You can give your accountant codes in two ways: a separate mapping table (category to code, once) or a code written on every transaction row. The mapping table is shorter and easier to maintain; row-level codes are more precise but take longer. Most small freelancers do best with a mapping table plus exceptions.
What is the difference in practice?
| Aspect | Mapping table | Code on every row |
|---|---|---|
| Size | One line per category | One code per transaction |
| Upkeep | Change one cell to change a category | Change every affected row |
| Precision | Same code for the same category | Can differ row by row |
| Best when | Categories behave consistently | Same category needs different treatment |
A graphic designer in Lyon who buys software, printing and travel will find that nearly everything in "Software" is treated alike. A mapping table is perfect. A consultant whose "Meals" category includes both client dinners and lunches alone might need a row-level flag for some of them.
How do I build the mapping table?
- List each category once, with a short note on what it covers.
- Ask your accountant for the code or ledger heading for each.
- Put the answers beside each category.
- In your transaction sheet, add a code column with =VLOOKUP(D2, $K$2:$L$40, 2, FALSE), where D is the category and K:L is the mapping table. Wrap it in IFERROR(..., "CHECK") so unmapped categories are obvious.
Then count the CHECK rows with =COUNTIF($E$2:$E$1500, "CHECK"). It should reach zero before you send anything.
When do I need row-level codes?
Only for exceptions. Suppose 120 transactions in "Meals" total €3,420.00 and 15 of them, totalling €512.50, are client meals that your accountant wants flagged differently. Add a column named Flag and mark just those 15. The remaining 105 rows, worth 3,420.00 - 512.50 = €2,907.50, follow the category default. You have not coded 120 rows; you have coded 15. Figures are invented for illustration.
How should I handle categories that change mid-year?
Say you split "Office" into "Rent" and "Supplies" in June. Do not rename history silently. Add a date column to the mapping table, or send two versions, noting the change date. Your accountant can then see that January to May used one scheme and June onward another, instead of wondering why totals jumped.
Can an app export the mapping for me?
If you use SpendixAI, the export centre has a categories and tax codes file, and categories can carry an accounting code as well as tax-related settings. It is built around India and INR, and the tax settings reflect Indian GST, so it is not made for a European practice. What matters for you is the idea: export the category setup, not only the transactions.
Testing the mapping with a small sample
Before trusting the table, test it on twenty random transactions. Look up each category by hand, confirm the code matches the mapping and check that nothing returns CHECK. If three of twenty fail, a spelling variant is probably hiding in the data, perhaps a trailing space or a plural. Fix those and test another twenty. This takes ten minutes and finds the same problems that would otherwise surface during your accountant's review, when they are slower and more awkward to correct.
Before you send
- Every category in the transaction file appears once in the mapping table.
- Unmapped rows are zero.
- Exceptions are flagged and listed with a one-line reason.
- The file has a date and a version label.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

