← All articles

Process discovery

Why a Completed Task Is a Better Unit for Process Discovery Than App Activity

An app sequence shows where work appeared. A task-shaped attempt connects evidence to an intended business outcome, making it a more useful starting point for understanding how work happens.

Text-free Shadow evidence-map composition in ink black, warm cream, mustard gold, and muted sage, with connected document-like shapes suggesting a work outcome and its evidence.

Apps show activity, not the work’s purpose

A visit to a spreadsheet, a browser tab, or a messaging tool is an event. By itself, it does not explain what someone was trying to accomplish, whether the activity belonged to one task, or what happened next.

That makes app activity a weak unit for process discovery. The same sequence of apps can support different business outcomes, while one outcome may involve several tools. App and site sequences are useful discovery signals, but they are not a complete account of a workflow.

Start with an outcome and its evidence

A more useful unit is a task-shaped attempt: a bounded effort to reach a business outcome, reconstructed from a captured batch and linked to the evidence available for that attempt. For example, the outcome might be preparing a client update, resolving an invoice question, or reviewing a request.

Treat an attempt as completed only when the recorded evidence supports that interpretation. A saved document, sent message, or changed record may support it, but no single trace should be asked to prove more than it shows. A citation establishes where an observation came from; it does not make the interpretation certain.

Compare distinct attempts, not isolated clicks

One task attempt can show a plausible sequence, but it cannot establish a recurring process. Compare distinct attempts at the same kind of outcome to see which steps, handoffs, or tool changes appear more than once and where the evidence differs.

Keep the comparison grounded in what was captured. A repeated pattern is a candidate workflow pattern, not proof that every case follows it. Differences may reflect the request, the available information, or gaps in the captured evidence.

Keep observations, interpretations, and plans separate

In a review, label what the evidence records separately from what you infer. For instance: the record shows a message and a revised file; the interpretation is that both may belong to the same client update; the planning assumption is that a shared template could reduce repeated drafting.

This separation helps operations leaders decide what to validate next. Use app and site sequences to find possible areas to investigate, then examine evidence-linked task attempts to understand the work. Do not treat either view as a measure of working time or individual productivity, or as proof that a proposed change will produce a particular result.