win-loss-brief

Win-Loss Brief

You are a deal-intelligence writer. Write a brief on why one specific closed deal was won or lost. This is a phase-2 deal-intelligence skill — it writes something forward for a manager/leader ritual (a QBR, a win-loss review), one closed deal at a time. It is not a whole-book digest like deal-risk-digest, and it does not grade an in-flight call the way meddicc or challenger do — the deal is already closed.

Before you start

• If .agents/gtm-context.md (or .claude/gtm-context.md) exists, read it first and don't ask for anything it already answers.
• Run this end to end without stopping to ask which drivers to include — apply the driver categories and confidence test below and report what the transcript/CSV actually supports, including an all-tentative brief if that's what the evidence supports.
• If the file has several closed deals and it isn't obvious which one to brief, ask which row — otherwise proceed without asking.

When to use this

• A manager preparing a QBR wants a written brief on why a specific deal was won or lost, not a live scoring of the call.
• A rep or manager reviewing a closed deal wants the loss drivers named with their actual evidence, not a generic post-mortem.
• Someone building a win-loss review deck needs one deal's story pulled from the call(s) and/or the CRM row, cited, not reconstructed from memory.

Inputs

Either a call transcript (or several from the same deal, read in chronological order) or a CRM export row for the closed deal, or both. Both together produces a fuller brief; neither is required alone.

Modes

Transcript mode (.txt, .vtt, .json, .md)

claude "run win-loss-brief on ./calls/deal-final-call.txt"

Read the transcript(s) in order. Pull drivers from what was actually said, citing the quote and its timestamp if present.

CSV mode (.csv)

claude "run win-loss-brief on ./exports/pipeline.csv"

Find the row for the deal (by name, or ask which row if the file has several closed deals and it isn't obvious which one). Match headers case-insensitively, ignoring _/-/space differences, and accept the synonyms listed in references/rubric.md. If a column the brief needs is missing entirely, say so once, up front, rather than guessing from another column.

A CSV-only brief is a structural claim, not a call-quality claim — the same disclaimer won-pipeline-check and the other *-pipeline-check skills use. It reports what the row's own fields say (outcome, reason, competitor, value, close date), not what happened on any call.

Both together

Read the transcript(s) first, then use the CSV row to fill gaps (a close date or value the call never stated) or to flag a mismatch between what the call implied and what the CRM row records (e.g. call reads as a clean win, row's reason column says "price"). Note any such mismatch explicitly.

Brief structure

1. Outcome — won or lost, close date, deal value. Cite the source (quote/timestamp, or column name + cell value).
2. What drove it — 3-5 cited factors, each tagged with one of the driver categories in references/rubric.md (competitive displacement, champion strength or loss, pricing/procurement friction, timing, technical fit). Every factor cites its quote/timestamp or its column name + cell value — a driver with no cited evidence does not ship.
3. What a similar deal should do differently — one or two forward-looking notes, each grounded in a driver actually named in section 2. No generic sales advice ("build more urgency") disconnected from what was cited.

Driver confidence

Tag each driver in section 2 as clear (the transcript or CSV states it plainly) or tentative (the closest thing said is a hedge, a guess, or an unconfirmed aside — quote it as such, don't upgrade it to a confident driver). See references/rubric.md for the full test. If every driver comes back tentative, say so explicitly — that's a finding about how thin the evidence is, not something to paper over with confident language.

Sample data

assets/sample-transcript.txt is a synthetic late-stage call from a deal that was ultimately lost. assets/sample-pipeline-won-loss.csv is a synthetic closed-deal export with both won and lost rows. Run the skill against either, or both, first.

Do not

• Don't ship a driver with no cited quote or CSV column/cell behind it.
• Don't upgrade a tentative driver to clear just to make the brief read more confidently.
• Don't grade an in-flight call against a live rubric — that's meddicc or challenger's job; this skill only briefs deals that already closed.

Related skills

• deal-risk-digest sweeps the whole open pipeline for risk; this skill briefs one already-closed deal.
• competitor-battlecard covers a competitive frame from a single in-flight call; this skill covers competitive displacement as one of several loss drivers, after the deal has closed.

What this does not do

No CRM connection, no API calls, no telemetry, no data retention beyond the current session. It reads the file(s) you point it at and nothing else.