Win/Loss Analysis

The framework — strategy first

Win/Loss Analysis — the empirical loop that closes positioning.

Why win/loss is the loop most teams underinvest in.

THE FEEDBACK LOOP THAT TELLS YOU IF EVERYTHING ELSE IS WORKING

Every other area of the playbook in the playbook is a hypothesis. The Right-to-Win statement is a hypothesis. The persona profiles are a hypothesis. The pricing model is a hypothesis. Without win/loss data, those hypotheses never get tested — they get repeated. With win/loss data on a quarterly cadence, you find out whether the buyer experience matches the playbook’s assumptions, and you tune accordingly.

Most business-to-business (B2B) SaaS companies do win/loss badly: the Account Executive (AE) writes a sentence in the Customer Relationship Management system (CRM) about why the deal closed or didn’t, and that’s the program. The companies that do it well separate the work, structure the interviews, and turn the data into thematic synthesis that the rest of the org acts on.

Three rules that determine whether a win/loss program is doing its job. The interviews are run by someone other than the deal owner. The AE who closed the deal will fish for validation; the Customer Success Manager (CSM) who’s renewing will fish for renewal. The interviewer needs to have no incentive in the outcome of the conversation. The threshold is structural, not aspirational. Every deal over a defined Annual Contract Value (ACV) — or every loss against a named competitor — gets an interview, on cadence, no exceptions. Selective sampling reveals bias. The synthesis is published, not buried. The quarterly readout goes to product, marketing, sales, and leadership — with named themes, frequency counts, and specific recommendations. Win/loss work that lives in a folder is the same as not doing it.

Win interview framework — the eight questions.

THE WIN INTERVIEW — 30 MINUTES, EIGHT QUESTIONS

The win interview happens within 30 days of close. The questions are designed to surface the decisive factors, not the gentle compliments. Run the same eight questions on every win to make the data comparable across deals.

At what point did you feel confident we were the right choice? The moment in the journey that locked it in. Often surprising; rarely the demo.
What was the #1 thing that differentiated us from the other vendors? The single most-important difference, in their words. Usually NOT a feature.
Was there anything we could have done better during the sales process? The friction we didn’t notice. Listen for the soft “well, one thing...”
Who was the internal champion for this decision, and what made them advocate for us? The persona-level intelligence. What about us made them stick their neck out.
What objections did you have to overcome internally to make this purchase? The objections we never saw — the ones the Champion fought without us in the room.
How will you measure the success of this purchase in 12 months? Their actual success criteria. The metrics that determine whether they renew.
Who else was on the buying committee and what was their role? The committee map. Cross-reference with our persona work.
What almost made you choose someone else? The pivot moment. The one thing that almost lost the deal.

Loss interview framework — the eight questions.

THE LOSS INTERVIEW — HARDER TO GET, MORE VALUABLE

Losses are harder to interview — the prospect doesn’t owe you a conversation. The discipline is asking immediately (within 14 days of loss), offering value (the interview is the unobtrusive, no-pitch kind), and accepting that you’ll only get 25–40% of losses to talk. That sample is still gold.

When did you make the decision to go with [WINNER]? The moment in the journey when you lost. Often before they told you.
What ultimately made them the right choice for you? The decisive factor. Sometimes it’s real (feature, fit, reference); sometimes it’s structural (relationship, geography, incumbent inertia).
What was the single biggest concern you had about going with us? The structural objection we couldn’t overcome.
Was price a factor, and if so, how? Always ask. Often the answer is “not really” even when the AE thought it was.
If we had done one thing differently, would the outcome have changed? The actionable insight. If yes, the gap is fixable; if no, the gap is structural.
How did our sales process compare to the competitor’s? The competitive process comparison. Sometimes the loss is about us, sometimes about them.
Was there a moment when you almost chose us instead? The pivot we almost won. The strongest indicator of where to invest in differentiation.
If a peer asked you about us, what would you tell them? The post-loss positioning of us in their network. This determines whether the loss becomes anti-marketing.

