Prompt Example

Language Model: ChatGPT-4 Data Analyst

Summary: Using synthetic data for 1,000 customers, analyze customer segmentation by demographics (age, gender, income), behavioral data (purchase history, browsing patterns), and psychographics (interests, values, lifestyle). Identify segments with the highest lifetime value, pinpoint segments at risk of leaving (churn), and identify customers who are high-value and at-risk.

Prompting techniques used:
Chain-of-Thought (CoT): Guide the language model to reason through each type of data (demographics, behavioral, psychographics) to categorize customers into segments. For each segment, the model calculates metrics like average lifetime value and churn risk.

Few-shot learning: Provide the model with examples of how customer data is analyzed to derive insights about customer segments. This includes examples of how to calculate lifetime value and churn risk based on behavioral patterns and demographic data.

Visualization Generator Pattern: Enhance understanding and decision-making by instructing the model to generate textual descriptions for data visualization. This includes descriptions for creating bar charts of customer lifetime value by segment or heat maps of engagement levels.