# Worksheet instructions

These files contain a blank template and synthetic worked example. They are planning records, not a payment import specification. Import IDs as text and preserve leading zeros. Amounts are USD. Keep bank details, tax identifiers, patient data, customer addresses, and confidential source content out of these files. Store evidence in an access-controlled system and use references here.

The example uses invented rates and agreements. Replace them with your approved terms, due dates, authority rules, and legal requirements. No formulas or external links execute in these CSVs. Calculate values in your controlled working copy and validate totals before relying on them.

## Session and cancellation decisions

One row represents one compensation component. A session can have linked components, but each must have a unique line ID. Use the session reference to detect duplicate exports; retain separate line IDs for authorized additions. Rate V1 applies to service dates before October 15; V2 applies from October 15 in this example. C1 is a separate cancellation rule. These are illustrative contract rules.

Approved amount is the amount authorized by the recorded decision. Potential amount preserves the submitted or possible amount while a decision is pending. A zero approved amount on a pending row does not mean no payment is owed. The decision owner must resolve that question by the applicable deadline.

The 10 attended lines total $920: six at $90 and four at $95. Add the approved $40 cancellation and $15 prior underpayment correction for $975. The other $40 cancellation is pending. Total potential amount is $1,015. Do not count it as approved or paid. All approved lines are unpaid in this example.

After payment, retain these records and connect the provider's actual payment reference in your system. If S-012 later qualifies, retain its source identity and create the authorized $40 obligation without selecting the earlier paid lines again. Store clinical evidence outside the ledger. A reference is not necessarily de-identified data.