Who runs the interviews — and why it’s not the AE.

The interviewer needs to have no incentive in the answer. The AE who closed the deal will hear what they want to hear. The CSM renewing the account will steer the conversation. The product manager whose feature was named will over-weight it. The interviewer needs to listen for what the buyer actually says, not for the answer that flatters the team.

WHO RUNS THE INTERVIEW	FIT	WHY IT WORKS (OR DOESN’T)
Product Marketing Management (PMM) lead	Strong fit	Has the right frame (positioning + competitive intel) and no deal incentive. The most common owner in mature B2B SaaS.
Head of Marketing	Good fit	Senior enough to get the call. Has the synthesis instinct. Risk: too close to the messaging to hear unflattering data.
External consultant or agency	Good fit (especially at scale)	Best for loss interviews specifically — the prospect is more candid with a third party. Adds cost; reduces internal pattern recognition.
Junior PMM analyst	Workable	Cost-effective; needs strict training to avoid leading questions. Best paired with a senior reviewer.
The AE who closed	Bad fit	Will fish for validation; will steer away from uncomfortable answers. Use only for the CRM “why we won” field, not for the structured interview.
The CSM renewing	Bad fit	Has renewal incentive. Will reframe complaints as opportunities to improve, not as the loss-pattern they actually are.

Cadence and threshold — the structural rules.

The most common win/loss failure mode: ad-hoc interviewing where the team picks “interesting” deals to debrief. That sample is biased toward the deals the AE remembers. The discipline is making the program structural — defined by a threshold and a cadence, not by anyone’s judgment.

THE STRUCTURAL CADENCE

The bar is: every deal that meets the threshold gets an interview, every quarter, without exception. The threshold itself is calibrated to the deal flow.

Wins: every closed-won over $X ACV (typical: $25K–$50K for mid-market SaaS). At higher volume, you can sample — but the sample is randomized, not curated.
Losses: every closed-lost over $X ACV AND every loss against a named top-3 competitor regardless of ACV.
No-decision: deals that died without a buying decision are the third bucket. Interview a sample; the absence of decision is its own signal.
Cadence: interviews within 30 days of close (wins) and 14 days of loss (losses). Synthesis published quarterly. Annual deep-dive on competitor-specific patterns.

The synthesis discipline — turning conversations into themes.

An interview is data. A theme is intelligence. The synthesis discipline turns 15–30 interviews per quarter into 5–10 named themes with frequency counts and specific recommendations. The structure:

Code every interview against a fixed schema. The same fields for every interview: decisive factor, competitor, persona champion, top objection, success criteria, almost-lost-because-of, etc. The schema makes patterns visible across the corpus.
Tag verbatim quotes. The themes need to be backed by direct quotes. A theme without quotes is a theory.
Count frequency. “3 of 12 wins cited integration depth as the decisive factor” is signal. “Some buyers mentioned integration” is not.
Cross-reference against the playbook. Does the theme match the Right-to-Win statement? Does it match a persona’s objection list? Does it match a pricing assumption? Where the theme diverges from the playbook, the playbook needs updating.
Recommend specific actions. Each theme ends with the action: update the Right-to-Win, refresh the persona, change a pricing tier, rewrite a battlecard, ship a feature, tune an SDR opener.

The quarterly readout structure

SECTION	WHAT IT COVERS
1. The number	Win rate this quarter vs. last quarter, by segment, by competitor. The one slide the CFO and CRO look at first.
2. Why we won	Top 5 decisive factors with frequency counts and verbatim quotes. Tagged to the persona who cited each.
3. Why we lost	Top 5 decisive factors with frequency counts. Differentiate “we lost because they were better” vs. “we lost because we didn’t execute” vs. “we lost because of structural reasons we can’t fix.”
4. Competitor-by-competitor	For each of the top-3 competitors: win rate this quarter, decisive factors when we beat them, decisive factors when they beat us, change vs. last quarter.
5. What changed from last quarter	The themes that are new this quarter. The themes that disappeared. The themes that intensified.
6. Recommended actions	5–7 specific actions, each owned by a named function (PMM, demand, product, sales enablement). With deadlines.

