← All articles

Operations

Use the Completed Task, Not App Activity, to Discover a Process

An app sequence shows where work appeared. A task with an evidence-linked outcome gives operations leaders a more useful unit for comparing how work gets done.

Text-free Shadow evidence-map composition of connected document and workflow cards on an ink-black field, with warm cream, mustard-gold, and muted-sage accents.

App activity is a clue, not a process

A sequence of sites and apps can help surface where a workflow might be happening. But opening a spreadsheet, switching to email, and visiting a business system does not, by itself, explain what the person was trying to accomplish or whether the work reached an outcome.

Treat those sequences as discovery signals: useful prompts for finding a workflow to examine, not proof of a process or a measure of work. Starting with app activity can make unrelated actions look like one procedure simply because they happened in the same tools.

Anchor the investigation in an outcome

A task with a recorded outcome is a more useful unit for process discovery. It gives the investigation a business question to follow—such as whether a request was resolved or a report was prepared—and a boundary for gathering relevant evidence.

Shadow reconstructs evidence-linked attempts from captured batches. That makes it possible to examine the steps and materials associated with a particular attempt, rather than treating all activity in an app as part of the same workflow. The boundary matters: include evidence that bears on the outcome, and avoid claiming more than the record supports.

Compare distinct attempts to find the recurring pattern

One attempt can show a plausible path, but it cannot tell you which steps are typical. Compare distinct attempts at the same kind of business outcome. Look for actions or handoffs that recur, places where the sequence differs, and points where evidence is missing.

A repeated pattern is a candidate workflow to review, not automatic proof that every team or case follows it. Keep unusual cases visible: they may reflect different requirements, exceptions, or simply incomplete evidence. The comparison should help leaders ask better questions, not flatten every attempt into a single idealized path.

Keep evidence, interpretation, and plans separate

For each finding, distinguish what the captured evidence shows from what you infer. A citation establishes where a detail came from; it does not make that detail certain or prove that the interpretation is correct.

Then label proposed changes as planning assumptions. A suggested handoff, checklist, or process adjustment is a decision to evaluate, not a step already present in the evidence. Keeping these layers separate lets operations leaders validate the recurring pattern with the people who understand the work and decide what to investigate or change next.