The terms an invoice auditor lives with.

The weight your invoice is actually billed against — the greater of the shipment's gross weight (what it weighs on a scale) and its volumetric weight (what it would weigh if it were as dense as the carrier's contracted conversion factor allows). On Road and Sea consignments it's also compared against a third value derived from loading metres (LDM) for pallet shipments.

Each mode has its own conversion factor written into the rate card — typically 150 kg/m³ for groupage road, 167 kg/m³ for IATA air, 1 000 kg/m³ for sea LCL, and a per-LDM equivalent (commonly 750–1 850 kg/LDM) for full and groupage pallet shipments. Whichever computation yields the highest number is the chargeable weight; that's the value bracket-lookups for base freight and weight-scaled surcharges use.

Why it matters in an audit: inflated chargeable weight is the single most common over-bill pattern. A carrier billing a 100 kg / 0.8 m³ road shipment at 150 kg (volumetric factor) when the contract specifies 167 kg/m³ produces a silent ~12% overcharge on every freight line and on every percent-of-freight surcharge stacked on top. Sentra recomputes the chargeable weight from gross + volume + LDM on every shipment and brackets the rate independently, so this category of error doesn't slip past.

A variable carrier charge passed through to reflect fuel-price movements, almost always quoted as a percentage of base freight and pegged to a published weekly or monthly index. Road in the UK uses the DfT weekly ULSD pump price; central Europe uses BAG, the German Federal Office for Goods Transport diesel-price track; sea uses the carrier's own BAF/CAF series; air uses an IATA-published fuel index.

Each contract specifies which index, which publication cadence (weekly / monthly), and the formula that converts an index step into a percentage. Some carriers compound the fuel surcharge with toll and security surcharges; others apply it only to base freight. Both shapes show up in real invoices.

Why it matters in an audit: the most common fuel-surcharge errors aren't "wrong percentage" — they're "wrong index week", "applied to the wrong base", and "compounded with surcharges the contract excluded". Sentra fetches the relevant index for the invoice's shipment date (not the billing date) and recomputes the surcharge from contract terms, then compares to the invoiced amount.

A single carrier invoice covering many shipments — German for "collective invoice" — typical of large carriers (DB Schenker, Kühne+Nagel, Rhenus, Dachser) billing customers on a weekly or monthly cadence. One PDF, one invoice number, one VAT total — but dozens or hundreds of individual consignments inside, each with its own route, weight, and surcharges.

From an audit perspective the unit of work is the shipment, not the invoice. A Sammelrechnung with 250 shipments needs 250 independent rate look-ups, 250 weight recomputations, and 250 surcharge validations — and any dispute letter has to reference the specific shipments at fault, not the invoice as a whole.

Why it matters in an audit: generic OCR pipelines treat a Sammelrechnung as one document and lose the per-shipment structure. Sentra splits a collective invoice into its shipments at extraction time, so every audit, every dispute, and every credit-note match operates at the shipment grain — the only grain at which "this charge is wrong" can be defended.

A canonical code that maps a carrier's free-text charge label to a fixed taxonomy Oracron maintains — 152 codes covering base freight, fuel, toll, security, customs, accessorials, demurrage, detention, and the long tail of carrier-specific surcharges. Each carrier labels the same thing differently — K+N's Zonentarif, Schenker's Frachtkosten Hauptlauf, DHL's Base rate are all the same canonical FREIGHT_BASE. The fee code is the lingua franca.

Mapping uses a three-tier strategy: deterministic rules first (regex on canonical labels), keyword overlap second, and a retrieval-augmented AI step last. Every mapping decision is logged with its tier and confidence, so a "wrong fee code" finding in an audit can be traced and corrected on the next invoice without touching the engine.

Why it matters in an audit: without a canonical code, "is the carrier billing the same fee twice?" is unanswerable — line A says Maut DE, line B says Toll Germany, and a naïve compare reads them as different charges. Sentra collapses both to TOLL_DE and surfaces the duplicate. The 152-code taxonomy is the spine that makes every downstream check possible.

A carrier-issued document that reverses or reduces a previously-invoiced amount. STORNO typically cancels a whole invoice (often a duplicate or wrong-customer issuance), while Gutschrift is a partial credit — the carrier accepting one or more disputed lines. The two terms blur in practice; some carriers use one label for both shapes.

Mechanically a credit note carries its own invoice number, references the originating invoice, and shows negative amounts. The originating reference is the critical piece: it's what lets the platform attribute the credit to the right dispute case and close the cash-recovery loop.

Why it matters in an audit: a credit note received without being linked back to its originating dispute leaves the case sitting open and the recovered amount uncounted in dashboards. Sentra extracts the originating-invoice reference at ingestion, matches it to the live dispute case, and credits the recovery to the right shipment — without that link, ROI numbers silently understate.

An audit verdict assigned to a shipment whose rate card has no comparable bracket — e.g. a 73 kg parcel on a route where the contract only specifies 0–50 kg and 100–500 kg brackets, or a Saturday delivery on a Mon–Fri-only contract. Sentra refuses to invent a benchmark by extrapolating outside the contract; the verdict is no-benchmark, with the gap clearly stated.

A no-benchmark verdict is not a failure — it's a disclosure. The shipment ships, the invoice is paid, but the audit log records: "we did not have a comparable contracted rate to validate this charge against." That visibility is what lets the finance team prioritise contract-coverage gaps: if 18% of last quarter's spend landed on no-benchmark rows, that's a procurement priority, not a dispute opportunity.

Why it matters in an audit: systems that auto-extrapolate produce confident-but-fictional findings — dispute letters get sent that the carrier rightly refutes, trust erodes. Sentra's no-benchmark verdict is the conservative path: disclose what the contract doesn't cover instead of inventing what it might have meant.