1. NPS Question (0-10 scale)
2. Open-ended follow-up: "What is the primary reason for your score?"
3. Categorical follow-up (optional): "Which area most influenced your rating?"
   - Product quality
   - Customer service
   - Pricing/value
   - Ease of use
   - Other (specify)
4. Demographic/segmentation questions (2-3 max)

# For B2B contexts:
"How likely would you be to recommend [product] to a colleague in a similar role?"

# For specific features:
"Based on your experience with [feature], how likely would you be to recommend it?"

# For competitive differentiation:
"Compared to alternatives you've considered, how likely would you be to recommend us?"

# Sample segmentation logic
survey_segments = {
    'new_customers': {
        'trigger': 'days_since_signup >= 30',
        'frequency': 'once',
        'timing': 'post_onboarding'
    },
    'active_users': {
        'trigger': 'last_activity <= 7_days AND tenure >= 90_days',
        'frequency': 'quarterly',
        'timing': 'relationship'
    },
    'support_interactions': {
        'trigger': 'ticket_closed AND satisfaction_rating IS NULL',
        'frequency': 'per_interaction',
        'timing': '24_hours_post_resolution'
    }
}

# Minimum sample size calculation
import math

def min_sample_size(confidence_level=0.95, margin_error=0.05, population_size=None):
    z_score = 1.96  # 95% confidence
    p = 0.5  # Maximum variability

    n = (z_score**2 * p * (1-p)) / margin_error**2

    if population_size:
        n = n / (1 + (n-1)/population_size)

    return math.ceil(n)

# For statistically significant NPS: minimum 100 responses per segment

-- NPS by customer segment
SELECT 
    customer_segment,
    COUNT(*) as responses,
    ROUND(AVG(nps_score), 2) as avg_score,
    ROUND(
        (SUM(CASE WHEN nps_score >= 9 THEN 1 ELSE 0 END) * 100.0 / COUNT(*)) - 
        (SUM(CASE WHEN nps_score <= 6 THEN 1 ELSE 0 END) * 100.0 / COUNT(*)),
        1
    ) as nps,
    ROUND(SUM(CASE WHEN nps_score >= 9 THEN 1 ELSE 0 END) * 100.0 / COUNT(*), 1) as promoter_pct,
    ROUND(SUM(CASE WHEN nps_score <= 6 THEN 1 ELSE 0 END) * 100.0 / COUNT(*), 1) as detractor_pct
FROM nps_responses 
WHERE survey_date >= CURRENT_DATE - INTERVAL '90 days'
GROUP BY customer_segment
ORDER BY nps DESC;

detractor_response:
  trigger: "score <= 6"
  timeline: "within 24 hours"
  owner: "customer_success_manager"
  actions:
    - personal_outreach
    - issue_investigation
    - resolution_plan
    - follow_up_survey

promoter_activation:
  trigger: "score >= 9"
  timeline: "within 48 hours"
  owner: "account_manager"
  actions:
    - thank_you_message
    - review_request
    - referral_program_invite
    - case_study_opportunity

# Executive Dashboard
- Current NPS score with trend arrow
- Response rate % (target: >15% for email)
- Sample size and confidence interval
- Promoter/Passive/Detractor distribution
- Top 3 improvement themes from detractors
- YoY and QoQ trends

# Operational Dashboard
- NPS by customer segment, product line, geography
- Response time metrics
- Action item completion rates
- Detractor win-back success rate
- Correlation with business metrics (churn, expansion)

# NPS trend analysis with seasonality
from scipy import stats
import pandas as pd

def nps_trend_analysis(df, periods=12):
    monthly_nps = df.groupby('month').agg({
        'nps_score': ['count', 'mean'],
        'promoter': 'sum',
        'detractor': 'sum'
    }).reset_index()

    # Calculate actual NPS
    monthly_nps['nps'] = (
        (monthly_nps['promoter'] / monthly_nps['count'] * 100) -
        (monthly_nps['detractor'] / monthly_nps['count'] * 100)
    )

    # Trend significance
    slope, intercept, r_value, p_value, std_err = stats.linregress(
        range(len(monthly_nps)), monthly_nps['nps']
    )

    return {
        'trend': 'improving' if slope > 0 else 'declining',
        'significance': p_value < 0.05,
        'monthly_change': slope,
        'r_squared': r_value**2
    }