Skip to content
Let’s talk

All technology

Find where time goes in a process before deciding what to automate

We measure processes, not people. From data your systems already record, such as ticket timestamps, workflow states and system logs, we show how long each step takes, where work waits, and which steps are repetitive enough to automate.

Example: time-in-step for a claims processTicket history, workflow logs and ERP timestamps are combined into a process model. The model produces a time-in-step report, aggregated by process step, which ranks automation candidates. It all runs inside the client's environment.Your environmentTicket historyWorkflow logsERP timestampsProcess modelTime-in-step reportAutomationcandidates

When clients need this

  • Everyone agrees a process is slow, but nobody can say which step is the problem.
  • You’re about to invest in automation and want evidence about where it will pay off.
  • A process spans several systems and teams, so no single dashboard shows it end to end.

How we build it

We work from records your systems already keep: when a ticket changed state, when an order moved, when a document was approved. We don’t install tracking software on employees’ computers and we don’t record screens.

Data is aggregated by process step. Reports show how long steps take and where work waits, not how fast any individual works. We agree the scope of what is measured with you, and with employee representatives where that applies, before we start.

The output is a short list of automation candidates, each with the measured time it costs today and what we would build to change it.

SELECT step,
       percentile_cont(0.5) WITHIN GROUP (ORDER BY hours) AS median_hours,
       count(*) AS claims
FROM (
  SELECT claim_id, step,
         EXTRACT(EPOCH FROM left_at - entered_at) / 3600 AS hours
  FROM claim_step_events
) AS t
GROUP BY step
ORDER BY median_hours DESC;
A typical first query: median hours each claim spends in each workflow step.

What we don’t do

  • We don’t measure, score or rank individual employees.
  • We don’t use keystroke logging, screen recording or emotion recognition.
  • We don’t collect more data than the process question needs.