Where the data feeds downstream.

Win/Loss doesn’t own a Brief section of its own — instead, it updates the other sections. The flow:

WIN/LOSS INSIGHT	WHICH BRIEF SECTION IT UPDATES
Decisive win factors that recur across deals	Brief Section 4 — the statement gets sharpened. Brief Section 6 (Brand) — pillars get reinforced.
Persona-level objections that the AE couldn’t overcome	Brief Section 3 (Personas) — objections list gets expanded. The disarming response gets workshopped.
Price-related losses	Brief Section 6 (Pricing) — discount policy gets reviewed; competitive positioning gets adjusted.
Competitor-specific patterns	Brief Section 5 (Competitive Intel) — battlecards get updated; the Market Watch agent gets new monitoring targets.
Product feature gaps cited in losses	Product team intake — with frequency count and ARR impact estimate. Not a Brief field; goes to product PMM directly.
Sales-process failure modes	Sales enablement — specific stage where deals are slipping. Not a Brief field; goes to sales ops.

Common failure modes.

THE FIVE WIN/LOSS PROGRAM FAILURE MODES

The AE writes a sentence in the CRM and calls it the program. The single most common failure. No structured interview, no thematic synthesis, no action loop.
Interviews happen but synthesis never does. 50 interviews sitting in a Notion database that nobody reads. The data exists; the intelligence doesn’t.
Synthesis happens but action never does. The quarterly readout is published, leadership nods, nothing changes. No named owners on the recommendations.
Selective sampling. The team picks “interesting” deals to debrief. The sample is biased toward the deals the AE wants to revisit, not toward the random truth.
The AE runs the interview. The AE fishes for validation. The data is poisoned. The themes look great until the next quarter when nothing improves.

Refresh cadence and ownership.

Win/Loss is a continuous program, not an annual exercise. The cadences:

Per-deal interview within 14–30 days of close. The interview happens, the structured coding happens, the verbatim quotes get captured.
Monthly review of interview pipeline. Are we hitting the threshold? Are losses being interviewed (the harder ones)? Owned by PMM.
Quarterly synthesis published to product / marketing / sales / leadership. The full readout. With recommended actions and named owners. Owned by PMM with sign-off from Head of Marketing.
Annual deep-dive on competitor-specific patterns. What changed in our win rate vs. each top-3 competitor over the year? What does it tell us about the next year? Owned by PMM with cross-functional review.

CAPTURE IN YOUR OPERATOR BRIEF

Win/Loss doesn’t own its own Brief section — it updates the others. The brief-builder cards below capture the program design: the threshold, the cadence, the owner, the synthesis target audience. Once the program design is signed, the agent below runs the loop.

Brief-builder: Win/Loss program design

Your Win/Loss program design

Threshold, cadence, owner, synthesis audience. Save to your Brief.

Win interview threshold (ACV minimum or competitor-triggered)

Loss interview threshold (ACV minimum or competitor-triggered)

Interview cadence (days from close)

Synthesis cadence and audience

Named owner

Brief-builder: Win/Loss interviewer

Who runs the interviews

Critical for unbiased data. The interviewer must have no incentive in the outcome.

Primary interviewer (PMM, Head of Marketing, external)

Recording & transcription tool

Where transcripts live

Brief-builder: Theme synthesis schema

How you code interviews and surface themes

The schema that makes patterns visible across the corpus.

Coding fields per interview (the fixed schema)

Quote-capture rule (verbatim, attributed by deal + persona)

Theme threshold (frequency count to surface a theme)

Brief-builder: Quarterly readout structure

The quarterly readout

How the synthesis gets published. The artifact that drives action.

Readout audience (who gets it and in what order)

Required sections in the readout

Action-item discipline (named owner, deadline, follow-up)

The prompt pack

Five prompts to run the loop.

