Credit card statement to CSV: choose the source and check the cycle
Updated · 5 min read
Corrected October 12, 2026: removed invented issuer formats and unsupported extraction and accounting promises.
The desktop apps export date, description, and amount as CSV. Open CSV in Excel to save an XLSX workbook. Statement support varies: verify the result against your own PDF.
Select the right record and cycle
Start with the company issuing your card and the applicable country and card product. A payment-network logo does not identify the issuer’s download menus.
| Your task | Source to check |
|---|---|
| Review the bill, payment due or interest calculation | The issued billing statement and its explanatory sections |
| Categorise posted spending for a defined period | A suitable native transaction file, or reviewed rows extracted from the PDF |
| Investigate a missing purchase | Transaction and posting information for the relevant cycles; pending activity separately |
Retrieval examples, with their own product limits: Amex US and U.S. Bank.
Extract transaction rows and preserve bill context
LocalExtract processes one PDF locally into date,description,amount CSV. Downloads and limits.
The three exported fields are not a complete bill: they do not separately represent payment due, opening or closing balance, credit limit, rewards, cardholder, currency or a full accounting entry. Read those details from the original statement and record verified context separately.
Keep the original bill, then:
- Check all transaction pages and any additional-cardholder sections.
- Exclude statement balances, subtotals and repeated page headings from transaction rows.
- Where two dates appear, establish purchase versus posting date before assigning the row to a cycle.
Resolve a purchase at the cycle boundary
Consider a separate, fictional date-check exercise: the issued bill covers March 2–April 1, and its inclusion rule uses the posting date. Apply that stated rule before comparing the card CSV with a calendar-month spending report.
| Purchase date | Posting date | Result under this exercise’s rule |
|---|---|---|
| March 1 | March 2 | Included in the March 2–April 1 cycle, despite the earlier purchase date |
| March 31 | April 2 | Outside this cycle; check the next issued bill rather than adding it to this one |
A March purchase report could therefore disagree with this bill while both follow their own date rules. Retain the purchase date when needed for spending analysis, but do not silently substitute it for the posting date used for cycle matching. Establish your own bill’s inclusion rule before applying this example.
Use an explicit amount convention
This synthetic worksheet uses increases in card debt as positive, and payments or refunds reducing that debt as negative, for one invented USD card account. Currency comes from this description because the three-field CSV has no currency column. Your issuer or importer may use a different convention.
date,description,amount
2026-03-04,Purchase,120.00
2026-03-06,Merchant refund,-20.00
2026-03-15,Card payment,-200.00
2026-03-20,Interest charge,5.00
For this invented complete cycle, an opening debt of 300.00 USD, purchase 120.00 USD, refund 20.00 USD, payment 200.00 USD and interest 5.00 USD lead to closing debt of 205.00 USD. Apply that relationship only when the source amounts, signs and posted-cycle coverage meet those stated assumptions.
Compare each known purchase, refund and payment with the PDF before transforming signs for an importer. A payment is not another purchase. A refund is not automatically a duplicate purchase, and a pending authorisation is not necessarily a posted charge. Do not add both an interest-summary figure and its already included transaction row.
Calculate spending without counting a repayment as another purchase
A spending report and a debt-rollforward control need different classifications. In the synthetic cycle above, keep the signed amount and add a separately reviewed Type in a working sheet: Purchase, Refund, Repayment and Interest. These types are defined by the exercise, not inferred by LocalExtract from the description.
With A = date, B = description, C = amount and D = Type, Excel’s SUMIFS can sum C for an established D value. =SUMIFS(C2:C5,D2:D5,"Purchase")+SUMIFS(C2:C5,D2:D5,"Refund") gives 100.00 USD for purchase 120 and refund −20. It excludes the −200 repayment and the +5 interest under this explicitly defined purchase-net-of-refunds measure. Interest-inclusive cost would be another measure, not the same total.
By contrast, all four signed movements total −95.00 USD, which explains debt changing from 300 to 205. Treating that −95 as net purchases would mix repayment funding with spending. SUMIFS cannot decide the correct financial type, identify duplicates or establish the statement cycle; its sum and criteria ranges must also have matching dimensions.
The synthetic spending-versus-debt worksheet records both controls and a repayment-included counterexample. Python reproduces these defined totals; Excel has not been executed. Formula separators may be semicolons in your Excel locale, while the quoted teaching labels must match your working-sheet values exactly. Preserve the source rows when building the report instead of deleting repayment evidence from the statement extract.
Use the shared formatting method
CSV formatting and import checks.
Formatting references: informational RFC 4180, Microsoft’s text/CSV import help and Intuit’s QuickBooks Desktop Web Connect instructions.
Finish with a traceable review
Download the card-cycle worksheet to record cycle boundaries, debt signs and unresolved entries. It is a manual aid, not an automatic validator.
Before accepting results, check the complete billing cycle, source pages, cardholder sections, purchase/refund/payment directions and the stated closing debt. Do not insert a balancing entry to conceal an unresolved row.
LocalExtract’s free evaluation covers a cumulative ten pages; Pro is USD 10/month or USD 60/year. Parsing and optical character recognition (OCR) happen locally, while account, subscription, update and periodic Pro verification functions require a network. No desktop batch queue, direct Excel workbook (XLSX) export or automatic accounting synchronization is provided.
LocalExtract Team
We build LocalExtract, a desktop bank statement converter. Questions or corrections? Contact · Editorial policy