How to Automate Deposits Reconciliation for PayPay and Rakuten Pay: Difference Alert Operations Guide

1) The goal of the Difference Alerts is to make sure you don’t miss anything

Difference alerts are not notifications meant only to increase the number of items. They are a mechanism that helps you identify the cause and run a recheck in the shortest possible time when the matching results fall outside the “expected range.” To avoid disrupting operations, first clarify the “definition of the differences” and the “handling of alerts.”

• Definition: As examples, we fix inspection criteria such as “the deposit amount does not match,” “there is no linkage to the target order,” and “the tax category and fee calculation differ from what was expected.”
• Category: We separate “Differences you can absorb automatically” from “Differences people should focus on.”
• Procedure: We will document the steps: notification → one-time isolation → re-verification → assigning cause labels.

2) Finish data formatting first

Payments data from PayPay and Rakuten Pay may not always match in item names, granularity, or aggregation units, even when they mean the same thing. Before automatic reconciliation, we will standardize the following formats as a “common normalization layer.”

1. Define an order matching key using the transaction ID (or reference key), order ID, and sales type (such as standard/refund).
2. Unify the calculation units for fees and taxes, and convert them into a format that can be compared with the amount on the reconciliation side.
3. The date and time allow for discrepancies between the time zone and the aggregation close (batch/daily), and determines the window for re-verification.

If this is ambiguous, difference alerts become “noise” and operations become exhausting. The shortest route is to align verifiable criteria first.

3) Design the boundary between automated matching and manual review

To operate a difference alert, you need to decide “how long to automatically keep tracking” and “when to hand it over to a person.” A recommended approach is to split the rules into three stages at the start.

4) Difference Alert Operation Template

To keep operations stable, we standardize the “way of looking” when discrepancies occur. The discrepancy alert content must include, at minimum, the following:

• Difference type (e.g., amount, linkage, tax category, fee, refund timing)
• Matching Key and Target Scope (Period, List of Order IDs/Transaction IDs, Target Currency)
• Automatically tested reconciliation conditions (which window, and what was recalculated)
• Next action (re-run, register exception rules, confirm with relevant departments, etc.)

If the necessary information is available, the responsible person can focus on root cause analysis. An alert should be “the entry point to a problem” and “a tool to reach a solution.”

5) Assuming multi-currency and Japanese tax requirements

In cases where the currencies differ between the payment side and the accounting side, discrepancy checks can become less consistent. In particular, this is affected by the base date for the exchange rate, rounding of fractions, and how taxes are handled. As an operating rule, let’s decide the following:

• Difference Determination Base Currency (e.g., JPY-Based) and Conversion Rules
• Tolerance (how much to allow for rounding differences, fee differences, and tax differences)
• Tax breakdown consistency check (including next-month shifts and the timing of refund postings)