Prompt 1

Win interview script personalized to the deal

Given a closed-won deal, generates a personalized 30-minute interview script with the eight questions tailored to the named competitors, the buying committee, and the deal context.

You are a Win/Loss interviewer for [COMPANY NAME]. I am providing a closed-won deal context: - Customer name (won’t use in interview): [CUSTOMER] - Industry / segment: [SEGMENT] - ACV: [ACV] - Sales cycle (days): [CYCLE] - Named competitor in the deal: [NAMED COMPETITOR] - The AE’s “why we won” field: [AE WHY WE WON] - Champion title: [CHAMPION TITLE] Draft a 30-minute interview script with the eight canonical win questions, personalized to this deal: 1. At what point did you feel confident we were the right choice? 2. What was the #1 thing that differentiated us from [NAMED COMPETITOR]? 3. Was there anything we could have done better during the sales process? 4. You championed this internally — what made you advocate for us? 5. What objections did you have to overcome to make this purchase? 6. How will you measure success in 12 months? 7. Who else was on the buying committee and what was their role? 8. What almost made you choose [NAMED COMPETITOR]? For each question, also give me: - The follow-up probe to use if the first answer is shallow - The signal that the answer is hitting a real theme vs. surface compliment Return as a script with timing per question. Total run-time: 30 minutes.

Prompt 2

Loss interview script personalized to the deal

Same as Prompt 1, but for losses. The harder interview to land — the script emphasizes minimal-pitch, candid-listening framing that gets prospects to talk.

You are a Win/Loss interviewer for [COMPANY NAME] conducting a loss interview. Deal context: - Prospect name (won’t use): [PROSPECT] - Industry / segment: [SEGMENT] - Lost to: [WINNING VENDOR] - ACV at close: [ACV] - Sales cycle (days): [CYCLE] - The AE’s “why we lost” field (likely incomplete): [AE WHY WE LOST] - Champion title (the person who almost championed us): [CHAMPION TITLE] Draft a 25-minute interview script with the eight canonical loss questions: 1. When did you make the decision to go with [WINNING VENDOR]? 2. What ultimately made them the right choice for you? 3. What was your single biggest concern about going with us? 4. Was price a factor, and if so, how? 5. If we had done one thing differently, would the outcome have changed? 6. How did our sales process compare to theirs? 7. Was there a moment when you almost chose us instead? 8. If a peer asked you about us, what would you tell them? Critical: the script should open with explicit no-pitch framing — this is a 25-minute listening conversation, not a save-the-deal call. Voice DOs: [VOICE DOS]. Voice DON’Ts: [VOICE DON’TS]. For each question, give me the follow-up probe and the signal that the answer is hitting a real theme vs. polite deflection.

Prompt 3

Theme synthesis across N interviews

Hand the prompt N coded interview transcripts. Returns 5–10 named themes, each with frequency count, verbatim quote support, and the cross-reference to the playbook section that should update.

You are a Win/Loss synthesis analyst for [COMPANY NAME]. I am providing [N] interview transcripts from the last quarter, coded against this schema: [CODING SCHEMA]. The corpus includes: - [N WINS] wins - [N LOSSES] losses - [N NO-DECISIONS] no-decisions Inputs from the Operator Brief: - Right-to-Win: [RIGHT TO WIN] - Personas: [ECONOMIC BUYER PERSONA], [CHAMPION PERSONA], [TECHNICAL EVALUATOR PERSONA], [END USER PERSONA] - Pricing position: [COMPETITIVE PRICE POSITION] - Top competitors: [TOP COMPETITORS] Surface 5–10 named themes from the corpus. For each theme: 1. Theme name (one short phrase) 2. Frequency count (X of N interviews cited it) 3. Persona attribution (which persona role cited it most) 4. Verbatim quotes (2–3 supporting quotes, attributed to deal type) 5. Cross-reference to playbook (which Brief Section should this update? Right-to-Win? Personas? Pricing? Competitive Intel?) 6. Recommended action (one specific change, with named owner) Then synthesize across all themes: - The single biggest opportunity to act on this quarter - The single biggest risk if no action is taken - One theme that contradicts our current playbook assumption (if any) Return as a markdown table of themes plus a 1-paragraph executive summary.

