Invoice auditing
Use Zuora AI to audit draft invoices and bill runs for anomalies, such as invoice line variances, missing expected charges, and abnormal invoice totals, before you post the invoices.
Invoice auditing helps you identify unexpected invoice item changes, missing expected charges, and abnormal invoice totals before invoices become final.
After a bill run completes, Zuora generates the invoices but does not post them. Auditing invoices at this stage helps you review potential issues before you post the bill run and make the invoices available for payment collection.
Prerequisites
-
To save validation results to Zuora, define the required custom fields on the
InvoiceItemobject for line variance validation or on theInvoiceobject for summary variance validation. If the custom fields do not exist, Zuora AI skips the persistence step. -
You must have the API Write Access permission under the Platform role to write validation results back to Zuora. If you do not have this permission, Zuora AI skips the persistence step.
Available invoice auditing skills
Zuora AI provides the following invoice auditing skills:
- Invoice line variance validator
- Reviews invoice items for amount and usage quantity deviations from the most recent posted billing period for the same charge.
- Invoice summary variance validator
- Reviews draft invoices for missing expected charges and abnormal spending totals compared to the most recent posted invoice in the same billing group.
You can use both skills together for broader pre-post coverage. The line validator identifies item-level pricing and usage anomalies, and the summary validator identifies invoice-level gaps and spending spikes.
Date Range of Data Processing
The invoice auditing skills use Data Query to retrieve historical billing data for validation. Data Query supports asynchronous, read-only SQL queries, and Data Query Live is subject to query processing limits.
When querying large transactional tables such as InvoiceItem, Data Query Live can reach the maximum input-record limit. After filters are applied, the maximum number of input records per table is 10,000,000. Zuora recommends adding filtering logic in the WHERE clause to keep the number of input records below 10 million; otherwise, the query can fail.
To address this behavior, the Invoice line variance validator and Invoice summary variance validator to automatically handle large volumes of historical invoice items that would previously fail because of Zuora Data Query's 10-million-row limit.
The validator uses an adaptive lookback window. It starts with a 13-month historical baseline and, if the query exceeds the row cap, automatically retries with progressively shorter windows (13 → 10 → 7 → 4 → 2 months) until the query succeeds. As a result, large-volume invoice item validations that previously returned an error can now complete without manual intervention.
The invoice auditing tool supports a maximum lookback window of 13 months. In high-volume scenarios, validation might still not succeed. If the result set still exceeds the limit at the minimum 2-month window, the validator displays a clear message that explains the constraint and recommends next steps, instead of failing silently.
For example, if you have 10 million invoice items within the last 4 months and the lookback window is reduced to 2 months, the validator might not be able to find the correct historical baseline for a quarterly billing period.
For more information about Data Query processing limits, see Overview of Data Query.