mcpField
C
customer-learning-notes

customer-learning-notes

github.com/lvtd-llcMarketing & growth160 installsUpdated 7w ago
Repository

About customer-learning-notes

A skill for turning customer conversations into traceable evidence and decisions. It helps teams distinguish observed facts from interpretations, assess confidence, and choose next research or product steps.

Key features

  1. 01Signal taxonomy for notes
  2. 02Evidence and interpretation separation
  3. 03Segment-based evidence grouping
  4. 04Contradiction and confidence review
  5. 05Belief and next-step updates

Use cases

  1. 01Synthesizing interview notes or transcripts
  2. 02Comparing customer segments and workarounds
  3. 03Assessing evidence strength before product decisions

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/lvtd-llc/skills --skill customer-learning-notes

After installing, type/customer-learning-notesin Claude Code to confirm it loaded.

Bundled filesagents/openai.yamlworkflows/synthesize-conversation-notes.md

Details

From the project README

Customer Learning Notes

Use this skill after customer conversations to turn raw notes into team-readable evidence. Good notes make it harder to misremember, overfit, or let one founder become the sole source of customer truth.

Source Traceability

Primary source: The Mom Test by Rob Fitzpatrick, especially chapter 8 and the conclusion. Guidance is paraphrased for this MIT repo; authoring notes used converted EPUB lines 3613-4445.

Signal Taxonomy

Use these labels when synthesizing notes:

LabelMeaning
PainProblem, obstacle, annoyance, risk, or cost
GoalDesired outcome, job to be done, or priority
WorkaroundCurrent manual process, tool stack, hack, or substitute
MoneyBudget, cost, value, purchase process, or decision owner
PersonSpecific stakeholder, competitor, team, buyer, or intro lead
FeatureRequest, buying criterion, integration need, or implementation clue
EmotionStrong excitement, anger, embarrassment, fear, or skepticism
Follow-upPromise, task, intro, research item, or next step

Synthesis Workflow

  1. Preserve concrete facts separately from interpretation.
  2. Pull out short, useful quotes only when they are needed for traceability, positioning, or internal alignment.
  3. Tag signals using the taxonomy above.
  4. Group evidence by segment, problem, workaround, budget, and commitment.
  5. Identify contradictions and mixed-segment noise.
  6. Update beliefs, risks, and the next three questions.
  7. Recommend whether to continue, narrow the segment, ask for commitments, or move to building/testing.

Confidence Levels

LevelUse When
HighRepeated behavior from a focused segment, with concrete cost or commitment.
MediumSpecific evidence from a few good-fit conversations.
LowOne-off quotes, mixed segments, opinions, or weakly anchored claims.

Output Format

# Customer Learning Synthesis

## Source Notes
- Conversations:
- Segment:
- Date range:

## Evidence
| Signal | Evidence | Segment | Confidence | Implication |
|--------|----------|---------|------------|-------------|

## Belief Updates
- Stronger / weaker / new / rejected:

## Decisions
- Product, segment, positioning, sales or access:

## Next 3 Questions
1. [Question]
2. [Question]
3. [Question]

Workflow

Use workflows/synthesize-conversation-notes.md when the user provides raw notes, transcripts, call summaries, or interview excerpts.

Quality Bar

  • Do not summarize notes into vibes.
  • Do not let one loud quote outweigh repeated behavior from a focused segment.
  • Do not mix segments without labeling them.
  • Do not treat notes as useful until they have been reviewed and turned into updated beliefs or decisions.

Comments

Markdown supported