What to check
A·S·F
What happens to a $400 refund on a $2,500 fuel purchase?
Is the refund netted against its category, or added as spend or dropped? Either of the last two leaves the total too high and off the P&L.
Netted within its category before the factor is applied, so it is priced as $2,100 of fuel. The report states how many credit rows were netted, and the Disclosure-Aligned ledger lists each one with its category. Reconciles to the GL by construction.
Electricity in California vs. Ohio: same number?
Does electricity use your state's grid, or a national average? A national average misstates a company on a clean grid and one on a coal-heavy grid alike.
A per-state factor, from the state you enter, anchored on that state's own commercial electricity price. California is 0.70 kg/$; Ohio is 4.55. A dollar of Ohio electricity carries 6.5× the emissions of a dollar in California.
A German supplier's ledger with EU-formatted numbers?
Does it read comma decimals and bracketed negatives, and show you how it read them?
Intake detects comma or period decimals, localized headers and bracketed negatives, and the report's Data Intake Decisions section prints how it read the file. A DATEV Buchungsstapel (SKR 03 or 04) is read natively, with input VAT split out by tax key, and priced with Germany's own grid factor and Open CEDA factors for Germany.
A €90 diesel receipt in Dublin: how many litres?
Is fuel bought abroad priced at that country's pump price, or divided by a US one?
Fuel and gas bought by spend are turned into litres, gallons or kWh at the official price for that country and month, net of recoverable VAT (EIA in the US, DESNZ in the UK, the EU Weekly Oil Bulletin and Eurostat in Europe, AIP in Australia, PPAC in India), then multiplied by a physical emission factor. A UK or EU ledger is not divided by a US pump price.
How confident is the classification on row 2,847?
Can you see how confident each row's match is, and which rows were guesses?
Every row carries a confidence score, and the Data Quality Summary gives the split by confidence band on every tier. Basic and Complete print each row's score for the top 20 rows by spend; Disclosure-Aligned for every row. Rows the engine could not identify are listed, never guessed.
Where does this number come from, exactly?
Can you reproduce any single row's number from its cited source?
Every priced row carries its factor, the rule that matched it, and the source document and factor pack behind it. Basic and Complete print the top 20 rows by spend; Disclosure-Aligned prints every row.