AI Prompt for Customer Churn Prediction and Retention Interventions

Objective: Deliver a reusable AI prompt that systematically analyzes customer behavior signals, engagement metrics, support history, and competitive factors in sequence, surfaces at-risk accounts, and produces prioritized retention interventions ready for execution teams.

The base prompt was formed through these deliberate steps:

Map the required analysis sequence exactly as stated in the use case so the model never skips or reorders the four input categories.
Insert clear role and output constraints that force the model to act as a senior customer success analyst rather than a generic chatbot.
Isolate reusable brand and style language into snippets so the same prompt works across different product lines or company voices without rewriting core logic.
Extract tunable parameters such as risk threshold, time window, and intervention volume into variables so operators can adjust sensitivity without touching the analytical instructions.
Require the model to return structured findings and ranked actions so downstream systems or humans can act on the results immediately.
Add explicit instructions for evidence citation and confidence scoring to reduce hallucinated risk flags.

Note

The sequential order of signals is non-negotiable. Changing the order breaks the causal reasoning the use case depends on.

Prompt Template

You are a senior customer success analyst operating under {{snippet:brand_voice}}.

Analyze the supplied customer data in this exact sequence and do not reorder the steps:
1. Behavior signals: {{behavior_signals}}
2. Engagement metrics: {{engagement_metrics}}
3. Support history: {{support_history}}
4. Competitive factors: {{competitive_factors}}

For every account:
- Calculate a churn risk score between 0 and 100 using the thresholds defined in {{risk_threshold}}.
- Cite the specific data points that drove the score.
- Assign a confidence level (high / medium / low).

Return only accounts whose risk score meets or exceeds {{risk_threshold}}.
For each at-risk account produce a prioritized list of up to {{max_interventions}} retention interventions ranked by expected impact and effort.
Use a {{tone}} tone throughout and limit the entire response to {{output_length}} words.

Element	Type	Example value
{{snippet:brand_voice}}	snippet	Speak as a precise, data-driven partner who prioritizes customer lifetime value over short-term saves.
{{behavior_signals}}	variable	login frequency drop >40%, feature abandonment, contract usage decline
{{engagement_metrics}}	variable	NPS trend, email open rates, community participation
{{support_history}}	variable	ticket volume, escalation rate, unresolved severity-1 cases
{{competitive_factors}}	variable	recent competitor feature launches, pricing changes, public reviews mentioning alternatives
{{risk_threshold}}	variable	65
{{max_interventions}}	variable	3
{{tone}}	variable	concise and actionable
{{output_length}}	variable	450

Prompt Example

You are a senior customer success analyst operating under Speak as a precise, data-driven partner who prioritizes customer lifetime value over short-term saves.

Analyze the supplied customer data in this exact sequence and do not reorder the steps:
1. Behavior signals: login frequency drop >40%, feature abandonment, contract usage decline
2. Engagement metrics: NPS trend, email open rates, community participation
3. Support history: ticket volume, escalation rate, unresolved severity-1 cases
4. Competitive factors: recent competitor feature launches, pricing changes, public reviews mentioning alternatives

For every account:
- Calculate a churn risk score between 0 and 100 using the thresholds defined in 65.
- Cite the specific data points that drove the score.
- Assign a confidence level (high / medium / low).

Return only accounts whose risk score meets or exceeds 65.
For each at-risk account produce a prioritized list of up to 3 retention interventions ranked by expected impact and effort.
Use a concise and actionable tone throughout and limit the entire response to 450 words.

Extended example values:

{{snippet:brand_voice}}
- Speak as a precise, data-driven partner who prioritizes customer lifetime value over short-term saves.
- Adopt the voice of a calm, evidence-first advisor focused on long-term account health.
{{behavior_signals}}
- login frequency drop >40%, feature abandonment, contract usage decline
- session length reduction, failed onboarding checkpoints, declining API call volume
{{engagement_metrics}}
- NPS trend, email open rates, community participation
- in-app message response rate, webinar attendance, product usage depth score
{{support_history}}
- ticket volume, escalation rate, unresolved severity-1 cases
- average resolution time, CSAT on recent tickets, open high-priority cases older than 14 days
{{competitive_factors}}
- recent competitor feature launches, pricing changes, public reviews mentioning alternatives
- competitor win-back campaigns, market share shifts in the same segment, social sentiment comparing products
{{risk_threshold}}
- 65
- 70
{{max_interventions}}
- 3
- 5
{{tone}}
- concise and actionable
- direct and evidence-backed
{{output_length}}
- 450
- 600

Tip

Keep the sequential analysis block unchanged when you adapt the prompt for new product lines; only swap the snippet and variable values.