objection-mining

name: objection-mining description: Aggregates objections from call notes across territory and builds a living response library. Trigger with "what are they pushing back on", "objection library", "log this objection", "what objections am I seeing", "objection patterns", or "update my objection library". version: 1.0.0

Objection Mining

Purpose: Turn individual objections from every call into a territory-wide pattern library. Every call is a data point. This skill connects them.

Input: ## Objections Logged sections in all accounts/*/discovery/ files. Output: memory/objection-library.md — append only, never overwrite.

STEP 0 — Confirm Scope

1. Single account: "Log the objection from my last call with [company]" → Read most recent discovery/ file for that account only
2. Territory sweep: "Update my objection library" or "What are they pushing back on?" → Scan all accounts/*/discovery/ files

STEP 1 — Scan Call Notes Files

Read all files matching accounts/*/discovery/*.md.

For each file, look for the ## Objections Logged section:

## Objections Logged
| Type | Verbatim Quote | Stage | Response Used | Outcome |

If no section found → skip it. If found → extract all rows.

Types: Timing / Budget / Authority / Trust / Fit / Competition

STEP 2 — Parse and Classify

For each objection row:

1. Confirm or correct the Type classification
2. Note the Stage (what deal stage was this raised at?)
3. Note the Response Used and Outcome
4. Check if new or a variant of an existing one

STEP 3 — Aggregate into Library

Read memory/objection-library.md (create if doesn't exist).

For each objection:

• Find the matching entry or create a new one
• Increment count
• Add the new verbatim quote
• Update win/loss pattern if outcome is known

Library format:

# Objection Library
*Last updated: YYYY-MM-DD | {N} objections tracked | {N} accounts*

---

## Timing Objections
**Count:** {N} | **Win rate:** {N}% | **Most common stage:** {stage}

### Most Common Responses (by win rate)
1. [Response that worked — with example]
2. [Second best response]

### Verbatim Quotes (latest 5)
- "[Quote]" — {Company}, {Stage}, {Date}

---

## Budget Objections
[Same structure]

## Authority Objections
[Same structure]

## Trust Objections
[Same structure]

## Fit Objections
[Same structure]

## Competition Objections
[Same structure]

---

## Top 5 Objections by Frequency

| Rank | Objection | Type | Count | Win Rate | Best Response |
|------|-----------|------|-------|----------|--------------|
| 1 | | | | | |

STEP 4 — Generate Response Templates

For any objection type where Count ≥ 3 AND no "Best Response" is documented:

Generate a template using the AAA framework:

Type: [Timing/Budget/Authority/Trust/Fit/Competition]
Pattern: "[Common version of this objection]"

AAA Response:
- Acknowledge: "[Validate without agreeing]"
- Ask: "[Reframe question to understand root cause]"
- Advance: "[Earn the next 30 seconds]"

Outcome goal: [What constitutes a win for this objection]

STEP 5 — Save

Append to memory/objection-library.md — never overwrite the full file.

Quality Gates

• All discovery/ files scanned for ## Objections Logged sections
• Each objection type-classified correctly
• Frequency counts updated
• Win rate updated where outcome is known
• Response templates generated for types with 3+ examples and no template
• memory/objection-library.md updated (append only)
• Top 5 frequency table updated