Version carrier rates so historical charges remain auditable

Carriers change base prices, surcharges, area lists and thresholds on different dates. One file named current_rates.xlsx destroys historical context. Every version needs an identifier, effective-from date and, once known, effective-to date.

Version selection follows the contractually relevant service date, not automatically the invoice date. Mid-month changes and retrospective corrections require distinct versions.

Automation translates a PDF or rate card into structured rules, but the structured form does not replace the original. Each executable rule links to the document section or table row from which it was derived.

Record who reviewed the translation and any assumptions made. A later reviewer can then distinguish a carrier variance from a misinterpreted rule.

Negotiated discounts, minimum charges, free allowances and excluded services must not live as comments beside the price logic. They alter the outcome and belong to the same versioned rule set.

Scope must be machine-readable when an exception applies only to one account, product or lane. Conflicts between standard and negotiated rules need documented priority.

Fuel and energy surcharges may change frequently. UPS describes weekly index-based adjustments, while DPD publishes variable energy and fuel components. The percentage should therefore not be embedded in a long-lived base-rate version.

Use a referenced time series containing source, reference period, effective date and calculation basis. The same applies to remote-area and postal-code matrices.

Rounding per line, surcharge or invoice can create thousands of differences at scale. Rounding level, decimals and calculation order must be versioned like prices.

For multiple currencies, preserve original amount, rate, source and rate date separately. Reporting conversion must never overwrite the audited historical value.

An approved historical version should not be overwritten. Corrections create a new version referencing its predecessor. Hashes of source and structured rules help reveal unintended changes.

Each version needs representative test shipments with expected results. Regression tests run before production, proving not just what is stored but how the rule behaves.