Prompt 4

Competitor-specific win/loss teardown

For a single named competitor, synthesizes all wins/losses against them in the last 12 months and returns the structural pattern: when we win, when they win, what’s changed quarter-over-quarter.

You are a competitive win/loss analyst for [COMPANY NAME]. Competitor: [NAMED COMPETITOR] Period: last 12 months Wins against them in this period: [N WINS] Losses to them in this period: [N LOSSES] Win-rate trend by quarter: [QUARTERLY TREND] I am providing the coded interview data for all deals against this competitor. Build a competitive teardown covering: 1. When we win against them — the patterns. Which segments, which deal sizes, which decisive factors. 2. When they win against us — the patterns. Same dimensions. 3. The structural advantages we have (where we win consistently) 4. The structural advantages they have (where they win consistently) 5. The change vs. last 12 months (themes intensifying, themes disappearing) 6. The single most-impactful action to improve our win rate against them 7. The battlecard updates the action implies Return as a competitive teardown document. The format used to brief the AE who’s about to enter a deal against this competitor.

Prompt 5

Quarterly leadership readout draft

Drafts the full quarterly leadership readout. Six sections, named owners on the action items, ready for sign-off by the Head of Marketing before publication.

You are drafting the quarterly Win/Loss leadership readout for [COMPANY NAME]. Inputs: - Q[X] win rate: [WIN RATE]; vs. Q[X-1]: [PRIOR WIN RATE]; by segment + by competitor - N interviews this quarter: [N] - Synthesized themes (from Prompt 3): [THEMES] - Recommended actions: [RECOMMENDED ACTIONS] - Functions involved: PMM, Marketing, Product, Sales, Sales Enablement Draft the full quarterly readout with these sections: 1. The number — win-rate by segment + by competitor, with delta vs. last quarter 2. Why we won — top 5 decisive factors with frequency counts and verbatim quotes 3. Why we lost — top 5 decisive factors, differentiated by structural vs. execution vs. fixable 4. Competitor-by-competitor — for each of top 3 competitors, win rate this quarter and change 5. What changed from last quarter — new themes, disappearing themes, intensifying themes 6. Recommended actions — 5–7 actions, each with named owner and 30/60/90-day deadline Length target: 3–5 pages including the data tables. Voice: operator-direct. Voice DOs: [VOICE DOS]. Voice DON’Ts: [VOICE DON’TS]. Forbidden language: [FORBIDDEN LANGUAGE].

The agent

Win/Loss Agent — the agent that runs the loop.

Win/Loss Agent

Runs the win/loss program end-to-end. Triggers structured 30-minute interviews on every closed-won and closed-lost > $50K within 14 days. Synthesizes themes quarterly. Rewrites positioning, pricing, and the product roadmap.

Who is this agent

Identity card

NameWin/Loss Agent

RoleWin/loss program operator — the rarest, highest-leverage discipline in B2B SaaS

OwnerHead of Product Marketing

Reports toVP Marketing + CRO

Versionv0.5 (supervised)

SurfaceClaude Project + Postgres (interview + theme corpus) + integration with transcription

Output target/win-loss/interviews/, /win-loss/themes/, /win-loss/quarterly-synthesis/

Review cadencePer-interview within 14 days; quarterly thematic synthesis; quarterly roadmap session

Mission

Run the rarest, highest-leverage discipline in B2B SaaS as a program, not a vibe. Trigger a structured 30-minute interview on every closed-won and closed-lost over $50K within 14 days of close. Conduct via a third party or PMM — never the AE who lost the deal. Synthesize themes quarterly. Surface named-account patterns. Feed positioning, pricing, and the roadmap with the highest-signal qualitative input in the company.

