What is Predictive Intelligence in ServiceNow?
A platform function that uses machine learning to power workflows by making predictions and recommendations
What does Predictive Intelligence learn from?
Contextual data and patterns to predict outcomes without manual intervention
What types of models does Predictive Intelligence support?
Classification, Similarity, Clustering, Regression
Name three problems Predictive Intelligence helps solve.
Intelligent routing, triage elimination, backlog reduction.
How does it improve service desk operations?
By predicting fields, identifying knowledge gaps, and reducing MTTR
What predictive capabilities does it offer?
Risk, SLA, resolution time, and similar ticket/article identification
What makes an organization ideal for Predictive Intelligence?
Over 10K records, manual triage, high ticket volume, frequent reassignment, desire for efficiency
What are signs it may not be a good fit?
Class imbalance, lack of data, platform customization, mixed languages, no dedicated resources
What are signs it may not be a good fit?
No missing info, consistent values, balanced distribution, 10–30K records, excludes abnormal periods.
Why is seasonality important in training data?
To ensure predictions reflect normal business behavior and avoid skewed models
What defines a good classification model?
High precision and coverage, well-distributed value sets, strong business process.
What indicates a poor model?
Outliers, uneven distribution, inconsistent precision and coverage.
Can Predictive Intelligence exclude certain classes during training?
Yes
What field types are recommended for model input?
Choice, String, and Reference fields
What is a word corpus used for?
To help the model interpret training and prediction data
What are stop-words in model training?
Words ignored during training to improve accuracy
Can Predictive Intelligence integrate with third-party apps?
No, only with ServiceNow applications like ITSM, HRSD, CSM, FSM, and PPM.
Why follow design recommendations for Predictive Intelligence?
To avoid missed opportunities and ensure long-term model performance.
“SMART DATA POWERS FAST PREDICTIONS”
🔠 Mnemonic Breakdown
- S – Service Desk Efficiency: Eliminates manual triage and reduces ticket backlog
- M – Machine Learning Models: Classification, Similarity, Clustering, Regression
- A – Automation Opportunities: Identifies patterns and gaps for workflow optimization
- R – Risk & SLA Prediction: Forecasts resolution times and service levels
- T – Training Data Quality: Requires clean, balanced, and representative datasets
- D – Distribution Matters: Good value spread ensures model accuracy
- A – Assignment Rules: Predicts fields like assignment groups and categories
- T – Triage Elimination: Speeds up routing and classification
- A – Article Matching: Finds similar tickets and knowledge articles
- P – Platform Integration: Works with ServiceNow apps (not third-party)
- F – Fit Criteria: High ticket volume, manual triage, desire for efficiency
- P – Precision & Coverage: Key indicators of model reliability
What does Task Intelligence predict at case creation?
Language and sentiment.
What types of cases can Task Intelligence detect early?
Duplicate and spam cases.
How does Task Intelligence support self-solve?
By auto-responding with knowledge articles or recommending resolutions.
What happens when a case needs agent involvement?
It categorizes the case and extracts document values (via DocIntel) to populate fields
