ConvoYield — Conversational Yield Optimization Engine

Treat every bot conversation as a yield-bearing financial instrument.

ConvoYield is a zero-dependency Python skill that gives any bot the ability to detect, score, and capture hidden monetary value from every single conversational exchange — not just the ones that end in a sale.

In finance, a yield is the income return on an investment. In conversations, the yield is the total value you can extract: revenue, data, referrals, engagement, competitive intel. Most bots capture less than 20% of available conversational yield. ConvoYield fixes that.

Why This Exists

Out of 13,700+ skills in the OpenClaw ecosystem, not a single one treats conversations as financial instruments. Every bot talks. No bot optimizes the value of what it says.

ConvoYield applies concepts from behavioral economics, financial engineering, and game theory to conversational AI:

Zero Cost to Run

• ZERO external dependencies — pure Python stdlib
• ZERO API calls — all analysis runs locally via pattern matching and heuristics
• ZERO tokens consumed — doesn't call any LLM APIs
• ZERO infrastructure needed — pip install and go

The Five Engines

1. Sentiment Arbitrage Engine

Detects emotional gaps that create revenue opportunities. A frustrated user mentioning a competitor isn't just venting — it's a $45+ conversion opportunity if handled correctly.

2. Micro-Conversion Tracker

Finds hidden money in every message. Between "hello" and "purchase," there are dozens of micro-moments worth $0.50-$15 each: email captures, budget reveals, pain point articulations, referral signals.

3. Momentum Scorer

Measures whether the conversation is gaining or losing steam. Positive momentum = push for conversion. Negative momentum = pull back and re-engage before you lose them.

4. Yield Forecaster

Predicts the total dollar value of the conversation in real-time. Imagine a dashboard showing: Estimated Value: $127.50 | Captured: $35.00 | At Risk: $92.50

5. Play Caller

Recommends optimal strategic "plays" from a 20-play playbook based on behavioral economics: anchoring, loss framing, social proof deployment, empathy bridges, urgency closes, and more.

Quick Start

from convoyield import ConvoYield

engine = ConvoYield(base_conversation_value=50.0)

# Process each user message
result = engine.process_user_message("I'm frustrated with Salesforce, it's way too expensive")

print(result.recommended_play)       # "competitor_displacement"
print(result.estimated_yield)         # 89.50
print(result.recommended_tone)        # "empathetic"
print(result.top_arbitrage.type)      # "frustration_capture"
print(result.risk_level)              # 0.21

# Record bot response for full state tracking
engine.record_bot_response("I hear you. What specifically isn't working?")

# Next message — yield COMPOUNDS
result = engine.process_user_message("The reporting is terrible and costs $500/month")
print(result.estimated_yield)         # 142.30 — value is growing!

Integration

Works with any bot framework:

# Discord, Telegram, Slack, OpenClaw — same pattern:
guidance = engine.process_user_message(user_text)

# Use guidance to shape your response:
# guidance.recommended_play → WHAT strategy to use
# guidance.recommended_tone → HOW to say it
# guidance.arbitrage_opportunities → WHERE the money is
# guidance.micro_conversions → WHAT value to capture
# guidance.risk_level → HOW careful to be

See examples/openclaw_skill.py for a complete OpenClaw skill wrapper.

Install

pip install -e .

Run Tests

pip install -e ".[dev]"
pytest tests/ -v

Run Demo

python examples/basic_usage.py

Architecture

convoyield/
├── __init__.py              # Public API
├── orchestrator.py          # Main ConvoYield engine
├── engines/
│   ├── sentiment_arbitrage.py   # Emotional gap detection
│   ├── micro_conversion.py      # Value-extraction tracking
│   ├── momentum.py              # Engagement trend analysis
│   ├── yield_forecaster.py      # Dollar value prediction
│   └── play_caller.py           # Strategic play recommendations
├── models/
│   ├── conversation.py      # Conversation state model
│   └── yield_result.py      # Analysis result model
examples/
├── basic_usage.py           # See it in action
├── openclaw_skill.py        # OpenClaw/MoltBot integration
└── batch_analysis.py        # Analyze conversation logs
tests/
└── test_convoyield.py       # 40 tests, 100% pass rate

License

MIT — Use it, sell it, build on it. Every bot deserves to know what its conversations are worth.