Goals & KPIs the agent moves

Leading indicators — the agent controls these

Interview-trigger compliance (closed deals > $50K with interview request sent < 14 days of close)≥ 90%

Quarterly synthesis shipped within 21 days of quarter close100%

Lagging indicators — downstream outcomes with review triggers

Interview-to-completion rate. Trigger: completion drops below 55% for 2 consecutive quarters pages the VP Marketing for interviewer-fit + incentive review.Target 60–75%

Theme-to-action conversion (themes assigned a named owner + due date within 30 days of synthesis). Trigger: conversion < 60% for a quarter pages the CMO for cross-functional follow-through review.Target ≥ 70%

What it does

Task list

Real-time When a CRM opportunity moves to Closed-Won or Closed-Lost with ACV > $50K, draft the interview request and route to PMM (or the third-party interviewer).
Daily Track interview booking + completion status. Nudge unbooked interviews after 7 days; escalate after 14.
Weekly Compile weekly digest: interviews booked / completed / pending / surfacing themes.
Weekly Theme tag every completed interview against the eight canonical question categories. Surface emerging themes.
Monthly Cross-interview theme synthesis. Pattern-mine for themes ≥ 3 interviews. Surface to ICP / Persona / RtW / Pricing / Brand / Competitive agents.
Quarterly Compile the quarterly synthesis — 3–5 themes with named accounts attached, frequency, direction (wins / losses / mixed), recommended actions split by owner (Product, Marketing, Sales).
Quarterly Roadmap session: present synthesis to Product + Marketing + Sales leadership. Drive owner assignments + due dates on each theme.
Event When a theme contradicts a Brief section, surface to Brief Sync Agent immediately (don’t wait for quarterly).
Event When a closed-lost interview surfaces a new competitor, route to Market Intelligence Agent within 24 hours.
Event When a theme matches Pricing Intelligence’s discount-drift signal, link the two for the quarterly session.

Schedule grid

Task	Frequency	Duration	Output goes to
Real-time interview-request drafting	Continuous (on opportunity close)	~5 min per draft	PMM or third-party interviewer
Daily booking + nudge status	Daily 09:00	~10 min	Head of PMM
Weekly digest + theme tagging	Weekly Fri 14:00	~45 min	Head of PMM + VP Marketing
Monthly cross-interview synthesis	Monthly 1st	~3 hours	VP Marketing + downstream agents
Quarterly synthesis compile	Quarterly Q+15 days	~6 hours	CEO + VP Marketing + CRO + CPO
Quarterly roadmap session	Quarterly Q+21 days	~2 hours	Cross-functional leadership

Triggers

Scheduled (cron-style):

Schedule	What it runs
`0 9 * * *`	Daily booking + nudge status
`0 14 * * 5`	Weekly digest + theme tagging
`0 9 1 * *`	Monthly cross-interview synthesis
`0 9 15 1,4,7,10 *`	Quarterly synthesis compile

Event-driven:

Event	What it runs
CRM opportunity moves to Closed-Won or Closed-Lost with ACV > $50K	Draft interview request + route within 24 hours
Interview unbooked at T+7 days from close	Nudge PMM + AE
Interview unbooked at T+14 days from close	Escalate to VP Marketing
Completed interview contradicts a Brief section	Route to Brief Sync Agent within 24 hours
Closed-lost surfaces new competitor not in Market Watch set	Route to Market Intelligence Agent within 24 hours

Who it works with

Inputs

Source	Type	Cadence	Required?
Operator Brief (Sections 2, 3, 4, 5, 6)	Markdown	Read every run	Required
CRM closed-won + closed-lost events	Webhook + API	Real-time	Required
Interview transcripts (Otter / Fireflies / Granola)	Text / JSON	Per-interview	Required
Eight-question framework (wins + losses)	Markdown	Versioned	Required — standard interview structure
Account Intel Hub closeout dossier	Markdown	Per-opportunity-close	Required — provides context for the interview
Revenue Attribution Engine per-deal touch history	JSON	Per-opportunity-close	Required

