Nexteam
Skip to content
← All learning cases

Excel & Financial Modeling / XFM-01

Clean, Map and Reconcile a Finance Dataset

Clean a small ledger export while preserving every exception.

Foundation · 45 minutes (estimate) · Original synthetic data

Prerequisites: Basic Excel and CSV imports

01 / Learn the method

Learning outcomes

Clean a small ledger export while preserving every exception. Calculate the result, reconcile it and communicate its limitations.

Method

Keep raw rows intact, then create a normalized layer with explicit parsing and duplicate rules. A repeated transaction ID is not enough to delete a row if its contents differ. Reconcile raw totals, excluded duplicates and accepted totals; keep unmapped or invalid rows visible instead of silently dropping them.

Smaller worked example

Raw amounts 100, 100 and 50 include one exact duplicated 100 record. Raw total 250 less excluded duplicate 100 equals accepted total 150. An unmapped account still belongs in accepted totals until its treatment is resolved.

02 / Put it to work

Business context and rules

Fictional Birch Agency, September 2026, USD. The companion CSV has seven rows: T01 sales 1,200.00; T02 expense (250.00); an exact duplicate T02; T03 sales 800.00; T04 expense -100.00 to unmapped account 6999; T05 expense -50.00 with invalid date 2026-09-31; T06 sales 300.00. Dates use ISO format. Account 4000 maps to Revenue, 6100 to Operating expense; 6999 is unresolved. Parentheses mean negative. Exclude only the exact duplicated T02; retain T04 and T05 in the reconciliation but flag them. Accept the signed amount of T05 for reconciliation without inventing a corrected date. Raw signed control total is 1,650.

Original synthetic inputs

InputValueUnit
T01 normalized amount1200USD
T02 normalized amount-250USD
Duplicate T02 normalized amount-250USD
T03 normalized amount800USD
T04 unmapped amount-100USD
T05 invalid-date amount-50USD
T06 normalized amount300USD
Raw signed control1650USD

Required deliverables

Import the raw CSV as text, preserve it, build typed amount / valid-date / account-mapping columns, and provide an exception log for duplicate, invalid date and unmapped account. The normalized Inputs tab is a reconciliation aid; submitting its totals without cleaning the raw CSV does not satisfy the task.

Use formulas for derived amounts and preserve source data. Put narrative deliverables in the workbook response area; expand it as needed. Compare amounts within 0.01 of the stated unit and percentages within 0.1 percentage point. No unsupported balancing plugs.

Required raw dataset

Download XFM-01_raw_v1.csv from the case download panel. It deliberately contains duplicate and invalid records. The workbook Inputs sheet contains numeric control aids, not a substitute for cleaning the raw file.

03 / Review your work

Try the assignment before opening the answer.

Open the worked answer and teaching notes

Worked numerical schedule

MeasureValueUnit
Raw signed total1,650.00USD
Accepted total excluding exact duplicate1,900.00USD
Known revenue total2,300.00USD
Known expense total including date exception-300.00USD
Unmapped accepted amount-100.00USD
Accepted mapping reconciliation0.00USD
Raw control residual0.00USD

Interpretation and recommended actions

There are 7 raw rows and 6 accepted records. Raw signed total 1,650 becomes accepted total 1,900 after excluding the duplicate -250. Revenue is 2,300, known expense -300 and unmapped -100. Keep T04 for mapping review and T05 for source-date correction; neither disappears from the total. Preserve the duplicate row and its exclusion reason in an audit trail. No final clean-date claim is appropriate until T05 is corrected by the source owner.

Scoring rubric - 100 points

DimensionPointsAwarding guidance
Calculation40Raw import / typed normalization 15; exact duplicate handling 10; totals / exceptions 15.
Interpretation25Correct application of the case rules 10; explain the business decision 10; identify evidence or limitations 5.
Controls / audit trail20Traceable formulas 8; independent reconciliation 8; explicit units and signs 4.
Communication15Decision and numerical headline 5; actions with owners and evidence 5; concise response covering all required deliverables 5.

Award method credit after an isolated arithmetic error rather than repeatedly deducting for it. Equivalent account labels and well-supported alternative recommendations are acceptable. Numerical tolerance is 0.01 in the stated units; no universal passing score is prescribed.

Common mistakes

Dropping an unmapped amount; converting an invalid date into October without approval; deleting all repeated IDs rather than exact duplicates.

Staged hints

Normalize amount text without changing signs. Compare complete duplicate rows. Reconcile accepted mapped plus unmapped amounts.

Instructor notes

Prerequisites: Basic Excel and CSV imports. Suggested use of the estimated 45 minutes: spend roughly 15% on the lesson and smaller example, 55% on the independent task, 20% on comparing approaches and 10% on the decision discussion. Timing is untested. Ask learners to explain why the numerical check is necessary but not sufficient.

For a simpler class, provide the model structure and work through one driver. For an extension, change one operational assumption and require a new reconciliation and recommendation. Verify the new key before distributing any variant. Open files in the intended spreadsheet application before class. Solutions are learning resources, not secure hiring examinations. Expert review remains pending.

Continue this learning path

XFM-02 · Build an Integrated Three-Statement Model

XFM-03 · Audit a Broken Financial Model

AI-assisted synthetic teaching case. Expert review and native Excel / Sheets testing pending.

Choose what you allow. Optional cookies are off until you enable them. You can browse, contact us, and request talent without accepting them.