Einstein Prediction Builder
Einstein Prediction Builder is the lightweight, admin-friendly predictive modeling tool inside Salesforce that builds simple binary or numeric predictions on a single Salesforce object without requiring a separate dataset or CRM Analytics licensing.
Definition
Einstein Prediction Builder is the lightweight, admin-friendly predictive modeling tool inside Salesforce that builds simple binary or numeric predictions on a single Salesforce object without requiring a separate dataset or CRM Analytics licensing. The admin picks an object (Account, Case, Opportunity), picks a field to predict (Will this case escalate? Will this opportunity close?), and Einstein Prediction Builder builds a model on the org's historical data automatically. The deployed prediction can be referenced from Flow, Apex, and Lightning record pages.
Prediction Builder is intentionally simpler than Einstein Discovery. It works on a single object (no joining datasets), supports binary classification (Yes or No) and numeric regression (a number), and uses a guided UI without exposing model internals. The trade-off is reduced flexibility: complex feature engineering, cross-object joins, and bias-aware variable exclusion all require Einstein Discovery instead. Prediction Builder is the right starting point for "will this case escalate" or "will this opportunity convert" questions where the data is on a single object and the admin wants a deployed prediction without learning AutoML concepts.
How Einstein Prediction Builder builds admin-friendly predictive models
Prediction Builder vs Einstein Discovery: when to use which
Prediction Builder and Einstein Discovery are two sibling AI tools in the Salesforce platform. Prediction Builder is single-object, binary or numeric, admin-friendly, included with Service Cloud Enterprise or Sales Cloud Enterprise. Einstein Discovery is multi-dataset, supports complex feature engineering, requires CRM Analytics licensing, and is data-science-aware. Pick Prediction Builder when the question is simple and the data lives on one object. Pick Einstein Discovery when the question requires joining data across objects, custom variable engineering, or explicit bias control.
How Prediction Builder builds a model
The admin opens Setup, Einstein Prediction Builder, New. The wizard asks: which object? (Case, Opportunity, custom object). Which field to predict? (a checkbox, picklist, or number field). What examples count? (closed cases with Resolution = Refund). What examples should be excluded? (test cases, internal demo records). After the configuration, Prediction Builder builds the model in the background, scoring it on historical data. The output is a prediction score (a probability for binary, a number for regression) that the admin can review before deployment.
Model quality scoring
Each Prediction Builder model gets a quality score: Insufficient Data, Acceptable, Good, Excellent. The score reflects model accuracy on held-out validation data. Excellent models predict the right outcome more than 80 percent of the time. Acceptable models are useful but not high-confidence. Insufficient Data means the org does not have enough historical examples to build a reliable model. The score is shown in the wizard before deployment; admins can choose to deploy or to refine the model definition based on the score.
Deployment and referencing predictions
After deployment, the prediction becomes available as a Prediction Definition (the same object type used by Einstein Discovery). Flow can call it through Get Predicted Values from Prediction Definition. Apex can call it through sObject.predict(). Lightning App Builder can surface the predicted score on the record page through the Einstein Predictions component. The integration paths are identical between Prediction Builder and Einstein Discovery, even though the model building experiences differ.
Top reasons: explainability built in
Like Einstein Discovery, Prediction Builder exposes Top Reasons: the variables most strongly correlated with the predicted outcome. A case predicted to escalate might surface Top Reasons like "Priority = High", "Case Origin = Phone", "Subject contains 'urgent'". The explanations are simpler than Discovery's full Insights but cover the same conceptual ground: which variables drive the prediction. Users see the Top Reasons inline with the predicted score on the record page, building trust in the model.
Refresh cadence and model retraining
Prediction Builder models retrain automatically every 7 days by default. The retraining uses the latest historical data, capturing changes in customer behavior, sales cycles, and case patterns. Admins can adjust the cadence or trigger a manual retrain. The automatic refresh is one of the operational advantages of Prediction Builder over a hand-built model: the admin sets up the prediction once and the platform keeps it current without ongoing data science intervention.
Use cases: where Prediction Builder shines
Prediction Builder fits a specific set of use cases. Will this case escalate (binary on Case)? Will this opportunity close in this quarter (binary on Opportunity)? What will the case resolution time be (numeric on Case)? What revenue will this opportunity bring (numeric on Opportunity)? Will this lead convert to an opportunity (binary on Lead)? All single-object, all answerable with the existing field set, all admin-buildable. For multi-object questions (will this customer churn given their account, contact, and case history?), use Einstein Discovery.
Building a prediction with Einstein Prediction Builder
Building a Prediction Builder model takes 15 to 30 minutes for the configuration plus a few hours for the platform to train and score the model. The output is a deployable prediction admins can reference from Flow, Apex, or Lightning page.
- Confirm licensing
Einstein Prediction Builder is included with Service Cloud or Sales Cloud Enterprise. Confirm the org has the relevant license.
- Open Einstein Prediction Builder
Setup, Quick Find Einstein Prediction Builder, click the link. The page lists existing predictions. Click New.
- Pick the object and prediction field
Choose the object (Case, Opportunity, custom object). Choose the field to predict (a checkbox, picklist value, or numeric field).
- Define the example set
Configure which records count as positive examples (Closed Cases with Resolution = Refund) and which should be excluded (test cases).
- Set predictors
Pick which fields the model can use as predictors. Exclude leak variables and any field that should not influence the prediction.
- Build and review the model
Click Build. Wait for the platform to train and score the model. Review the quality score: Insufficient, Acceptable, Good, Excellent.
- Deploy the model
Click Deploy. The prediction becomes a Prediction Definition referenceable from Flow, Apex, and Lightning App Builder.
- Surface the prediction on the record page
Lightning App Builder, edit the record page, drag the Einstein Predictions component, configure the Prediction Definition reference.
The checkbox, picklist, or numeric field the model predicts. Binary or regression.
The historical records used as positive examples for training the model.
The input fields the model uses to predict the target. Excluded fields are not considered.
Platform-assigned model quality rating: Insufficient Data, Acceptable, Good, Excellent.
Simple explanation of which variables drive each prediction, shown inline on records.
Automatic retraining schedule, default 7 days. Adjustable per prediction.
- Prediction Builder is single-object only. Multi-object joins require Einstein Discovery instead.
- Insufficient Data means the org does not have enough historical examples. The platform needs hundreds to thousands of records of each outcome class for a useful model.
- Leak variables (fields only known after the outcome) produce great-looking models that fail in production. Audit the predictor list before deployment.
- Bias concerns apply. Fields that proxy for protected characteristics should be excluded unless explicitly justified.
- The 7-day refresh cadence is automatic. Models retrain whether the admin wants them to or not. Adjust the cadence for predictions where stability matters more than freshness.
Trust & references
Straight from the source - Salesforce's reference material on Einstein Prediction Builder.
- Einstein Prediction BuilderSalesforce Help
- Create a PredictionSalesforce Help
- Deploy a PredictionSalesforce Help
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What does Einstein Prediction Builder enable?
Q2. Who is Prediction Builder aimed at?
Q3. What's a common use case?
Discussion
Loading discussion…