Affiliate Reporting: Check Pagination Before Trusting an Export Total

A report can contain valid rows and still be incomplete. In affiliate media buying, a downloaded file or copied table may represent only part of a result set. Before using it to compare ecommerce CPS or digital-product campaigns, establish what the export includes and whether every required page was retrieved.

The workflow below is a proposed completeness check. It makes no assumption about a particular network's reporting interface or whether that interface paginates exports.

Record the report contract

Write down the account, reporting period, time zone, filters, selected columns, and row definition. Determine from the interface or documentation whether the export covers all matching rows, the current page, or a separate export limit. If the scope is unclear, mark it unresolved rather than assuming the file is complete.

A screen can show a limited preview while an export includes everything, or the two can have different scopes. Establish their definitions before treating a difference as an error. Keep the original downloaded files unchanged so a later reconciliation can be reproduced.

Track page boundaries explicitly

When retrieval does require multiple pages, record the page number or continuation token, row count, and first and last row identifiers for each batch. Use a stable sort where the system supports one. If results can change during retrieval, prefer a fixed reporting snapshot or closed period where available; otherwise note that rows may move between page boundaries.

• Confirm that each required page or continuation step was requested.
• Check that retrieval ended according to the system's documented completion signal.
• Look for repeated row identifiers across adjacent pages.
• Investigate gaps against a trustworthy expected count or independent control total.
• Keep legitimate repeated events when the row definition allows them; do not deduplicate on campaign name alone.

Use counts as evidence, not a guarantee

In a hypothetical report showing 245 matching rows with 100 rows per page, a complete retrieval would contain batches of 100, 100, and 45 if the result set stays fixed. Two full batches contain 200 rows, leaving 45 unaccounted for. A file that stops at 200 rows is therefore incomplete under those assumptions even if its commission total looks plausible.

Matching 245 rows is necessary in this example but is not sufficient: repeated rows could replace missing ones. Compare stable identifiers at the report's actual level of detail and reconcile a suitable control total using identical filters, currency, status, and reporting period. A count mismatch identifies a question to resolve; it does not by itself identify which rows are wrong.

Keep incomplete data out of allocation decisions

Label the dataset complete, partial, or unresolved, and retain the evidence supporting that label. If a retrieval fails halfway through, do not present the partial sum as a full-period total. Do not scale spend from an apparent campaign ranking until the missing portion has been understood.

Teams responsible for performance marketing media buying can add the completeness label to their regular reporting handoff. Advertisers and affiliate partners then have a precise diagnostic request: the report settings, expected scope, retrieved pages, and discrepancy. This turns a vague disagreement about revenue into a checkable question about the dataset behind it.