Process Intelligence from Video
Process mining reconstructs a process from the events a system happened to log. Most operational work never reaches a log — it happens on a factory floor, across a dozen applications, or in the gaps between systems. Video is the only record of that work, and process intelligence is what turns it into something you can measure.
The work that leaves no event log
A process-mining tool builds its picture from timestamps a system wrote down. That works well when a case moves through one platform and every transition is a database row. It breaks in three ordinary situations: work performed by hand at a bench or on a line, work split across applications that share no identifier, and the gaps between recorded events — the waiting, the re-checking, the switching back and forth — where a great deal of cycle time actually goes. In each case the log is silent and the recording is not.
Where the two approaches apply
| Situation | Event log exists? | Best suited |
|---|---|---|
| Order-to-cash inside one ERP | Yes, complete | Process mining |
| Assembly or inspection at a station | No | Video process intelligence |
| Quoting across ERP, spreadsheets and PDFs | Partial, no shared case id | Video process intelligence |
| Ticket lifecycle in a service desk | Yes | Process mining |
| Time inside a step — why one case took longer | Rarely | Video process intelligence |
What a study produces
Each analysed session returns a timeline of labelled activities with a measured duration against every one, exportable as CSV. For screen-recorded work it also reports which applications the time went to and how long the operator held focus before switching. Where a documented method exists, the session can be compared against it and the points of divergence listed. Running the same process twice and comparing the two timelines is what turns a single observation into a baseline.
Reading a bottleneck honestly
The longest activity in a timeline is not automatically the constraint. A step can be long because it absorbs a delay created upstream, because it batches several cases at once, or because it is genuinely the slowest operation. Process intelligence narrows the question by showing exactly where duration accumulates and letting you replay the moment it happened; the diagnosis stays with the engineer. This is the same discipline as any time study — the measurement is objective, the interpretation is professional judgement.
Where it sits next to a formal time study
A predetermined motion time system such as MTM or MOST derives a standard time from a described method, before the work is observed. A stopwatch study measures a live operator and applies a performance rating. Video process intelligence sits between them: it measures actual elapsed durations across whole sessions without an observer present, and applies no performance rating, which makes it well suited to establishing a baseline and monitoring drift, and less suited to setting a normalised standard time on its own. See work measurement for how the techniques compare.