Einstein Discovery12% of the exam

Model card, Model Manager monitoring and refresh

Documenting a deployed prediction with a model card, monitoring accuracy and live data in Model Manager, setting Model Quality Notifications, and keeping a drifting model current with automatic refresh and rescoring.

Official exam objective
Monitor and interpret a Model Card to improve or maintain model performance.

~6 min read · 6 practice questions ·

What the exam expects

  • Configure a model card for a prediction definition and expose it on the record page.
  • Know which sections a model card holds and which it fills in automatically.
  • Set up performance monitoring so Model Manager compares predicted with actual outcomes.
  • Configure Model Quality Notifications and interpret live data alerts for missing and out-of-bounds values.
  • Counter drift with automatic model refresh, rescoring and a new model version.

Key terms: Model Manager (the Studio tool that views, configures and manages deployed prediction definitions and models); model card (documentation of a prediction's training statistics, intended use and limitations); terminal state (the condition after which a record's actual outcome is final); drift (a deployed model becoming less accurate as the business, data or requirements change); out-of-bounds value (a value not found in the training data).

Key facts

Model card required field
Detail
Title; empty sections are hidden
Training Metrics section
Detail
Shown by Einstein; per selected class for multiclass
Sharing a card offline
Detail
View Model, then Download as PDF
Monitoring prerequisite
Detail
Automated prediction writeback
Terminal state conditions
Detail
Equals or Not Equals, case-sensitive
Evaluation frequency
Detail
Weekly or Monthly
New version deployed
Detail
Statistics reset; earlier history cleared
Out-of-bounds value
Detail
Not found in the training data
Refresh frequency
Detail
Weekly or Monthly, or Refresh Now
Deployment Interruption Threshold
Detail
No auto-deploy above the out-of-bounds percentage
Sources for this section (4)
  • Why might a refresh job's history show runs that didn't happen?
  • When are a prediction definition's performance statistics reset?
  • Which statistics can Replace Missing Values use for a numeric variable?
  • What does orange show in the Live Data Alerts bar charts?
  • When a definition holds several models, how do you see another model's card details?

The questions are free to read. Answers with their sources and self-grading are included with the pack.

Keep reading with the CRM Analytics and Einstein Discovery Consultant pack

  • Core concepts
  • Exam traps
  • Worked scenario

The pack includes:

  • All 26 lessons of the CRM Analytics and Einstein Discovery Consultant guide, with reading progress
  • Interactive decision walks and step-by-step sequence steppers
  • Practice exams purchase adds inline bank questions and topic practice (6 questions for this lesson)
  • An official source cited for every claim
  • An AI tutor that answers from the guide
  • Choose guide-only, practice-only, or save $4 with both

Practice for the CRM Analytics and Einstein Discovery Consultant exam