Outputs

Output	Format	Target path	Audience
Interview request drafts	Markdown + Slack message	/win-loss/requests/<opp-id>.md	PMM or third-party interviewer
Per-interview write-ups	Markdown	/win-loss/interviews/<opp-id>.md	Internal corpus + downstream agents
Weekly digest	Markdown	/win-loss/digests/YYYY-WW.md	Head of PMM + VP Marketing
Monthly theme synthesis	Markdown	/win-loss/themes/YYYY-MM.md	ICP + Persona + RtW + Pricing + Brand + Competitive agents
Quarterly synthesis	Markdown + chart bundle	/win-loss/quarterly-synthesis/Q<n>.md	CEO + VP Marketing + CRO + CPO + Board if material
Theme-to-action tracker	Markdown table	/win-loss/action-tracker.md	VP Marketing + each owner

↑ Upstream — agents/sources that feed this one

Operator Brief (human-maintained). Sections 2-6 provide the lens for interpreting interview signal.
Signal Router. Routes closed-won + closed-lost events to trigger interview requests.
Account Intel Hub. Closeout dossier provides full account history for the interview prep.
Revenue Attribution Engine. Per-deal touch history surfaces which marketing touches were most influential.

↓ Downstream — agents/humans that consume its output

ICP Researcher Agent. Receives ICP-trigger themes that may require scorecard refresh.
Persona Researcher Agent. Receives persona-dimension themes (new objections, trigger events).
Right-to-Win Council. Receives positioning-contradiction themes.
Market Intelligence Agent. Receives new-competitor surfacing + competitive theme patterns.
Pricing Intelligence Agent. Receives price-related loss themes.
Brand & Positioning Agent. Receives brand + voice themes.
Proof Library Agent. Receives reference-candidate flags from closed-won interviews.
Brief Sync Agent. Receives Brief-contradicting themes for PR drafts.

Human escalation paths

Trigger condition	Escalate to	Within
Interview compliance < 80% in a quarter	VP Marketing + CRO	Same quarter
Quarterly synthesis missed Q+21-day deadline	VP Marketing + CEO	Immediate
Closed-lost interview reveals a contract-blocking issue (legal, security, compliance)	VP Marketing + CRO + General Counsel	< 24 hours
Theme contradicts core positioning (RtW)	VP Marketing + CEO	< 48 hours
Theme appears in ≥ 5 interviews with no named action owner	VP Marketing + cross-functional leadership	< 14 days

How to build it

System prompt

You are the Win/Loss Agent for [COMPANY]. YOUR JOB Run win/loss as a program, not a vibe. Trigger interviews within 14 days of close on every deal > $50K. Conduct via PMM or third party - never the AE. Synthesize themes. Feed positioning, pricing, and roadmap with the highest- signal qualitative input in the company. INPUTS (always read in this order) 1. /operator-brief.md (Sections 2-6) 2. /win-loss/eight-questions.md - the canonical interview framework (wins and losses) 3. /accounts/closeout/<opp-id>.md - account context for interview prep 4. /win-loss/interviews/* - historical interview corpus 5. Today's CRM close event (the trigger) OUTPUTS - /win-loss/requests/<opp-id>.md (interview request draft) - /win-loss/interviews/<opp-id>.md (per-interview write-up) - /win-loss/themes/YYYY-MM.md (monthly synthesis) - /win-loss/quarterly-synthesis/Q<n>.md (quarterly) THE EIGHT-QUESTION WIN FRAMEWORK 1. At what point did you feel confident we were the right choice? (Decisive moment) 2. What was the #1 thing that differentiated us from the other vendors? (Decisive factor in their words) 3. Was there anything we could have done better during the sales process? (Friction we didn't see) 4. Who was the internal champion for this decision, and what made them advocate for us? (Champion profile) 5. What objections did you have to overcome internally to make this purchase? (Hidden objections) 6. How will you measure the success of this purchase in 12 months? (Their success criteria) 7. Who else was on the buying committee and what was their role? (Committee map) 8. What almost made you choose someone else? (Pivot moment) THE EIGHT-QUESTION LOSS FRAMEWORK 1. When did you make the decision to go with [WINNER]? (Moment you lost) 2. What ultimately made them the right choice for you? (Decisive factor) 3. What was the single biggest concern you had about going with us? (Structural objection) 4. Was price a factor, and if so, how? (Price reality check) 5. If we had done one thing differently, would the outcome have changed? (Actionable vs. structural) 6. How did our sales process compare to the competitor's? (Process comparison) 7. Was there a moment when you almost chose us instead? (Pivot we almost won) 8. If a peer asked you about us, what would you tell them? (Post-loss positioning) RULES 1. Never interview via the AE who worked the deal. 2. Interview within 14 days of close. 3. Every theme cites the interview IDs + verbatim quotes. 4. Quarterly synthesis: 3-5 themes max, each with named accounts, frequency, direction, action-by-owner. 5. Never soften losses. Operator-direct. ESCALATION - Theme contradicts RtW: VPM + CEO within 48h. - Compliance/security issue surfaced: VPM + GC within 24h.

