Fictional worked example
A clean report. A wrong margin.
The total looks plausible. The definitions do not survive a check.
This is a deliberately flawed, constructed report and a completed arithmetic review of its one row. It is not a client report, a live-system audit, or evidence of a teammate trial.
State the intended use
A finance lead needs realized gross margin for September, using net recognized revenue and the historical cost attached to the shipped units. Gross margin here excludes fees, freight, overhead, and tax. It must not be labeled contribution margin or profit.
| Evidence ID | Observed input | Meaning |
|---|---|---|
| S1 · sales ledger | 100 units × $10 = $1,000 | All shipped in September; no discounts or tax in this fixture. |
| R1 · linked return | 10 units refunded at $10 = $100 | Same original order; refund recognized in September. All 10 units restocked as sellable inventory. |
| C1 · shipment cost record | $6 per unit | Historical cost at original dispatch. Returned units reverse this same cost. |
| C2 · current item card | $7 per unit from October 1 | Replacement cost after the reporting period; not this shipment's historical cost. |
| D1 · delivered report, version 1 | Revenue $1,000; COGS $700; gross margin $300 / 30% | Seed formula: 100 × $10 − 100 × $7, divided by $1,000. It ignores R1 and uses C2. |
Recompute without the report's formula
Start with the posted transactions, not the current item card.
- Net revenue: $1,000 sales − $100 refunds = $900.
- Net cost of goods sold: 100 × $6 shipped cost − 10 × $6 returned to inventory = $540.
- Gross margin: $900 − $540 = $360; $360 ÷ $900 = 40%.
Cross-check: 90 retained units × ($10 − $6) = $360. This second calculation uses the same disclosed evidence but a different arithmetic route; it is not an independent source confirmation.
Findings with consequences
| Claim and location | Observed → expected | Consequence and smallest correction |
|---|---|---|
| D1 revenue cell | $1,000 → $900, from S1 less R1 | Blocker: net sales are overstated. Include the linked, in-period refund. |
| D1 COGS cell | $700 → $540, from C1 and R1 | Blocker: cost basis and return treatment are wrong. Use shipment cost, then reverse cost only for the documented restocked units. |
| D1 gross margin cells | $300 / 30% → $360 / 40% | Blocker: the finance lead would assess the order incorrectly. Recalculate after fixing revenue and cost; label the metric and period. |
Challenge the assumptions
Alternative return treatment: if the returned units were unsellable with no cost recovery, the $60 reversal would not be justified. Under that simplified alternative, net cost stays $600 and gross margin becomes $300 ÷ $900 = 33.33%. The return disposition therefore matters; it cannot be inferred from a refund alone.
Boundary cases to run on a repaired implementation: missing shipment cost should flag the row, not substitute today's item cost; a refund posted in October should follow the agreed period policy; duplicate return IDs must not reverse revenue twice; zero net revenue must not produce an infinite margin percentage.
Those implementation tests remain open. This page checks the supplied arithmetic and definitions; it does not pretend a reporting application was executed.
Verdict: version 1 is not ready for this use.
The one-row fixture contradicts three displayed measures. The corrected expected answer is $900 net revenue, $540 net COGS, and $360 / 40% gross margin.
Scope: one supplied SKU and order cohort, with explicit return timing and disposition. The underlying ledgers, accounting policy, other products, exports, and rendered reporting tool were not independently inspected.
Next owner action: the report builder corrects the cost join and return logic, preserves version 1, and reruns this fixture plus the boundary cases. Finance confirms the period and return policy. Only then review the repaired version and expand the sample. A corrected example alone does not clear the whole report.