--- name: call-transcript-analyzer description: > Transforms a raw sales or discovery call transcript into a rich, interactive HTML deal intelligence report. Use this skill whenever a user pastes or uploads a call transcript and wants analysis, a follow-up email, win probability, use case ranking, or any kind of post-call debrief. Also triggers for phrases like "analyze this call", "create a report from this transcript", "what did we learn from the call", "score this sales call", or "make a deal report". Optionally accepts a discovery prep doc alongside the transcript for deeper customer fit analysis. The output is a branded, tabbed single-file HTML page with scores, signals, service fit checklist, use cases, improvement feedback (including conversation flow analysis), a follow-up email, next actions, an overall win assessment, and a LinkedIn content tab for marketing use. --- # Call Transcript Analyzer Turn any sales or discovery call transcript into a complete deal intelligence report — a polished, interactive HTML page the user can open in a browser, share, and act on immediately. ## Inputs **Required:** - Call transcript (pasted text or uploaded file) **Optional:** - Discovery prep doc (pre-call research / context doc). If provided, cross-reference it with the transcript to surface gaps: things flagged as important pre-call that were never covered on the call. Use it to give a fuller customer fit picture beyond what was said live. ## What this skill produces A single-file HTML report with 10 navigable tabs: 1. **Deal Snapshot** — key facts extracted from the call (company, stage, stakeholders, urgency) 2. **Interest & Buying Signal Analysis** — scored interest metrics with visual bars, positive signals vs. risk flags 3. **Service Fit** — checklist-based scoring against the user's service/product capabilities 4. **Use Cases** — identified use cases ranked by priority and likelihood, each expandable with root cause + opportunity 5. **Value Proposition** — concrete value delivered per use case, anchored to what the prospect actually said 6. **What We Could Have Done Better** — honest improvement feedback including conversation flow analysis 7. **Follow-Up Email** — a ready-to-send email with a one-click copy button 8. **Next Actions** — split into "before next call" and "in the next call" columns 9. **Overall Assessment** — deal quality, competitive risk, velocity, win probability with reasoning 10. **LinkedIn Content** — privacy-first marketing content based on the call's pain points (see below) ## Brand configuration Default palette (replace with your own brand colours): ``` --yellow: #f3af00 --blue: #207796 --light-blue: #dff3fa --charcoal: #201600 --white: #ffffff Font: Muli (Google Fonts) ``` ## How to analyze a transcript Work through the transcript methodically before writing any HTML. Extract these dimensions: ### People & roles - Who is on the call from each side? - Who is the decision-maker / budget holder? - Who is the internal champion (the one who feels the pain most)? - Who are the technical or operational stakeholders? - Who was missing that should have been there? ### Pain & problems - What operational or business problems did they describe? - Were problems stated voluntarily (unprompted = high authenticity) or only after being asked? - How specific were the descriptions? Vague frustration vs. concrete numbers/incidents? - What is the financial or operational cost of each problem (even if stated implicitly)? ### Buying signals — positive Look for: unprompted disclosure of pain, specific metrics shared, emotional reactions (surprise, relief, excitement), senior time committed, proactive questions about next steps, references to budget cycles or timelines. ### Buying signals — risk / caution Look for: late arrivals, vague responses, mentions of existing vendors or tools, resource/bandwidth concerns, lack of budget signal, no timeline shared, passive language ("maybe", "we'd have to see"). ### Use cases For each distinct problem area: name it, assign a priority tier (Critical / High / Medium / Low), estimate the likelihood that this is a real underlying issue (0-100%), and articulate: - What they said (direct evidence from the transcript) - Root cause (why the problem exists structurally) - Our opportunity (how the offering addresses it) ### Scores (0-100%) Assign scores to these dimensions based on evidence from the call: - Overall Interest (headline score) - Problem Urgency - Solution Fit - Decision Authority Present - Timeline Clarity - Budget Signal - Champion Strength Use green (>=65%), orange (40-64%), red (<40%) for colour coding. --- ## Service Fit Checklist (Tab 3) Score the prospect against each specific capability of your offering (workshops, consulting, training, SaaS, etc.). For