Predicting Net Promoter Score (NPS) to Improve Patient Experience at Manipal Hospitals Custom Case Solution & Analysis

1. Evidence Brief: Data Extraction and Classification

Financial Metrics

Revenue Impact: Patient retention is 5 to 25 times less expensive than acquisition; a 5 percent increase in retention can grow profits by 25 percent to 95 percent.
NPS Scale: Calculated as percentage of Promoters (score 9-10) minus percentage of Detractors (score 0-6).
Market Position: Manipal Hospitals is the second-largest multi-specialty healthcare provider in India with over 7,000 beds across 27 hospitals.
Operating Costs: Significant capital is tied to bed turnover rates and length of stay (LOS) efficiency.

Operational Facts

Data Volume: The case utilizes a dataset of approximately 14,000 patient records containing 34 variables.
Patient Segmentation: Two primary categories are Outpatient (OPD) and Inpatient (IPD).
Key Variables: Patient age, gender, department (e.g., Cardiology, Orthopedics), admission type, length of stay, and specific wait times for billing and pharmacy.
Current Process: NPS surveys are conducted post-discharge, resulting in reactive rather than proactive service recovery.
Infrastructure: Centralized HIS (Hospital Information System) captures operational timestamps but lacks real-time predictive integration.

Stakeholder Positions

Executive Leadership: Focused on transitioning from a volume-based model to a value-based, patient-centric model to drive long-term growth.
Data Science Team: Tasked with identifying which operational drivers (e.g., discharge time, billing speed) most accurately predict a Detractor.
Front-line Staff: Experience friction during peak discharge hours (11:00 AM – 2:00 PM), which correlates with lower NPS.
Patients: Express highest sensitivity toward administrative delays rather than clinical quality.

Information Gaps

Clinical Outcomes: The data focuses on administrative touchpoints; the correlation between medical success (recovery rates) and NPS is not explicitly quantified.
Staff Sentiment: No data provided regarding nurse or physician burnout levels, which likely impacts patient interaction quality.
Competitor NPS: Benchmarking data for other major Indian hospital chains (e.g., Apollo, Fortis) is absent.

2. Strategic Analysis: From Reactive to Predictive Experience

Core Strategic Question

Can Manipal Hospitals transition from retrospective surveys to a real-time predictive model that identifies and mitigates patient dissatisfaction before discharge?

Structural Analysis

Service Value Chain Analysis: The hospital primary value is clinical, but the perceived value is heavily weighted toward administrative ease. The bottleneck exists in the Support Activities—specifically Technology Development and Procurement (Pharmacy). When administrative friction exceeds clinical benefit, Promoters shift to Passives or Detractors. The current feedback loop is too slow to allow for service recovery, meaning a Detractor is lost permanently before the hospital even knows they are unhappy.

Strategic Options

Option 1: Real-Time Intervention Engine. Integrate the predictive model into the HIS to flag high-risk Detractors 24 hours before scheduled discharge. This requires a dedicated Guest Relations task force to resolve issues (billing, insurance) in person.
- Trade-off: High operational cost in headcount vs. significant improvement in NPS.
- Requirement: Low-latency data processing and 24/7 staffing.
Option 2: Operational Bottleneck Elimination. Use model insights to re-engineer the discharge and billing process, targeting a 30 percent reduction in wait times for all patients rather than individual interventions.
- Trade-off: Long-term structural improvement vs. lack of personalized patient care.
- Requirement: Process redesign and potential capital expenditure in pharmacy automation.

Preliminary Recommendation

Pursue Option 1. The predictive model shows that specific administrative delays are the primary drivers of Detractors. By identifying these patients in real-time, Manipal can deploy targeted service recovery. This addresses the immediate revenue risk of losing a patient’s lifetime value and prevents negative word-of-mouth in a highly competitive market.

3. Implementation Roadmap: The 90-Day Execution Plan

Critical Path

Week 1-4: Data Pipeline Integration. Connect the Random Forest predictive model to the live HIS feed to generate daily risk scores for all IPD patients.
Week 5-8: Pilot Launch. Deploy the intervention model at the flagship Bangalore hospital. Assign Guest Relations Managers (GRMs) to visit every patient flagged as a potential Detractor (probability > 0.7).
Week 9-12: Feedback Loop Validation. Compare actual NPS scores of the intervention group against a control group to measure the delta in Detractor reduction.

Key Constraints

Data Accuracy: The model is only as effective as the timestamps entered by nursing staff. Manual overrides or late entries will break the predictive accuracy.
Staff Buy-in: Clinical staff may view Guest Relations interventions as an interference with medical discharge protocols.

Risk-Adjusted Implementation Strategy

To mitigate the risk of staff friction, the intervention should be framed as an administrative support function. GRMs will handle insurance documentation and pharmacy delivery—tasks that currently burden nurses. This ensures the strategy improves both patient experience and operational efficiency. If the pilot shows less than a 10 percent reduction in Detractors, the model must be retrained to include physician-patient interaction variables.

4. Executive Review and BLUF

BLUF

Manipal Hospitals must deploy a real-time predictive NPS engine to halt patient churn. Administrative delays in billing and discharge are the primary drivers of Detractors, not clinical failures. By identifying high-risk patients 24 hours before discharge, the hospital can execute targeted service recovery. This shift from reactive measurement to proactive management will reduce Detractors by an estimated 15 percent within the first year, securing long-term patient lifetime value and protecting market share against rising private competition.

Dangerous Assumption

The analysis assumes that administrative service recovery can compensate for clinical dissatisfaction. If a patient is unhappy with the medical outcome, a faster billing process will not convert them into a Promoter. The model risks over-indexing on what is measurable (timestamps) while ignoring what is meaningful (clinical empathy).

Unaddressed Risks

Data Integrity Risk: Probability: High. Consequence: Severe. If front-line staff manipulate discharge timestamps to meet KPIs, the predictive model will receive corrupted data, leading to incorrect intervention targets.
Resource Dilution: Probability: Medium. Consequence: Moderate. Focusing Guest Relations exclusively on predicted Detractors may lead to a decline in service for Passives, inadvertently pushing them into the Detractor category.

Unconsidered Alternative

The team failed to consider a tiered pricing or service model. Instead of treating all Detractors equally, Manipal could prioritize interventions for high-value segments (e.g., long-stay surgical patients or repeat chronic care patients) where the financial impact of churn is highest. This would ensure the highest return on the Guest Relations payroll.