Win/loss interview synthesis (L3)

Record win/loss calls, LLM extracts themes monthly, posts to PMM Slack. Replaces the consultant.

WORKFLOW1Conduct and Transcribe In…erviewsManual2Synthesize Themes with LL… CitationsManual3Human Validation and Tagg…ngManual4Distribute Monthly Insigh…s to SlackManual5Conduct Quarterly Roadmap…ReviewManual
5 steps, in order, with the tool that owns each one.
Adoption ladderSix levels from Starter to Rebuilt. This item sits at level 3.L1 StarterOne tool, no workflow changeL2 AssistedAI drafts, humans approveL3 IntegratedWired into CRM and SlackL4 OrchestratedMulti-step, owned, measuredL5 AutonomousAgent runs, human auditsL6 RebuiltThe process itself changes
This playbook belongs at L3 Integrated. Running it above your level is how pilots stall.
Measures of successthemes shipped to product; competitive win rate changePROVE IT WORKEDthemes shipped to productcompetitive win rate change

The steps

  1. 01

    Conduct and Transcribe Interviews

    The Product Marketing Manager (PMM) must secure a minimum of 10 win/loss interviews monthly. To automate this, use a tool like Gong, Chorus, or Grain to record and transcribe calls. Configure your CRM (Salesforce or HubSpot) to trigger a 'Win/Loss Interview Request' sequence via Outreach or Salesloft once an Opportunity moves to 'Closed Won' or 'Closed Lost.' • Set up a Zapier automation: When a Deal stage changes to 'Closed,' send a Slack notification to the PMM with the buyer's contact info. • Use a booking link (Calendly) specifically for these interviews to avoid manual back-and-forth. • Ensure you have a standardized question script covering: 'Why us?', 'Why them?', 'Product Gaps,' and 'Sales Experience.' Owner: Product Marketing Manager. Estimation: 5-8 hours of interviewing + 2 hours of setup. Prerequisites: CRM access and a call recording tool. Pitfall: Relying on Sales to book the calls; Sales often cherry-picks 'happy' customers. You must automate the request directly to the buyer. Definition of Done: 10 completed transcripts stored in a dedicated folder or workspace.

  2. 02

    Synthesize Themes with LLM Citations

    Move transcripts into an LLM-ready format. If using OpenAI or Anthropic, you can use their Batch API or a playground interface. The goal is to extract specific themes while enforcing 'Grounding' (citations). Use this prompt template: 'Analyze the following 10 interview transcripts. Identify the top 5 recurring reasons for winning and losing. For every claim, provide a direct quote and the transcript timestamp. Format: [Theme Name] - [Summary] - [Evidence: "Quote" - Time 00:00]. Focus on pricing, feature gaps, and competitor mentions.' • Go to the OpenAI Playground, select 'Chat' mode, and use 'GPT-4o.' • Paste the transcripts into the 'System' or 'User' message box. • Adjust 'Temperature' to 0.3 to keep the output factual rather than creative. Owner: PMM or RevOps. Time Estimate: 1 hour. Prerequisites: Access to GPT-4/Claude 3.5. Pitfall: Grouping too many interviews in one prompt can lead to 'hallucinations' or the LLM forgetting the middle transcripts (Long Context Window issues). Definition of Done: A structured document containing 5-7 major themes with at least 2 citations per theme.

  3. 03

    Human Validation and Tagging

    The PMM must manually validate the LLM's output. Even high-performing AI can misinterpret sarcasm or project a minor complaint as a major theme. • Review the 'Evidence' quotes provided by the AI. Click into the original recording tool (Gong/Chorus) to listen to the 30 seconds surrounding that quote to ensure context is preserved. • Map these themes to specific departments: Product (Feature Request), Sales (Enablement/Closing Skill), or Marketing (Messaging/Positioning). • Create a 'Theme Matrix' in Google Sheets or Notion. Columns: Theme, Frequency, Impact (High/Med/Low), Action Item, and Owner. Owner: Product Marketing Manager. Time Estimate: 2 hours. Prerequisites: LLM output and access to call recordings. Pitfall: Taking the AI summary at face value without clicking the timestamps; the 'why' is often hidden in the tone, not just the text. Definition of Done: A verified insight report where every theme has been human-confirmed and assigned an impact score.

  4. 04

    Distribute Monthly Insights to Slack

    Deliver the insights where the team actually works. Set up a dedicated Slack channel (e.g., #win-loss-insights) and post a monthly 'State of the Buyer' summary. • Format the Slack post using blocks: 🏆 TOP WIN THEME: [LLM Theme] ❌ TOP LOSS THEME: [LLM Theme] 🎙️ BUYER VOICE: [Insert clearest 'Loss' quote] 🛠️ ACTION: [Proposed roadmap fix or messaging change] • Use Slack's 'Schedule Message' feature to post this during high-engagement hours (Tuesday at 10 AM). • Mention key stakeholders (@ProductHead, @VPofSales) to ensure it doesn't get buried. Owner: PMM. Time Estimate: 30 minutes. Prerequisites: Slack access and Validated Report. Pitfall: Posting a massive PDF wall of text. People won't read it. Use bullet points and focus on one 'Killer Quote.' Definition of Done: Monthly summary posted to the designated Slack channel with confirmed engagement from leadership.

  5. 05

    Conduct Quarterly Roadmap Review

    Convert qualitative feedback into quantitative roadmap inputs. Schedule a 45-minute 'Quarterly Insight Review' with the Product and Sales Leadership teams. • Prepare a slide deck comparing themes over the last 3 months. Highlight 'Trend Shifts' (e.g., 'Competitor X is appearing in 40% more losses than last quarter'). • Use the 'Competitive Win Rate' metric from your CRM to show how these themes correlate with revenue. • Drive a specific outcome: Product should commit to adding one high-loss feature to the backlog, and Marketing should commit to updating one sales battlecard. Owner: PMM. Time Estimate: 3 hours (prep + meeting). Prerequisites: 3 months of synthesized data. Pitfall: Letting the meeting become a 'venting session' for Sales. Keep it focused on the evidence-based themes and the data provided by the LLM synthesis. Definition of Done: Meeting minutes showing at least one roadmap adjustment or sales training session scheduled based on the data.

Next playbooks

Unfamiliar terms are defined in the AI and Revenue Dictionary. Related frameworks live in the framework library.

Share this playbook

Posting to Instagram or TikTok? Copy the link, it carries the title, summary and share image.

Arrives weekly by email. Free. Unsubscribe anytime. By subscribing you agree to our Privacy policy and Terms. We never sell or share the list.