# Batch AI Predict {% hint style="info" %} Batch AI Predict is a priced offering. Reach out to your CSM or Account Manager to learn more. {% endhint %} Batch AI Predict calculates predictive scores for each of your profiles using machine learning models trained on your data: purchase behavior, browsing events, campaign reactions, subscription status, and more. Scores are refreshed on a schedule you define and are available across the Batch interface for targeting, personalization, orchestration, and analytics. ## How it works The Scoring Engine processes your customer and event data to generate scores at profile level. Each score is stored as a profile attribute and can be used anywhere you can apply a targeting condition or a personalization variable. For product affinity scores (cross-sell, upsell, replenishment, product recommendation), you can define the categories or products to include in the model. ## Predictive Scores Batch AI Predict uses predictive AI to create personalized scores for each of your customers. Scores are refreshed on your preferred schedule and are available throughout the Batch interface for targeting, personalization, orchestration, and analytics. Each score is stored as a profile attribute and can be used anywhere you can apply a targeting condition or a personalization variable. ### Available scores {% hint style="info" %} **Parameterized scores:** Some scores can be computed for any group of products you define: a category, a sub-category, a brand, or a specific selection of SKUs. For example, a product propensity score can be calculated independently for "skincare", "haircare", and "fragrance", generating a separate attribute on each profile for each scope. {% endhint %} #### Products Recommendation (**Parameterized score)** Identify, for each customer, the top N products or product categories to promote in order to maximize repeat purchase rate and immediate conversion. **Output format:** List of N product IDs #### Cross-sell Propensity (**Parameterized score)** Identify, for each customer, the top N products or product categories to promote in order to maximize acquisition on products the customer has never purchased. **Output format:** List of N product IDs #### Product Propensity (**Parameterized score)** Identify each customer's propensity for a given product or product category (i.e., the probability that the customer will purchase this product or category in the coming months). **Output format:** Decimal between 0 and 1 #### Churn Decisive Moment Identify customers at risk of becoming inactive in the coming months and the optimal past date after which they are highly likely to become inactive. **Output format:** Date #### Second Purchase Date Identify the date after which a one-time buyer is most likely to return and make a second purchase. **Output format:** Date #### Replenishment Date Identify, for each customer, the date after which they are most likely to have finished a "consumable" product. **Output format:** List of SKUs × date #### Promotion Sensitivity Identify each customer's sensitivity to promotions (i.e., the ratio between the probability that the customer makes a purchase when not exposed to a promotion vs. the probability that they make a purchase overall). **Output format:** Decimal between 0 and 1 #### Discount Recommendation Identify, for each customer, the optimal promotion level in order to maximize both conversion rate and the associated gross margin. **Output format:** Promotion ID #### Subscription Churn Identify, for each customer, their risk of unsubscribing before the next subscription renewal (i.e., the probability that the customer churns). **Output format:** Decimal between 0 and 1 #### Top Client at Risk Identify in advance the top historical customers who are at risk of reducing their spending in the coming months. **Output format:** Boolean #### Future Lifetime Value Identify, for each customer, the total future amount they are likely to spend in the coming months or years. **Output format:** Amount in € #### High Potential Prospect Identify, for each prospect, their likelihood of being converted soon (i.e., the probability that the prospect will make their first purchase soon). **Output format:** Decimal between 0 and 1 #### High Potentiel Customer Identify, for each customer, their likelihood of becoming a high-value customer soon (i.e., the probability that the customer will generate significant revenue or place high-value orders in the near future). **Output format:** Decimal between 0 and 1 #### Best Send Time {% hint style="info" %} Best Send Time is a beta capability for now. Reach out to your CSM or Account Manager to learn more. {% endhint %} Identify, for each recipient, what is the best time to send a message to him or her within a given time window. **Output format**: Score only usable in Automation Builder, through a Best send time step. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://doc.batch.com/getting-started/features/customer-engagement-platform/batch-ai/batch-ai-predict.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.