Record which duplicate candidate you kept, and why
A duplicate review CSV connecting both candidate rows, source files, matching fields and the retained or excluded decision, with a fictional overlap example.
Matching amounts do not prove that two rows describe the same sale. This log keeps both candidates visible while you inspect references, dates, staff and item detail. It records a decision about the working calculation without deleting either source export.
The useful part
Identify both rows precisely and explain the evidence for keeping or excluding the candidate. A shared invoice number alone is not enough.
Your working files
Download blank templateCSV · Column headings only. Add your own records to a copy.Download fictional exampleCSV · Completed sample records in AUD. Use for practice, not a live close.Treat a match as a question
A repeated service may be legitimate, and an invoice may contain several different items. Begin with a candidate pair, not an assumption that the second row is wrong. Keep the source filename and row number for each side so another reviewer can reproduce the comparison even when the invoice reference is repeated throughout an export.
Capture both sides of the comparison
| Field | What to enter |
|---|---|
| Review ID | Unique identifier for the candidate pair. |
| Retained file / Retained row / Retained reference | Exact source location of the row already counted. |
| Candidate file / Candidate row / Candidate reference | Exact source location of the additional row being reviewed. |
| Matching fields | Fields actually compared, including item detail where available. |
| Candidate commission AUD | Amount the additional row would contribute if included. |
| Candidate decision | Retain, exclude or unresolved; this describes the candidate only. |
| Reason | Evidence supporting that outcome, not merely “duplicate”. |
| Reviewer / Reviewed date | Person and date of the decision. |
| Evidence reference | Location of supporting exports or additional checks. |
An overlapping file adds $80.00 twice
In the fictional example, the same service appears on row 2 of a weekly file and row 7 of an overlapping export. The reference, date, earning staff, item, category, amount and supplied tax agree. Each copy would contribute $80.00 under the supplied rate. Excluding the candidate leaves the original $80.00 counted once.
| State | Retained line | Candidate line | Combined commission |
|---|---|---|---|
| Before review | 80.00 | 80.00 | 160.00 |
| Candidate excluded | 80.00 | 0.00 | 80.00 |
Look for evidence that changes the answer
If the source cannot distinguish two identical-looking legitimate items, ask for better item-level evidence. A matching algorithm cannot settle that uncertainty by itself.
- Compare item-level references, not only the invoice total.
- Check whether either file includes a different date range or attribution filter.
- Inspect quantities, categories and refund signs when repeated rows look similar.
- Keep both lines when evidence shows distinct activity, and record why the match was misleading.
Preserve the record you reviewed
The log does not remove rows or change a Trimsum workspace. Apply a checked decision through the relevant review controls and keep this file with the evidence pack if useful. When a candidate overlaps an earlier finalised run, inspect that history before treating it as a duplicate within the current draft. Never alter an original export just to make the totals reconcile.
Put it to work.
Compare records before deciding which activity should be counted.
Find candidate duplicate rows