Blog · AI and self-service analytics · Finance and FP&A teams
Set up a financial-statement analysis request using a ledger export, account mapping and budget. Rebuild a worked quarter comparison, reconcile the returned figures, and review the explanation before it enters a finance report.
ChatGPT financial analysis should begin with mapped financial records and finish with independently reconciled figures. Give the request a specific period, comparison and sign convention, then check the returned statement lines before using the explanation. The finance reviewer owns the classifications and the conclusions.
Use a question such as: How did operating profit change from the prior quarter, and why did it miss the current-quarter budget? That is specific enough to test against a ledger and approved plan. A broad request to assess financial health leaves too many choices unstated.
OpenAI's official spreadsheet-consolidation guidance describes supplying source exports, explicit matching rules and reporting definitions, then inspecting formulas, unmatched records and assumptions. Apply that approach to finance records by making the account mapping and statement reconciliation part of the requested output.
For this workflow, prepare:
Keep workbook versions and export timestamps with the review. Use an approved company workspace and follow your company's rules for financial-file uploads. This article uses a compact income-statement comparison; financial analysis covers the broader set of questions.
A fictional US industrial distributor compares the quarters ending June 30 and September 30, 2026. Every amount below is USD. Revenue and expense accounts have already been converted to positive presentation amounts using the approved account mapping; retain the original debit and credit fields in the supporting export.
Place the following table in columns A:E, with headers in row 1 and account records in rows 2:6. These are the complete account-level totals for this simplified example, derived from the underlying transaction rows.
| Account | Statement line | Prior actual (USD) | Current budget (USD) | Current actual (USD) |
|---|---|---|---|---|
| 4000 Product sales | Revenue | 800,000 | 900,000 | 870,000 |
| 4010 Service revenue | Revenue | 200,000 | 220,000 | 230,000 |
| 5000 Materials | COGS | 420,000 | 470,000 | 465,000 |
| 5100 Direct labor | COGS | 180,000 | 190,000 | 195,000 |
| 6100 Sales and administration | Operating expenses | 220,000 | 235,000 | 240,000 |
Do not sum every numeric row into a supposed statement total: revenue and expense lines have different roles. Check that each account has one approved mapping, and separately retain any unmapped account for review. An account can be split across reporting lines only with an explicit allocation that reconciles to the original amount.
The request can be plain business language:
Compare current actuals with current budget and prior actuals using the supplied account mapping. Show revenue, COGS, gross profit, operating expenses and operating profit. Show the spreadsheet formulas and contributing account rows. Use actual minus budget for variance amounts, and label whether each movement helps or hurts profit. Report unmapped accounts and reconciliation differences. Separate observed movements from explanations requiring additional evidence.
Request an answer tied to these files rather than assumptions about industry norms. If the returned explanation names pricing, volume or customer churn, ask which records establish that cause. Statement lines alone do not distinguish those drivers.
The general AI financial-analysis workflow explains that division between computed evidence and interpretation. Here the review is anchored to the supplied statement mapping; AI analytics provides the broader file-based approach.
Start a summary in A10:D15 with headers in row 10. Put Revenue in A11, COGS in A12, Gross profit in A13, Operating expenses in A14 and Operating profit in A15. Columns B:D hold prior actual, current budget and current actual.
For revenue, enter this in B11 and copy across to D11:
=SUMIFS(C$2:C$6,$B$2:$B$6,$A11)
Use the same formula pattern in B12 and B14, matching their respective labels in column A, then copy across. Microsoft's SUMIFS documentation specifies the sum range followed by criteria ranges and their criteria. Keep those ranges aligned with the same source rows.
Calculate gross and operating profit in B13 and B15, then copy across:
=B11-B12
=B13-B14
The rebuilt statement is:
| Statement line | Prior actual (USD) | Current budget (USD) | Current actual (USD) | Actual minus budget (USD) |
|---|---|---|---|---|
| Revenue | 1,000,000 | 1,120,000 | 1,100,000 | -20,000 |
| COGS | 600,000 | 660,000 | 660,000 | 0 |
| Gross profit | 400,000 | 460,000 | 440,000 | -20,000 |
| Operating expenses | 220,000 | 235,000 | 240,000 | 5,000 |
| Operating profit | 180,000 | 225,000 | 200,000 | -25,000 |
The variance formula in E11 is =D11-C11, copied down. A positive expense variance hurts profit, while a positive revenue variance helps it. The same subtraction convention therefore needs different business labels for revenue and costs.
Current gross margin is 440,000 / 1,100,000 = 40.0%. Budget gross margin is 460,000 / 1,120,000 = 41.1%, rounded to one decimal place. Use =D13/D11 for the current figure and =C13/C11 for budget. The gross profit margin method explains the denominator in more detail.
First reconcile each source-period revenue, COGS and expense total to its approved statement. Use the same currency, entity scope, close status and budget version. Agreement within the rebuilt workbook is necessary, but agreement to an independent approved total is the stronger check.
Next rebuild the budget miss from account movements. Product revenue is $30,000 below budget; services are $10,000 above it. Materials are $5,000 below budget and labor is $5,000 above it, leaving total COGS unchanged. Operating expenses exceed budget by $5,000.
Operating profit variance = -30,000 + 10,000 - (-5,000 + 5,000) - 5,000 = -25,000
That agrees with $200,000 actual operating profit minus the $225,000 budget. Against the prior quarter, operating profit rose by $20,000: $200,000 minus $180,000. Both statements can be true: profit improved sequentially and missed plan.
Assign follow-up ownership from that evidence. Sales reviews the product revenue gap; operations reviews the offsetting materials and labor movements; finance reviews the expense overspend. Invoice quantities, prices and cost records can establish causes. A fluent explanation does not replace those records.
Covirage is not affiliated with OpenAI or Microsoft.
Bring a mapped ledger export and the matching budget or prior-period workbook. Covirage maps the columns, its tools compute and check the comparisons, and the external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.
See FP&A reporting in Covirage for the supporting finance workflow.
Prepare the closed-period ledger export, approved budget, account-to-statement mapping and comparison statements. Include period labels, currency, account codes and stable row identifiers. Note the sign convention and any excluded entities so the request has a defined scope.
Rebuild its numerator and denominator from the reconciled statement lines, using the same period and scope. Inspect the underlying formula rather than the displayed percentage alone. A correctly calculated ratio can still mislead when an account mapping or period boundary is wrong.
It can establish the amount and location of the change. A cause such as lost customers or lower prices needs customer, product or invoice evidence. Ask for those supporting rows before including a causal explanation in a management report.