Use Lead Scoring → Rescoring to inspect scoring jobs and evaluate whether scores align with actual outcomes. Scoring operations require authorized workspace access; running them can consume credits or change scores. Begin by reviewing existing jobs and the published policy rather than starting another job automatically.

What you’ll see

The Rescoring page, where you rescore a group of leads or roll back a change. Rescoring uses one AI credit per lead.

Review and run an authorized operation

  1. Open Logs & rescore. Refresh loads policy versions; Refresh jobs loads recent and running jobs. Inspect each job’s scope, policy version, status, Processed, Updated, Unchanged, and Failed counts.
  2. For a bounded operation, use Create rescore job. Choose Scope and the matching Filter value. Review Dry run and Include research before Queue rescore. A dry-run job still invokes server work; do not assume it is a cost-free local preview.
  3. Rescore all leads is a separate full-workspace action. Its confirmation states Yes, rescore & charge credits and the source checks payment readiness. Proceed only with explicit approval for that scope and cost.
  4. Process page advances a nonterminal job; it is not a status refresh. Inspect current progress before using it.
  5. In Calibration & outcomes, use Refresh to inspect score bands and suggestions. Record outcome writes an actual lead outcome with score/version context. Compare versions compares champion and challenger versions.

Confirm and recover

Queued is not completed. Recheck job status and failed counts before claiming a rescore finished. Policy publication and rescore queueing can have different outcomes. The visible Versions list is not proof of a rollback button or a completed rollback workflow. If a job fails or its state is unclear, inspect its existing record and ask support instead of duplicating the operation.

Related: Dashboard, Leads.

Screenshots

The screenshot for this guide is shown above, under “What you’ll see”.