Predictive NPS (PNPS)
summary
Claims to predict NPS for every customer rather than the usual 10–15% survey response by producing 100% coverage scores from combined CRM billing usage feedback and complaint data, using a productized backend that runs as microservices in a Kubernetes cluster on AWS with Docker, Jenkins, Bitbucket, Postgres and API endpoints. Models start with a linear baseline then random forest LGBM and neural nets, use calibration and report AUC/Gini, precision and recall, variable importance and downloadable diagnostics for Close‑the‑Loop and churn prevention.
AI-written summary · about
chapters (16)
- 0:00Predictive NPS intro
- 1:21Introduction
- 2:03Why CX matters today
- 3:28Understanding NPS
- 5:56Low NPS response rates
- 8:04PNPS as solution
- 9:32How PNPS works
- 13:05PNPS data sources
- 14:09Modeling process stages
- 15:37Descriptive analysis & variables
- 17:03Modeling techniques
- 18:06Model evaluation metrics
- 19:50Calibrating predictions
- 21:15PNPS tech stack
- 25:58Cloud deployment & automation
- 28:27Q&A
related talks
13mBeyond Surveys: Real-Time User Satisfaction Prediction
26mImproving customer experience through journey...
56mPredictive Analytics for Practitioners
18mTransforming Financial Decision Intelligence Through Advanced NNs
07mScaling Predictive Manufacturing Analytics: Kubernetes-Powered CRM
43mPredictive Maintenance: From POC to Production With Spark
11mDrug Launch Success Through Prompt-Driven Sales Data Analytics
02h 03mStarting Your First Predictive Analytics Project