01 / DATA PREPARATION
Define the population before running tests.
Start with a clear review question and a defined period. Understand what the source export represents: purchases, refunds, voids and pending transactions may require different treatment. Retain an unchanged source copy and document the choices used to prepare the analytical population.
- Identify transaction dates, amounts, cardholders, merchants and available reference fields.
- Confirm the meaning of each field rather than relying on its column name alone.
- Review missing values, duplicate source records and inconsistent formats.
- Consider currency and transaction status before combining or comparing amounts.
- Record exclusions and their rationale so another reviewer can understand the scope.
A file that can be imported is not necessarily ready for every analytic. Missing information may limit which comparisons can be made.
02 / ANALYTICAL SIGNALS
A flag starts a question, not a conclusion.
An unusual amount, repeated transaction or cluster of purchases can have more than one explanation. A review candidate identifies a pattern worth examining; it does not establish fraud, non-compliance or an error.
Read the rule and its parameters alongside the source transactions. Consider business purpose, transaction timing, merchant context and the organization's applicable policies. Similar-looking records may reflect a legitimate recurring activity, a correction or a genuine issue that needs follow-up.
- Separate the observed pattern from the interpretation you are testing.
- Prioritize review using context as well as the analytical signal.
- Seek corroborating evidence and record plausible alternative explanations.
- Keep unresolved questions distinct from confirmed conclusions.
03 / REVIEW DOCUMENTATION
Make the reasoning traceable.
A useful review record connects the analytical signal to the evidence and the eventual disposition. Another reviewer should be able to follow what was examined, what was learned and why a conclusion was reached.
- Reference the relevant source transactions and the analytic that generated the candidate.
- Record the parameters and scope used for the analysis.
- Distinguish facts, explanations received, assumptions and outstanding questions.
- Identify the reviewer, follow-up actions and review status.
- Record the rationale when a candidate is explained or requires further work.
An exported dashboard can summarize activity, but it does not replace supporting evidence or the reviewer's documentation.
04 / DATA HANDLING
Use only the data you need.
Purchasing-card exports can contain sensitive organizational and personal information. Follow your organization's approved handling, access and retention requirements. Local analysis does not remove the need to secure the computer and any generated reports.
Keep working files in an approved location, restrict access to people who need it and use sanitized examples when asking for support. Do not send confidential transaction datasets through ordinary email.
This material is general educational guidance, not a substitute for your organization's methodology, professional judgment or applicable requirements.
DOCUMENTATION IS DIFFERENT
Guidance for the product itself.
Product documentation explains setup, supported fields, workflows, audit tests and troubleshooting. Educational resources explain the review approach and reasoning around the analysis.
A public documentation library is not currently available. The PCard product page provides a verified workflow and capability overview without offering a software download.
Explore the PCard workflow →