each item, mark: - ✅ **Confirmed** — evidence from the call or prep doc that this applies - ⚠️ **Possible** — partial signal, needs validation - ❌ **Not evidenced** — no signal Generic Service Fit Checklist (customise per engagement type): 1. Problem is a genuine priority (not just a "nice to have") 2. Budget signal is present or implied 3. Decision authority is in the room (or identified) 4. Timeline urgency makes this actionable within 1–2 quarters 5. Scope matches your delivery sweet spot (not too small, not too unwieldy) 6. Champion strength — someone internally wants this to happen 7. Data / inputs / access to run the engagement are realistic 8. Organisational readiness (management buy-in, not just one eager sponsor) 9. Willingness to adopt standard method vs. demand heavy customisation 10. Expansion potential — can this engagement grow into a bigger relationship? **Important:** If the call surfaces a one-off scope or a proof-of-concept rather than a genuine engagement, **flag this clearly** with a warning box: "⚠️ Engagement Risk: The proposed project may be a one-off. Consider whether this fits your ideal client profile." Overall Service Fit Score: calculate as % of checklist items Confirmed or Possible. --- ## Improvement Feedback — Conversation Flow Analysis (Tab 6) In addition to standard improvement feedback, always include a dedicated **Conversation Flow** section: - Did the prospect take control of the agenda or redirect the conversation unexpectedly? - If so: at what point, and what was the trigger? - What specific re-directing question or bridge could have been used to bring the conversation back? - Example recovery moves: "That's a great point — I want to make sure we come back to that. Before we go deep there, can I ask you about X for just 2 more minutes?" or "I love where you're going — let me just anchor us back to the bigger picture first." Be specific and honest. Reference what actually happened in the transcript and suggest concrete alternatives. --- ## Follow-up email - Subject line specific to the call outcome, not generic - Open with a genuine observation from the call, not boilerplate - Reference the 2-3 use cases most strongly evidenced - Propose a concrete next step (session type, duration, agenda hint) - Under 250 words — warm and specific, short enough to be read - Include a [proposed date/time] placeholder --- ## Win probability Single percentage with a 2-3 sentence rationale grounded in specifics from the call. --- ## LinkedIn Content Tab (Tab 10) Generate privacy-first marketing content based on the call. The goal is to turn what was heard into a story that can be published externally — for the content/marketing owner to review before posting. **Rules:** - Never identify the actual company or individuals - Change the industry sub-sector slightly (e.g. automotive → industrial equipment) to obscure identity - Build a fictional but realistic ICP (Ideal Customer Profile) that this scenario would plausibly apply to - The story should feel authentic and grounded, not generic **Content structure to generate:** ### 1. The Problem Story (LinkedIn post format, ~200 words) - Hook: A provocative or empathetic opening line - The pain: Describe the operational challenge in vivid terms (symptoms, daily friction, cost) - The insight: What most companies get wrong about this problem - The solution angle: How your offering addresses it (generic framing) - CTA: Soft call to action ("Sound familiar? Let's talk.") ### 2. Use Case Brief (internal reference for content team) - Problem / Use Case name - Which businesses this is most prevalent in - Symptoms to look for - Key metrics that could be improved - Capability that addresses it ### 3. Content Tags Suggest 5 relevant LinkedIn hashtags. Include a copy-to-clipboard button for the LinkedIn post text. --- ## Output: HTML report Build one self-contained HTML file. The file must: - Use only the brand colours and a single Google Font (loaded from Google Fonts CDN) - Be fully self-contained — no external JS dependencies except Google Fonts - Work on mobile (responsive layout, horizontal-scroll tabs) - Have a copy-to-clipboard button on the email tab and on the LinkedIn content tab - Animate tab transitions with a CSS fadeUp keyframe - Have expandable cards (use cases, improvements) toggled by inline JavaScript - Display a prominent warning banner on Tab 3 if Service Fit Score is below 50% or if a one-off engagement risk is detected ## Delivering the output Save the HTML to `/mnt/user-data/outputs/_deal_report.html`. Use `present_files` to share it with the user. Follow with a 2-3 sentence plain-language summary: overall interest score, Service Fit score, and the single most important thing to fix before the next call.