss/win-loss.md. This is the evidence base under positioning, messaging, and ICP — every claim in those docs should trace back to something a real buyer said here.
When to use
- Day 4 of Example 1: before
/icp-research + /positioning, so the strategy reads from real buyer language
- You have a fresh batch of 5+ won/lost call transcripts
- A quarter closes and you want to refresh why deals moved
- Churn spikes and you need the pattern, not the anecdote
When NOT to use
- For a single account's interview prep (use
/customer-interviews — not in V2 quickstart)
- For behavioural simulation of a buyer (use
/icp-behavioural — not in V2 quickstart)
- When you have fewer than ~5 transcripts — patterns need volume; below that you get anecdotes, not signal
Two rules apply before any analysis
How it works
- Inputs: sales-call transcripts (Gong, Fireflies, Otter, Granola, Zoom/Avoma VTT, or pasted text), each tagged with the deal outcome (won / lost / churned). Optional:
marketing/icp/ICP.md to frame patterns by segment.
- Normalize each transcript to speaker-attributed turns with timestamps where present.
- Pick a mode:
- Single — deep analysis of one transcript
- Batch (default) — aggregate 5–20 transcripts into patterns with frequency counts
- Comparison — won vs lost (or retained vs churned) side by side
- Extract patterns across six dimensions: product, messaging, GTM / sales process, pricing, competition, customer context.
- Score confidence by frequency: a pattern needs 2+ occurrences across different deals; High = 3+ deals, Medium = 2, Low = single mention. Aim for ≥5 wins and ≥5 losses before trusting a pattern.
- Writes to
marketing/win-loss/win-loss.md (overwrites prior canonical; git history preserves prior versions).
Invoke
/win-loss-analysis
Then paste or point to the transcripts and tag each outcome. Or:
/win-loss-analysis — here are 8 won + 6 lost transcripts: [paste / paths]
Example output
See marketing/win-loss/win-loss.md for the PulseAnalytics example seed. Notice: patterns are grouped by dimension; each carries a frequency + a verbatim quote with speaker; the closing section routes findings to positioning / messaging / ICP.
Dependencies
- Reads from: sales-call transcripts (required);
marketing/icp/ICP.md (optional, for segment framing)
- Reads via Granola MCP (optional): meeting transcripts, if wired
- Writes to:
marketing/win-loss/win-loss.md (canonical; positioning, messaging, and ICP read from here)
Customization
Split the analysis by segment when your ICP has more than one (enterprise vs mid-market lose for different reasons). Add a competitor column to the competition dimension once you're losing to a named rival repeatedly — that feeds /competitor-research.
Where this fits in the Example 1 chain
Day 1-3: /competitor-research × N → per-competitor files
Day 3: /competitor-aggregate → competitor canonical
Day 4: /win-loss-analysis (THIS SKILL) → win-loss canonical
Day 5: /icp-research reads win-loss + competitors → canonical ICP
Week 2: /positioning + /product-messaging read win-loss quotes for real buyer language
Refresh cadence
Monthly if deal volume is high; quarterly otherwise. Refresh sooner on a churn spike or a new competitor showing up repeatedly in lost deals.
1---2name: win-loss-analysis3description: Analyze sales-call transcripts to extract why deals are won and lost across six dimensions — product, messaging, GTM/sales, pricing, competition, and customer context. Produces aggregated patterns with verbatim buyer quotes, frequencies, and recommendations. Writes to marketing/win-loss/win-loss.md as the evidence base under positioning, messaging, and ICP. Triggers - "win loss analysis", "why are we losing deals", "why do we win", "analyze sales calls", "churn analysis", "deal review"4---5
6# win-loss-analysis — Day 4 research skill
7
8The Example 1 Day 4 skill. Reads your won + lost sales-call transcripts and writes a pattern-level analysis to `marketing/win-loss/win-loss.md`. This is the evidence base under positioning, messaging, and ICP — every claim in those docs should trace back to something a real buyer said here.
