Problem Statement
The goal is to help a customer success team proactively identify accounts that are at risk of churn. By analyzing activity logs, health scores, and renewal timelines, the system classifies churn risk levels and recommends appropriate interventions. This enables customer success managers to focus on high-value and high-risk accounts with timely, relevant outreach.
Steps
Ingest Account Activity Data
INGESTCollect and parse account activity logs, support ticket counts, health scores, and renewal dates from the provided sources. Standardize the inputs and validate that all required fields are present.
Retrieve Historical and Value Indicators
EXECUTERetrieve account value tier, historical churn patterns, and any past intervention outcomes to inform prioritization. This contextual data supports more accurate risk assessment.
Generate Churn Risk Classification
EXECUTEAnalyze usage activity, health scores, support tickets, and renewal timing to determine risk level. Identify leading indicators contributing to the classification and produce a structured interpretation.
Validate Intervention and Prioritization
EXECUTECheck that recommended outreach steps match defined rules and prioritize high-value accounts when risks are similar. Ensure results follow constraints and generate the final recommended timing and approach.
Prompt Configurations
Churn Analysis and Recommendation Prompt
You are an AI assistant supporting a customer success team. Analyze the provided account data: activity logs, logins in the last 30 days, support ticket count, health score, contract renewal date, and account value tier. Classify churn risk as low, medium, or high using only the defined rules: high risk = <2 logins in 30 days OR >3 support tickets OR health score <40; medium risk = 2–5 logins OR health score 40–60; otherwise low. Identify leading indicators without inventing any data. Recommend an appropriate intervention (email, call, or discount offer) and suggest ideal outreach timing based on renewal proximity. Tone must be data-driven, proactive, and empathetic. Only use information explicitly provided.
Example Inputs/Outputs
Account: Acme Corp Logins (30 days): 1 Support tickets: 4 Health score: 32 Account value: High Renewal date: 2026-04-15
Churn Risk: High Leading Indicators: Very low login activity, multiple support tickets, low health score Recommended Intervention: Personalized call from CSM Ideal Timing: Within the next 3 days due to upcoming renewal