Explain why a report and export disagree
Reconcile population, period, unit, state and rounding in order.
Compare population, period, state, unit and rounding in that order. Do not add a balancing entry until the difference has a record-level explanation.
First make the two totals comparable
Write the contributor and project filters, reporting zone, start/end boundaries, review state, granularity and export time for each total. A daily report and a period summary may include different corrections or people. Compare one contributor and one short period first. If that aligns, widen the scope methodically instead of changing several settings at once.
Use a reconciliation bridge
For a fictional source total of 480 minutes and recipient total of 450, look for identified records or transformations accounting for the 30-minute difference. Perhaps one 30-minute entry is outside the recipient’s accepted state. That explains the difference without changing the original duration. Keep unknown differences separate from explained ones; do not label the whole batch correct merely because the final numbers can be forced to match.
Check mathematical transformations after scope
A value of 7.30 decimal hours is 438 minutes, not seven hours thirty minutes. Display precision and per-row rounding can produce smaller discrepancies too. Harvest and Clockify document differing rounding/report behaviors. Inspect raw values and the transformation stage before choosing a remedy. A mismatch of a few minutes is not automatically harmless or automatically a rounding issue.
Retain the reproducible explanation
Save the two file versions, settings, affected IDs and calculation. If a product defect is suspected, send its support team a minimal fictional reproduction or appropriately redacted evidence through the authorized process. Do not upload private worker records to a public forum. If the cause remains unknown, mark the batch unresolved rather than silently offsetting the total.
Continue this work-record check
Use the linked guide for the next decision in this workflow. Keep your original records separate from experiments and record any unresolved requirement before changing a live process.
Sources and boundaries
Primary documentation supports the dated product facts. Proposed checks and fictional examples are our editorial method, not observed product results.
- Harvest — totals · checked September 23, 2026
- Clockify — round · checked September 23, 2026