01 / Section
Choose an outcome as the unit
When operations leaders want to understand how work happens, start with a business outcome: for example, a customer request resolved or an invoice processed. That gives the review a clear question: what steps and decisions contributed to this result?
An app or website visit is a weaker starting point. The same application can support many different tasks, and a task can cross several tools. Activity alone does not show whether a business outcome was reached or how.
02 / Section
Reconstruct the attempt from evidence
Shadow reconstructs evidence-linked attempts at business outcomes from captured batches. Reviewing an attempt as a sequence helps a team connect the outcome under discussion to the recorded material that may explain how it was reached.
Keep three things distinct: what the captured evidence records, what reviewers interpret it to mean, and what they assume or plan to test next. A citation helps a reader trace a claim back to its source. It proves provenance, not that the interpretation is certain or complete.
03 / Section
Compare attempts, not isolated clicks
One attempt can reveal a possible workflow, but it cannot establish a recurring pattern. Compare distinct attempts aimed at similar outcomes. Look for steps that recur, decisions that differ, and handoffs or exceptions worth examining.
Treat a pattern as a finding to validate, not a rule that applies to every case. Captured batches can be partial, and different attempts may reflect different circumstances. Record the evidence behind a pattern and the questions that remain open.
04 / Section
Use app and site sequences as discovery signals
App and site sequences can help suggest where a workflow may begin, which tools may be involved, or what to investigate next. Keep those sequences separate from reconstructed task attempts: they are discovery signals, not proof that a specific task was completed.
A practical review can begin with a defined outcome, inspect its evidence-linked attempts, compare several distinct attempts, and then use app or site sequences to guide further questions. This keeps the discussion focused on how work may be done without treating activity as a measure of people or as a complete account of the process.
