AI Customer Health Scoring for Customer Success

AI-based Customer Health Score

Customer Health Score (CHS) is a comprehensive indicator of how engaged a customer is with a product and the likelihood of renewal or churn. For B2B SaaS and subscription services, it is a leading indicator of NRR (Net Revenue Retention). The ML approach allows for the transition from CSM intuition to an objective, repeatable assessment.

Signals for the model

Product Usage:

• DAU/WAU/MAU: activity of the main users of the account
• Feature adoption: percentage of key features adopted in the last 30 days
• Depth of use: advanced features vs. basic ones
• Stagnation: Decline in usage over the past 4 weeks

Support & Success signals:

• Open tickets without a response > 3 days - negative signal
• NPS, CSAT scores for the last 6 months
• The EBR (Executive Business Review) took place – a positive signal
• Escalations: management-level complaints

Commercial indicators:

• Renewal date proximity: < 90 дней → повышенное внимание
• Expansion or contraction in the last 12 months
• Invoice payment delays: late payments
• Contract modifications: attempts to revise the terms

Relationship signal:

• Sponsor changes: the champion left the account - high risk
• Multi-threading: how many contacts the CSM knows (< 2 = single-threaded)
• Last meaningful interaction: when was the last time there was a real conversation

Feature Engineering

def compute_customer_health_features(account_id, lookback_days=90):
    usage = get_product_usage(account_id, lookback_days)
    support = get_support_tickets(account_id, lookback_days)
    commercial = get_crm_data(account_id)

    return {
        # Usage trends
        'usage_trend_slope': np.polyfit(range(lookback_days), usage['daily_active_users'], 1)[0],
        'feature_adoption_score': len(usage['active_features']) / total_key_features,
        'power_user_ratio': usage['high_frequency_users'] / usage['total_seats'],

        # Support health
        'open_critical_tickets': support[support['priority'] == 'critical']['count'],
        'avg_resolution_time_days': support['avg_resolution_time'],
        'recent_nps': support['last_nps_score'],

        # Commercial
        'days_to_renewal': (commercial['renewal_date'] - today).days,
        'logo_expansion_12m': commercial['arr_change_12m'],
        'payment_delay_days': commercial['avg_payment_delay'],

        # Relationship
        'sponsor_change_6m': commercial['sponsor_changed_flag'],
        'contacts_known': commercial['known_contacts_count'],
        'days_since_last_call': (today - commercial['last_substantive_contact']).days
    }

Models and architecture

Composite Score (correct baseline):

def rule_based_health_score(features):
    score = 100 # start at 100

    # Usage penalties
    if features['usage_trend_slope'] < -0.1:
        score -= 20
    if features['feature_adoption_score'] < 0.3:
        score -= 15

    # Support penalties
    if features['open_critical_tickets'] > 0:
        score -= 25
    if features['recent_nps'] and features['recent_nps'] < 7:
        score -= 15

    # Commercial risk
    if features['days_to_renewal'] < 60 and features['logo_expansion_12m'] < 0:
        score -= 20

    return max(0, min(100, score))

ML model floor: LightGBM or Logistic Regression trained on historical "renewed/departed" data after 12 months. Advantage vs. rule: identifies nonlinear interactions (for example, a decrease in usage alone is not a risk, but when combined with an upcoming renewal, it is critical).

Temporal validation:

# Walk-forward: we train on cohorts < 12 months ago, we predict on later ones
# Metric: AUC of 90-day churn prediction
# Baseline: just the renewal date + the latest NPS

Risk segmentation and actions

Risk Tiers:

Automated Playbooks:

def trigger_playbook(account, health_score, reason_codes):
    if health_score < 30:
        crm.create_task(owner='csm_manager', type='urgent_review', account=account)
        slack.notify('#csm-alerts', f"CRITICAL: {account.name} score={health_score}")
    elif health_score < 50 and 'usage_decline' in reason_codes:
        gainsight.enroll_in_playbook(account, 'activation_campaign')
    elif health_score > 80 and account.days_to_renewal < 90:
        crm.create_opportunity(account, type='expansion', amount=account.arr * 0.2)

Integration with CS tools

Gainsight, ChurnZero, Totango: These platforms already have native health scoring. A custom ML model exports the score via an API → replaces or supplements the built-in rules.

CRM (Salesforce, HubSpot): Custom field "AI Health Score" on the Account object. Updated weekly. Used in reports and forecasting.

Product Analytics (Amplitude, Mixpanel, Heap): Usage data source. Separate API/export for B2B account-level aggregation.

Timeframe: Feature pipeline + rule-based scoring + Salesforce integration — 3-4 weeks. ML model training and validation + automated playbook triggers + Gainsight integration — 6-8 weeks.