See where the AI reduces effort and where classification or replies need tuning.
You need to know whether Einstein for Service is reducing effort or creating extra work. This guide shows you how to measure classification accuracy, routing corrections, and handle time across AI-assisted and manual cases, so you can tune the features that need improvement.
Show faster, more consistent service inside Salesforce with automation metrics.
Resolve common requests end to end with a scoped AI agent that escalates the rest.
Tune which fields Einstein predicts and the data it learns from for the team's case types.
Pull case volumes, classification accuracy, and routing corrections for the period. In your reporting tool, filter cases by the analysis period. Extract total case volume, the number of cases auto-classified, classification accuracy, and the count of routing corrections.
Measure handle time on cases using suggested replies versus manual. Segment cases into two groups: those where the agent used a suggested reply, and those drafted manually. Calculate average handle time for each group.
Identify case types where the AI helps most and least. Break down classification accuracy, routing correction rate, and handle-time by case type. Look for patterns showing where the AI reduces effort and where it needs tuning.
Tune classification, routing, or replies for the weak areas. For case types with low classification accuracy, review and adjust the classification rules. For high routing correction rates, refine routing logic. For minimal handle-time improvement, review reply templates or knowledge articles.
Track the trend over time. Set up a recurring report that refreshes regularly. Compare classification accuracy, routing correction rate, and handle-time period-over-period.
Period: [analysis period] Total cases: [volume] Einstein auto-classified: [count] Classification accuracy: [correct] / [total] Routing corrections: [count] Handle time (AI-assisted replies): [duration] average Handle time (manual drafts): [duration] average Case type breakdown: - [Type A]: [accuracy rate], [correction rate] - [Type B]: [accuracy rate], [correction rate] Action: Tune classification rules for [Type B] cases; review routing logic for [Type B] queue. Next review: [date]
For the period, show classification accuracy, the rate of manual routing corrections, and handle time on AI-assisted versus manual cases, by case type, to find where to tune.
Get a complete, on-brand draft grounded in the team's knowledge to review and send.
Send each case to the correct queue automatically, removing manual redirection.