Is 04/05 April 5 or May 4? Send three fake customers through first, check every field, and write the format down before the real move.
Swipe or tap Next · checked on 2026-10-04
A date like 04/05 is just numbers. Some places write the month first, so it's April 5. Others write the day first, so it's May 4. Your software picks one based on its settings. It doesn't ask you.
ISO, the international standards group, uses this exact problem as its example: 01/05/22 could mean January 5, 2022, or May 1, 2022. Its fix is a format called ISO 8601: year, then month, then day. 2027-03-12 can only mean March 12, 2027.
Here's where it bites. The same file can be read two ways, depending on settings:
So a file can be fine, get opened once "just to look," and come out different. Nobody gets an error message. That's the scary part.
Three is enough. Each one tests something different. Put TEST in every name so nobody calls them, books them or counts them. Use contact details that can't reach a real person:
No customer IDs in your systems? Skip that line.
The three fake customers (type them into your old system) · Part 1 of 2 · Copy copies all 29 lines
TEST CUSTOMER 1: which way?
First name: TEST
Last name: Dateone
Phone: (XXX) 555-0101
Email: test1@example.com
Customer ID: DEMO-0001
Appointment: 03/12/2027 9:00 AM
Should show: March 12, 2027
TEST CUSTOMER 2: one meaning
First name: TEST
Last name: Datetwo
Phone: (XXX) 555-0102
Email: test2@example.com
Customer ID: DEMO-0002The three fake customers (type them into your old system) · Part 2 of 2 · Copy copies all 29 lines
Appointment: 03/25/2027 2:30 PM
Should show: March 25, 2027
TEST CUSTOMER 3: other boxes
First name: TEST Zoë
Last name: O'Brien-Smith
Phone: (XXX) 555-0103
Email: test3@example.com
Customer ID: 00412
Appointment: 2027-03-12 9:00 AM
Should show: March 12, 2027,
ID 00412, zeros kept
XXX = your own area code.Can't add customers to the old system? Add the three as new rows in a copy of the export file instead. Or let the prompt further down build them for your exact columns.
Type them in like any new customer. That tests the whole trip: what the old system sends out, not just what the new one takes in.
Use the same export button and settings you'll use for the real move. Then don't open and save the file in Excel or Sheets on the way. Opening it is how dates get re-read and zeros get dropped. Need to look inside? Open a copy, or open it in a plain text app like Notepad on Windows.
The new system will ask which column goes where. Slow down here. Look hard at any column called Number, ID or Date. "Number" might be a customer ID, not a phone number. If the import screen lets you pick a date format, pick the one your file really uses. Write down what you picked.
Open the records in the new system. Don't stop at the import summary. "Imported 3 of 3" means three rows went in. It doesn't mean they went in right. Check each field against the "It should show" column above:
Change one thing: the mapping, the date format on the import screen, or the date setting on the export. Delete the fakes from the new system and send them through again. Repeat until all three come out right.
Fill in the format sheet below. Delete the three fake customers from both systems, or tag them TEST so no report counts them. Now do the real move, with the exact same settings.
I've connected a lot of systems. The date box always looks the most innocent. So check it first.
One page. Fill it in after the test, keep it with your notes, and hand it to whoever runs the next import. No logins on it.
Format sheet · lines 1-11 of 29. Copy all copies the whole file; read it in full on the page.
FORMAT SHEET
From: _____________ (old system)
To: _____________ (new system)
Tested on: ________ By: ________
DATES
Old system exports: ___________
New system reads: ___________
Import setting we picked: ______
Safe format: YYYY-MM-DD
Time zone, old: ______ new: ______Want a hand? Paste this into Claude or ChatGPT. It builds the fake customers for your exact columns, then checks the results with you. Only paste fake data. Never paste real customers into a chat.
The list-move test prompt · lines 1-4 of 16. Copy all copies the whole prompt; read it in full on the page.
I'm moving my customer list from [OLD SYSTEM] to [NEW SYSTEM]. Before the real move, help me run a test with three fake customers. I will only share fake data with you, never real customers.
Here are the column names from my export file (the first line only):
[PASTE THE FIRST LINE OF YOUR EXPORT FILE HERE]The AI can only compare what you paste. Opening the records in the new system and looking is still your job.
Coastal Line Roofing is moving its customer list from an old scheduling app to a new customer system. The office manager adds the three fake customers to the old app, exports, and imports. Here's what the new system shows:
Two fixes. On the import screen, the date setting was day first. She switches it to month first. Then she maps the old app's "Number" column to Customer ID instead of Phone. She deletes the fakes, sends them through again, and all three come out right.
Her format sheet now says: dates go in month first, or as YYYY-MM-DD. "Number" goes into Customer ID. Both systems set to Eastern time. Nobody shows up on the wrong Tuesday.
"3 of 3 imported" counts rows. Open the records and look at each field.
When the day and month match, it reads right either way and proves nothing. Use 03/12 to see which way it reads, and 03/25, which can only mean one thing.
That's where zeros vanish and dates get re-read. Look at a copy, or open it as plain text.
A real name in a test can get a call, a text or a reminder. Use TEST names, example.com emails and 555-0100 to 555-0199 numbers.
Next year somebody runs the same import from memory. The format sheet is the memory.
What each page backs up is listed at the end of the full guide.
Not exciting? Correct. Neither is explaining why everyone showed up on the wrong Tuesday.
Want help setting this up? Receipts Group builds these systems.
Open the full guide as a page (every file in full, plus the table of contents).