Skip to main content
Log2MapLog2Map

Task Mining

Apps, screens, and the task graph

Map user, session, and action. The graph is the operator’s path through applications, not the pizza process spine.

Start Task Mining opens a Draft. Upload task-mining-sample.csv (not the pizza process file) and map:

  • User ID / Employee / Actor → user_id
  • Action / Event Name → event_type (not element)
  • Timestamp → timestamp
  • System / Application → app

Session ID is optional: map session_id when the file has it. Save project stays disabled until analysis has run.

After Run analysis, check Insights: 5 users and 42 sessions. Extra metric Raw events is 434. The task graph should show about 18 nodes named like Copy phone number · CRM, not “Open app → Choose pizza.” If you map Action to element, nodes become Phone field · CRM instead.

Left-click selects a node. Right-click opens a compact popover (Runs, Automation as a percent, Sessions, Users). That popover is not the Insights extra metric named Automation — that extra metric is a count of candidates.

Read one session, for example PIZZA-TM-001 / op_anna.

Map the operator file

Start Task Mining, upload task-mining-sample.csv, map the four required fields plus session_id, and confirm Insights shows 5 users and 42 sessions. Nodes should be applications, not pizza activities.

Start Task MiningDownload UI-event CSV

Reference in Docs