You are here:
Configure a Propensity to Buy Template (Beta)
Configure the Propensity to Buy (beta) template to automatically create the data transforms, output data model objects, and predictive model required to generate propensity scores. Query the scores to assess a customer's likelihood of making a purchase within your specified prediction window.
Required Editions
| Available in: All Editions supported by Data 360. See Data 360 edition availability. |
| USER PERMISSIONS NEEDED | |
|---|---|
| Allow users to manage models in AI Models | Enables you to create, update, and delete models in AI Models. |
| PERMISSION SETS | |
|---|---|
| Data Cloud Architect | Admin-level access to all AI Models features, including the ability to create, update, delete, and activate models. |
| Data Cloud User | Restricted access to use a model, including getting predictions and improvements derived from a model. |
Set Up and Query the Template
-
Configure and run the template.
- Click the App Launcher, search for Data Cloud, and then select it.
- Click the AI Models tab.
- From the Predictive section, click Create from Template.
- Select Propensity to Buy, and then click Next.
-
Specify a name and description for your template configuration.
The system names all generated data transforms and model assets based on the configuration name that you provide.
- Select the target data space, the product that you want to evaluate, and the time horizon (prediction window). Optionally specify a product category.
- To run the template configuration, click Train.
-
Monitor the template installation.
To monitor the template installation, from Setup, go to App Hub > Scoring, and Templates.
https://<org-domain-url>/lightning/setup/AppHub/home (Org Setup → App Hub)
https://<org-domain-url>/lightning/setup/ScoringFramework/home (Org Setup → Scoring)
https://<org-domain-url>/lightning/setup/TemplateStudio/home (Org Setup → Templates)
- The template installations create these seven assets: 3 transforms, 3 output DMOs, and 1 model. To verify that all assets are successfully created and executed, go to these Data 360 tabs: DataTransforms, Data Model Object, and AI Models.
-
Propensity Training Transform
-
Propensity Inference Transform
-
Propensity Scoring Transform
-
Propensity Scoring Model
-
PropensityTraining__dlm(output DMO)
-
PropensityInference__dlm (output DMO)
-
PropensityScoring__dlm (output DMO)
Asset names vary depending on the configuration name that you specify. For example, if you name your configuration "Shirt", the template automatically creates these assets.
Shirt Scoring Transform Definition
ShirtPropensityScoringData1__dlm
Shirt - Propensity Scoring Model
-
Query propensity scores.
To validate the generated scores that predict purchase likelihood within your specified prediction window, run a query on the
PropensityScoring__dlm
data lake object (DLO) in the Query Editor. For example, run this query on your data space to return the first 1,000 records.SELECT Id__c, ProductPropensityScore__c FROM PropensityScoringData__dlm ORDER BY Id__c DESC LIMIT 1000Replace PropensityScoringData__dlm with the name of your output DMO with predictions. For example, ShirtPropensityScoringData1__dlm.
To verify that your template assets were successfully created and executed based on your configuration name, first go to the Data Transforms tab and verify that these assets exist. Then go to the AI Models tab and verify that your model is active. Check that your template created the seven assets based on your configuration name.
Considerations
Data Requirements
Use at least 6 months of purchase history and behavioral engagement data, such as browse and cart events, for the individuals in your data space. Model Builder requires at least 400 records to build a model. Limited transaction history or sparse engagement data can result in scores that don't reflect purchase intent.
Individuals with no purchase history or minimal engagement data receive low-confidence scores. Treat these scores as insufficient data rather than low purchase intent. Products need historical sales data to produce meaningful propensity scores.
Score Interpretation
The ProductPropensityScore__c value represents the relative likelihood that an individual purchases the specified product or product category within the configured prediction window. Most individuals receive low scores because the base purchase rate for a single product is typically below 5%.
Segment scores by decile or percentile for activation. For example, target the top 10–20% of scored individuals instead of applying a fixed numeric threshold. Score distributions vary by product and data volume.
Time Horizon and Freshness
The prediction window affects the model's predictions. Shorter windows favor recent behavioral signals, such as active cart events, while longer windows weigh historical purchase patterns more heavily.
Scores are point-in-time predictions and become unreliable after the prediction window passes. To keep scores current, rerun the inference transform at a cadence equal to or shorter than the configured prediction window.
Scheduling and Execution
The template creates Training, Inference, and Scoring batch data transforms that run in sequence. The transforms aren't automatically scheduled after initial creation. Manually rerun or schedule the transforms and model retraining to keep scores current. Execution time increases with data volume.
Scale
The template generates scores for each individual-product or individual-product category combination. The output DMO (PropensityScoring__dlm) can grow significantly when a data space contains many individuals and products. Plan your storage and query strategy accordingly.
Each template configuration targets one product or product category. To score multiple products, create a separate configuration for each product or select a product category during configuration.
Credit Consumption
The template consumes credits when it creates transforms for training and scoring data. Because this template is deterministic, it doesn't consume agentic credits.
Other
The template supports only the standard Data Cloud objects and fields listed in Propensity to Buy Template (Beta). You can't add custom DMOs or custom fields to the model configuration. Scoring is batch only. Real-time, event-triggered scoring isn't supported.
