CSV date handling
Which CSV Date Should You Use? Order, Payment, Refund, Fulfilment, and Payout Dates Explained
Use the date that matches the business question, not whichever date column appears first. Use order date for demand, payment or capture date for collected-payment activity, fulfilment date for operational delivery, refund date for return activity, and payout or bank-arrival date for cash settlement. Keep the other dates, document the timezone and period boundaries, and label the resulting metric clearly. When comparing reports, align their date basis before comparing totals; otherwise both reports can be correct and still disagree.
Short answer
Use the date that matches the business question, not whichever date column appears first. Use order date for demand, payment or capture date for collected-payment activity, fulfilment date for operational delivery, refund date for return activity, and payout or bank-arrival date for cash settlement. Keep the other dates, document the timezone and period boundaries, and label the resulting metric clearly. When comparing reports, align their date basis before comparing totals; otherwise both reports can be correct and still disagree.
Upload
Bring the CSV you already use.
Understand
Start with the brief and evidence.
Act
Ask, edit, filter, and export.
A practical starting point
Use the date that matches the business question, not whichever date column appears first. Use order date for demand, payment or capture date for collected-payment activity, fulfilment date for operational delivery, refund date for return activity, and payout or bank-arrival date for cash settlement. Keep the other dates, document the timezone and period boundaries, and label the resulting metric clearly. When comparing reports, align their date basis before comparing totals; otherwise both reports can be correct and still disagree.
Concrete examples
Example 1: An order and its payment fall in different months
An order is created at 11:58 p.m. on 31 August and captured at 12:03 a.m. on 1 September in the reporting timezone. It belongs in August demand by order date and September captured-payment activity by capture date. Moving both metrics into one month hides the operational delay.
Example 2: A later refund changes a different period
A RM240 order is placed and paid on 20 July, then RM60 is refunded on 4 August. A report of July orders retains RM240 of order activity, while an August refund-event report shows negative RM60. A cohort view can separately show that the August refund relates to a July order.
Example 3: Fulfilment measures workload, not demand
A retailer receives 80 orders on Friday and fulfils 50 on Friday plus 30 on Monday. The order-date view shows 80 Friday orders. The fulfilment-date view shows the warehouse workload split across two days. Neither view should be relabelled as the other.
Example 4: Payout timing does not equal payment timing
Payments captured over a weekend become part of a Tuesday payout and reach the bank later. A payment-date chart describes customer collection activity; a payout-effective or bank-posted chart describes settlement. Comparing them by calendar day without a reconciliation bridge creates a false discrepancy.
Example 5: A timezone moves a row across the daily boundary
A source timestamp of 2026-08-31T16:30:00Z is 12:30 a.m. on 1 September in Singapore. A UTC report counts it in August, while a Singapore-local report counts it in September. Both can be internally consistent, so the dashboard must name its timezone.
Examples are illustrative and are not customer results.
Common questions
What is the safest way to work with this CSV problem?
Use the date that matches the business question, not whichever date column appears first. Use order date for demand, payment or capture date for collected-payment activity, fulfilment date for operational delivery, refund date for return activity, and payout or bank-arrival date for cash settlement. Keep the other dates, document the timezone and period boundaries, and label the resulting metric clearly. When comparing reports, align their date basis before comparing totals; otherwise both reports can be correct and still disagree.
Keep exploring
See what your own CSV is saying.
Try the browser workspace with a real file and keep the public guide open when you need help.
Start free