SYNTHETIC EDUCATIONAL EXAMPLE. Not customer data or a RowSpeak run.
Scope: DemoStore, USD, September 1-30 2026 inclusive, UTC dates.
One order per orders row; one event per processor row; one payout per payout
row; one bank credit per bank row. Refund and fee events are already negative.
Opening processor balance and sent-but-not-posted payouts: both zero.
Ending processor balance: E09 = USD 200, not yet assigned to a payout.
All other settlement flows are zero. No tax, shipping, FX, reserves, disputes,
failed payouts, unpaid orders or prior-period adjustments in this fixture.
The paid_sales field is a simplified before-refund paid amount, not a claim
about accounting revenue. September data completeness is stipulated for this
exercise; a real export needs independent confirmation of completeness.
INPUTS: orders.csv, processor-activity.csv, payouts.csv, bank.csv.
EXPECTED OUTPUTS: expected-bridge.csv and expected-payout-matches.csv.
Do not upload the expected outputs during a blind tool evaluation.
Stress test: in a COPY of bank.csv change B02 credit from 3920 to 3910.
Expected: bank total 8990; P02 short by 10; overall unexplained difference 10.
Do not invent a fee to force the records to agree.
Verified with Python Decimal; no RowSpeak execution is claimed.