Tools & integrations

Platform / tool	Used for	Required?
Claude Project + Postgres (interview + theme corpus)	Reasoning + theme synthesis	Required
Otter / Fireflies / Granola API	Interview transcription	Required
Salesforce / HubSpot API + webhook	Closed-won + closed-lost trigger events	Required
Calendly / Chili Piper API	Interview scheduling	Required
Slack API	Request routing + weekly digest	Required
Third-party interviewer relationship (Klue / Crayon / Skalata or similar)	Optional	Optional but recommended for closed-lost

Guardrails — what it must not do

Never interview via the AE who worked the deal. Hard rule.
Never share customer verbatim quotes outside the marketing + sales scope without consent.
Honor interviewee anonymity where requested. Some quotes attribute; some don’t.
Never use interview content in marketing without explicit consent at the time of interview.
Never compress interview themes — the verbatim signal IS the signal.
Never penalize an AE for interview content surfacing in losses — the program exists to learn, not assign blame.
Honor data retention — interview content stored in compliance with the customer’s data terms.

Evals + hallucination defense

Evals — output quality checks:

Interview compliance. Quarterly: % of closed deals > $50K with interview booked < 14 days. Target ≥ 90%.
Interview completion. Quarterly: of booked, % completed. Target ≥ 70%.
Theme-to-action. Quarterly: of named themes, % with named owner + due date. Target ≥ 80%.
Action close-out. Quarterly: of themes with owners, % closed by next quarter. Target ≥ 60%.

Hallucination defense — specific checkpoints:

Themes must cite specific interview IDs + quote ranges. No paraphrased generalities.
Frequency claims must trace to specific interviews in the corpus (“8 of 12 interviews” with IDs).
Named-account attribution must respect the interviewee’s consent flag.
When evidence is mixed, surface both sides explicitly.
Never extrapolate from one interview to a pattern.

Maturity curve + first-run checklist

v0.1 — Manual-assistAgent drafts interview requests on-demand. PMM runs interviews + synthesis. Useful from day 1 to formalize the discipline.

v0.5 — SupervisedReal-time trigger on. Daily booking watch. Weekly + monthly + quarterly synthesis. PMM signs off on every quarterly. Default ship state.

v1.0 — Semi-autonomousAfter 90 days of clean evals + theme-to-action ≥ 80%, agent auto-routes monthly synthesis to downstream agents without PMM gate. Quarterly synthesis stays human-approved.

First-run checklist — 5 steps from spec to running agent:

Author the eight-question wins and losses framework + interview-request template with Head of PMM.
Establish the interviewer roster — PMM team + third-party partner for closed-lost.
Wire CRM webhook to trigger requests on close events. Test with the next 3 deals.
Run the first quarter end-to-end with full PMM review. Tune theme-tagging.
Subscribe VP Marketing + CRO + CPO to quarterly synthesis. Schedule the quarterly roadmap session. Log every run.

Win/Loss Analysis.