# Workflow time investigation Follow one bounded kind of work from entry to verified completion. This worksheet contains completed hypothetical teaching records and a blank working copy. It provides no customer measurement, saving, adoption result, or release approval. ## Completed hypothetical boundary | Field | Entry | | ----------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------- | | Work | Domestic invoice price-variance cases | | Start | Case enters the agreed queue | | Finish | Authorized posting is confirmed by ledger readback | | Declared units | Handling minutes and elapsed minutes; all four cases enter and complete in the same illustrative window | | Handling | Sum of declared active work, including reopening and corrections | | Elapsed | Declared duration from entry to verified completion | | Approval | Required forward-path review; distinct from posting and readback | | Evidence basis | Hypothetical timing table v1, not a timestamped source-system trace | | Current state | Teaching baseline supplied; proposed changes unimplemented and unmeasured | | Owner roles to confirm before a real test | Purchasing source owner, accounts-payable queue/service owner, finance conflict owner, designated reviewer, finance controls owner | ## Completed case table | Case | Return reason | Return events | Handling minutes | Elapsed minutes | Elapsed minus handling | | ----- | ---------------------------------------------------- | ------------: | ---------------: | --------------: | ---------------------: | | RX-1 | None; evidence complete | 0 | 8 | 20 | 12 | | RX-2 | Purchase-order revision missing; two repeat requests | 2 | 14 | 180 | 166 | | RX-3 | Invoice and purchase-order amounts conflict | 1 | 12 | 75 | 63 | | RX-4 | None; required approval is a forward handoff | 0 | 10 | 30 | 20 | | Total | Four entered and completed cases | 3 | 44 | 305 | 261 | Checked calculations over the same four cases: - Returned cases: RX-2 and RX-3; `2 / 4 = 50%`. - Return events per entered case: `3 / 4 = 0.75`. - First-pass completions without a return: RX-1 and RX-4; `2 / 4 = 50%`. - Handling: `8 + 14 + 12 + 10 = 44 minutes`. - Sum of elapsed durations: `20 + 180 + 75 + 30 = 305 case-minutes`. - Sum of differences: `12 + 166 + 63 + 20 = 261 case-minutes`. First-pass completion does not mean processing without a person. Approval remains required. The 261 case-minutes are neither staff hours available to save nor a shared workday's duration. Overlapping cases and parallel work need separate measurement. The difference contains possible waiting or unobserved work; it does not establish its cause. ## RX-2 investigation record | Question | Current answer | Evidence still needed | | -------------------------------------------- | -------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | | What active work occurred? | 14 handling minutes are declared in the teaching table | Permitted observation or a checked handling record for each real activity | | What elapsed duration occurred? | 180 minutes are declared | Entry and verified-completion events on an agreed clock | | Which reason was recorded? | Missing purchase-order revision; two repeat requests | Request/return records and the source owner's confirmation | | What explains the 166-minute difference? | Unknown; no interval breakdown is supplied | Queue events, approved correspondence, paired walkthrough, and any remaining unknown intervals | | Does the source reason explain every minute? | Not established | Separate request latency, queueing, other activity, and gaps lacking evidence | | Who can permit a source repair? | Purchasing source owner and AP service owner must decide | Purchasing confirms which purchase-order record applies, who may read it, and how current it must be. The AP service owner agrees which cases may enter a test. | ## Proposed redesign, not tested | Current work | Proposed disposition | Protection retained | Owner and test | | ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------- | | Reviewer discovers a missing revision | Repair presence and freshness checks at entry | No guessed revision; missing or stale evidence stays held | Purchasing source owner and AP service owner; test missing, stale, and current evidence | | Repeated requests and duplicate status entry | Combine tracking in one owned queue; remove duplicate entry only if approved | Assignment, reminders, history, and escalation | Queue owner; demonstrate that held work keeps its owner and escalation | | Conflicting invoice and purchase-order amounts | Finance still decides how to resolve the mismatch; compare evidence retrieval or optional drafting separately | Check arithmetic and compare source amounts; a plausible explanation cannot overrule a mismatch | Finance owner confirms who may decide and what evidence is needed; test both an unresolved conflict and a properly approved correction | | Approval, posting, and ledger readback | The designated reviewer still approves and posts under current permissions; check the ledger separately | Do not mark complete until required approval and confirmation of the recorded change are present | Reviewer and controls owner; test stale approval, failed posting, and missing readback | The clock still starts at the original queue entry. A new holding queue cannot hide difficult cases by restarting their clock after source repair. The safe current path remains available. Source and service owners have not approved or run the proposed test in this exercise. ## Your working copy ### Start with these questions 1. Which event starts a case's clock, and what independently checked event ends it? 2. What active work can you observe, including correction and reopening? 3. Which intervals show a queue, a return, or a necessary control, and which remain unknown? 4. What smaller source or queue repair could you test without removing required approval? ### Record the timing evidence - Which workflow and version are you investigating, who owns operation, and what is the evidence cutoff? - Which cases qualify, and which are excluded? - Which start and finish events apply, and what independent evidence confirms completion? - Will you use calendar time or business time, in which unit and timezone? Do source clocks differ? - Which measurement window applies, and how will pending, rejected, held, and late-completing cases remain visible? - How were cases sampled, how many are included, and what can that sample not establish? - Which access, evidence destination, redaction, and retention rules are approved? | Case / interval | Start and end source events | Active work and its method | Elapsed gap | Queue, rework, necessary control, or unknown | Source / revision / limitation | Owner and next question | | --------------- | --------------------------- | -------------------------- | ----------- | -------------------------------------------- | ------------------------------ | ----------------------- | | | | | | | | | Separate person-minutes of effort from case-clock active intervals when people work in parallel. Do not subtract summed simultaneous effort from elapsed duration and label the answer waiting time. | Return definition | Numerator | Denominator | Source and revision | Treatment of repeats and forward approvals | | ------------------------------ | --------- | ----------- | ------------------- | ------------------------------------------ | | Cases with at least one return | | | | | | Return events per entered case | | | | | | First-pass completions | | | | | - Which intervals remain unknown or disputed, which evidence would resolve them, and who will obtain it? - Why does the current step exist? Should it be kept, repaired, combined, or removed, and which alternatives will you compare? - Which necessary approval, source check, escalation, or recovery control must remain? - Which simpler source, queue, or rule repair will you compare before proposing model work? - What test is proposed, for which cases and comparison path? Keep its start and finish unchanged. - How will you measure reviewer and source-owner effort, errors, backlog, support, and full cost? - Who may decide, which approval is still needed, what safe fallback remains, and what would stop the test? - What is the next action, who is accountable, when is it due, and when will it be reviewed? Keep proposals and measured results in separate revisions. A filled timing sheet does not authorize removing a control or releasing a system.