Customer Value Leakage Engine

A clean-room Python proof of concept that finds where customers lose expected value during onboarding, activation, service handoffs, and early product use.

What it does

• Generates deterministic synthetic feedback
• Measures prevalence, tNPS, effort, and unresolved rates
• Ranks leakage signals with a transparent 0-100 priority score
• Identifies the strongest leakage theme by journey stage
• Maps each signal to a concrete operational action
• Produces JSON and an executive HTML brief

Why it matters

Traditional CX reporting often explains satisfaction after the fact. This engine focuses on the front end of the relationship, where unclear onboarding, failed activation, mismatched expectations, and broken handoffs can destroy value before customers fully adopt the product.

Run

python run_demo.py
python -m pytest -q

Open:

outputs/value_leakage_brief.html

Priority logic

The score combines signal prevalence, tNPS, effort, and unresolved rate. It supports prioritization but does not establish causality.

Clean-room notice

All feedback, journeys, scores, themes, outputs, and recommendations are fictional and independently created. No employer data, customer data, private prompts, or confidential metrics are included.