REPORTING TIME WORKSHEET
A recording aid for https://meenozen.com/writing/measure-ai-reporting-time

Use one row per report. Start with a baseline and an AI-assisted report of comparable scope. Repeat across cycles. The online tool summarises two reports at a time. The blank CSV contains two empty rows and no invented measurements.

FIELD GUIDE
comparison_scope_and_acceptance_standard: Specify data sources, coverage, analysis, audience and acceptance conditions. Use the same definitions across the comparison.
comparability_confirmed: Record yes only after checking scope, quality standard, timing boundaries, stage definitions and exceptional work. Otherwise leave blank or record unconfirmed.
report: A non-confidential label and reporting period.
workflow_version: Describe the baseline method or AI-assisted method and relevant version changes.
Stage person-minutes: Sum active work across all contributors, including prompting and supervision. Start with an activity log containing role, activity, minutes and timestamps. Assign each activity once. A later repeat check is new work, so record that additional time.
  data_preparation: Source retrieval, cleaning and assembling inputs.
  drafting_and_analysis: Analysis and preparing the initial narrative, including prompts.
  source_and_interpretation_checks: Checking figures, sources and interpretations, including checks after revision.
  corrections: Fixing identified errors; keep checking those fixes in the checks stage.
  stakeholder_revisions: Requested changes other than error correction.
  approval_and_release_preparation: Final approval work, formatting and delivery preparation.
recurring_person_minutes_total: Sum all six stages only when all six are recorded. Leave blank if any stage is missing. Do not add setup or elapsed time. The online export calculates this field; the blank CSV has no formulas.
setup_and_training_person_minutes_separate: Record relevant one-off setup, training and configuration outside recurring stage totals. Explain what is included in notes. Review this effort before expanding a workflow.
elapsed_wall_clock_minutes: Measure once from an agreed start event to final acceptance. Record the actual start and end timestamps and timezone in the notes. Include waiting, unattended processing and overnight periods consistently. Never calculate this by adding people's work or stage durations.
acceptance: accepted, pending or changed. Accepted means the report meets the agreed standard. Pending includes a draft needing corrections. Changed means the scope or quality standard changed and a like-for-like claim needs reconsideration.
corrections_waiting_missing_data_and_exceptions: Record first-review acceptance or rejection, material corrections, unresolved source issues, waiting reasons, missing logs, unusual complexity, timing boundaries and other differences that could explain the result.

MISSING AND INVALID VALUES
Blank means unknown or unrecorded. Zero means measured and none required. Use non-negative minutes with a full stop for decimals. Do not use thousands separators or units in time cells. Do not fill missing measurements with zero. The online tool blocks export for invalid time entries; it allows incomplete records. Spreadsheet software does not enforce these checks automatically.

OVERLAPPING WORK WALKTHROUGH
An analyst checks sources while an editor revises the narrative. Record both people's active minutes in their respective stages. Their work contributes to combined human effort even while it overlaps. The report's turnaround is one continuous interval from the agreed start to acceptance. Do not add the two people's durations to calculate that interval. One person multitasking must allocate their actual active minutes rather than count the same minute twice.

REJECTED REPORT WALKTHROUGH
A draft contains unsupported conclusions. Keep acceptance pending and record the first-review rejection. Add the corrections to the corrections stage and the further checks to the checks stage. Continue measuring turnaround until the report meets the agreed standard. If the standard or scope changes, mark changed and select a comparable report before claiming a saving. These walkthroughs explain the method; they are not measured client outcomes.

INTERPRETATION
Compare recurring person-minutes and elapsed wall-clock minutes separately. The difference is baseline minus AI-assisted. A positive difference means less recorded time in the AI-assisted report; a negative difference means more. Never add these two measures together. Do not compare incomplete totals.
If drafting falls but checking rises, inspect the sources and rework before expanding.
If human effort falls but turnaround remains similar, inspect waiting and approval delays.
If quality is unresolved, finish review before interpreting speed as an improvement.
If setup is substantial, keep it visible and assess it across an appropriate number of repeated reports without assuming future savings.
Repeat across comparable reporting cycles and consider workload, familiarity, tool changes and exceptional work. A pair of reports does not establish causation. Recovered capacity is not automatically a cash saving.

EXPORT AND PRIVACY
The online worksheet holds entries in page memory, with no save-to-server operation. Reloading or leaving clears them. Download before leaving. The CSV quotes fields and preserves blank entries. Text starting like a spreadsheet formula is prefixed with an apostrophe for safety; that apostrophe may appear in the spreadsheet. The CSV is a summary, not an underlying activity log, and contains no ROI or cost formula. Keep confidential details out of labels and notes.