9
10---
11
12## When to use
13
14- Day 4 of Example 1: before `/icp-research` + `/positioning`, so the strategy reads from real buyer language
15- You have a fresh batch of 5+ won/lost call transcripts
16- A quarter closes and you want to refresh why deals moved
17- Churn spikes and you need the pattern, not the anecdote
18
19## When NOT to use
20
21- For a single account's interview prep (use `/customer-interviews` — not in V2 quickstart)
22- For behavioural simulation of a buyer (use `/icp-behavioural` — not in V2 quickstart)
23- When you have fewer than ~5 transcripts — patterns need volume; below that you get anecdotes, not signal
24
25## Two rules apply before any analysis
26
27- **Redact PII first** — [`.claude/rules/pii-redaction.md`](../../rules/pii-redaction.md). Mask end-customer names, emails, and account numbers before processing; keep roles, company, and deal context.
28- **Bind every claim to evidence** — [`.claude/rules/evidence-bound-outputs.md`](../../rules/evidence-bound-outputs.md). Every pattern cites a verbatim quote + speaker, or it lowers its confidence. No invented quotes.
29
30## How it works
31
321. Inputs: sales-call transcripts (Gong, Fireflies, Otter, Granola, Zoom/Avoma VTT, or pasted text), each tagged with the deal outcome (won / lost / churned). Optional: `marketing/icp/ICP.md` to frame patterns by segment.
332. Normalize each transcript to speaker-attributed turns with timestamps where present.
343. Pick a mode:
35 - **Single** — deep analysis of one transcript
36 - **Batch** (default) — aggregate 5–20 transcripts into patterns with frequency counts
37 - **Comparison** — won vs lost (or retained vs churned) side by side
384. Extract patterns across six dimensions: **product**, **messaging**, **GTM / sales process**, **pricing**, **competition**, **customer context**.
395. Score confidence by frequency: a pattern needs 2+ occurrences across different deals; High = 3+ deals, Medium = 2, Low = single mention. Aim for ≥5 wins and ≥5 losses before trusting a pattern.
406. Writes to `marketing/win-loss/win-loss.md` (overwrites prior canonical; git history preserves prior versions).
41
42## Invoke
43
44```
45/win-loss-analysis
46```
47
48Then paste or point to the transcripts and tag each outcome. Or:
49
50```
51/win-loss-analysis — here are 8 won + 6 lost transcripts: [paste / paths]
52```
53
54## Example output
55
56See [`marketing/win-loss/win-loss.md`](../../../pulse-analytics-example/win-loss/win-loss.md) for the PulseAnalytics example seed. Notice: patterns are grouped by dimension; each carries a frequency + a verbatim quote with speaker; the closing section routes findings to positioning / messaging / ICP.
57
58## Dependencies
59
60- **Reads from:** sales-call transcripts (required); `marketing/icp/ICP.md` (optional, for segment framing)
61- **Reads via Granola MCP (optional):** meeting transcripts, if wired
62- **Writes to:** `marketing/win-loss/win-loss.md` (canonical; positioning, messaging, and ICP read from here)
63
64## Customization
65
66Split the analysis by segment when your ICP has more than one (enterprise vs mid-market lose for different reasons). Add a competitor column to the competition dimension once you're losing to a named rival repeatedly — that feeds `/competitor-research`.
67
68## Where this fits in the Example 1 chain
69
70```
71Day 1-3: /competitor-research × N → per-competitor files
72Day 3: /competitor-aggregate → competitor canonical
73Day 4: /win-loss-analysis (THIS SKILL) → win-loss canonical
74Day 5: /icp-research reads win-loss + competitors → canonical ICP
75Week 2: /positioning + /product-messaging read win-loss quotes for real buyer language
76```
77
78## Refresh cadence
79
80Monthly if deal volume is high; quarterly otherwise. Refresh sooner on a churn spike or a new competitor showing up repeatedly in lost deals.
majiayu000