← Market Analyst · c7
findings.md
Cycle 7 Findings: Avoma Notes Structure
Contents
Based on 7 sample calls (Brooks/Dropbox, Starbucks, Surepath, Wunderkind, Dropbox POC, Airbnb, Cam)
Note Structure Overview
Each call's notes are a flat array of blocks — headers (header-2) with nested content blocks (lists, paragraphs, transcript-type blocks with timestamps). Two types of sections:
1. Standard Sections (template-driven)
These appear consistently across sales/demo calls (5 of 7 samples):
| Section | Block Key | Present in | Content Quality |
|---|---|---|---|
| Participants | cb8hm | 7/7 | Company-grouped attendee lists |
| Agenda | — | 5/7 | Standard template items |
| Key Takeaways | highlight_category_key_takeaways | 7/7 | 4-6 timestamped highlights |
| Background | highlight_category_background | 7/7 | Context on relationship/history |
| Current Workflow | highlight_category_current_workflow | 7/7 | Prospect's existing tools/processes |
| Business Need | highlight_category_need | 5/7 | What they need from Qpoint |
| Impact | highlight_category_impact | 5/7 | Expected value/outcome |
| Feature Mention | highlight_category_feature_mention | 5/7 | Specific features discussed |
| Budget | highlight_category_budget | 5/7 | Pricing discussion (often empty) |
| Timeline | highlight_category_timeline | 7/7 | Timelines, deadlines, next milestones |
| Risk | highlight_category_risk | 7/7 | Concerns, blockers, risks identified |
| Pain Point | highlight_category_pain_point | 5/7 | Explicitly extracted pain points |
| Positive Moment | highlight_category_positives | 5/7 | Positive signals, praise, enthusiasm |
| Competitor | highlight_category_competitor | 5/7 | Competitive mentions (often empty) |
| Team Size | highlight_category_team_size | 5/7 | Team/org size (often empty) |
| Decision Making | highlight_category_decision_making | 5/7 | Who decides, what process |
| Action Item | highlight_category_action_item | 7/7 | Person-attributed action items with timestamps |
| Follow up | highlight_category_next_meeting_date | 5/7 | Next meeting/follow-up timing |
2. Technical Chat Template
The 2 technical/internal calls (Cam, Surepath) use a different template:
| Section | Block Key |
|---|---|
| Problem Statement | highlight_category_problem_statement |
| Proposed Solution | highlight_category_proposed_solution |
| Alternatives Considered | highlight_category_alternatives_considered |
| Technical Tradeoffs | highlight_category_technical_tradeoffs |
| Stakeholder Feedback | highlight_category_stakeholder_feedback |
| Decisions Made | highlight_category_decisions_made |
| Open Questions | highlight_category_open_questions |
3. AI-Generated Custom Sections (per-call)
Each call gets 2-4 additional AI-generated sections marked with machine_title: true. These are topic-specific and vary per call. Examples:
- "AI Security Insights", "Deployment Strategy", "Data Classification Focus"
- "Retention and Storage Solutions", "Integration with Existing Infrastructure"
- "Market Positioning", "Future Collaboration"
Transcript Blocks
Many note items are type: "transcript" with a data.time field (seconds into call). This links each insight directly to the source moment in the transcript. Other items are type: "paragraph" (no timestamp — typically template text or AI summaries).
High-Value Fields for Our Analysis
Already extracted by Claude (overlap)
- Pain Points — Avoma has its own list; could serve as validation baseline
- Feature Mentions — Similar to our feature requests
- Competitor mentions — Same concept, different extraction
- Key Takeaways — Overlaps with our summary
NEW data not in our current extraction
- Action Items — Person-attributed with timestamps. We don't extract these at all.
- Budget signals — Explicit pricing discussions with quotes
- Decision Making — Who has authority, what the approval process looks like
- Timeline — Concrete next steps with dates
- Current Workflow — What tools/processes they use today
- Positive Moments — Direct quotes showing enthusiasm
- Background — Relationship history, how they found Qpoint
- Risk — Specific concerns and blockers from prospect's perspective
- Follow up — Next meeting dates and commitments
- Participants — Company-grouped with roles (cleaner than our attendee parsing)
Data quality observations
- Notes correctly interpret transcription errors (e.g., "Centimeters" → "SIEM")
- Empty sections are common (Budget, Competitor, Team Size often have no content)
- AI-generated custom sections provide surprisingly specific topical analysis
- Timestamps enable linking notes directly to transcript moments
- Action items are attributed to specific people — excellent for follow-up tracking
How This Helps Our Analysis
1. Improved extraction quality
- Use Avoma notes as supplementary input alongside raw transcripts
- Notes pre-extract pain points, needs, timeline, budget — reducing Claude's extraction work
- Timestamped references let us verify/ground-truth our own extraction against Avoma's
2. New dashboard visualizations
- Action item tracking — who owes what across all calls
- Budget/pricing view — which calls mentioned specific price points
- Decision maker mapping — who has authority at each company
- Follow-up timeline — when next meetings are expected
- Risk/blocker heatmap — what concerns come up most
3. Better PMF signals
- Positive Moments give us verbatim value echoes we might miss
- Current Workflow reveals the status quo we're displacing
- Background shows relationship depth and referral chains
- Timeline + Decision Making together signal deal velocity
Consistency Assessment
Across 7 samples:
- Very consistent (7/7): Participants, Key Takeaways, Background, Current Workflow, Timeline, Risk, Action Items
- Consistent for sales calls (5/7): Business Need, Impact, Feature Mention, Budget, Pain Point, Positive Moment, Competitor, Decision Making
- Template-dependent (2/7): Problem Statement, Proposed Solution, etc. (technical chat template)
- Always unique (7/7): AI-generated custom sections (2-4 per call)