Affiliate Tracking Discrepancies: How to Reconcile Orders

An affiliate says they drove 40 orders. Your platform shows 27. Before anyone changes a commission rate or promises a make-good payment, find out whether both sides are counting the same thing.

I'd start with a small reconciliation: the placement, the reporting window, the order status and the evidence behind the missing transactions. That gives the affiliate manager a useful conversation to have with the partner and a specific question for the tracking team. A screenshot of two different totals is a starting point, not a diagnosis.

Put the disagreement into one working record

Consider a fictional tea retailer working with a gift newsletter. The publisher reports 40 orders associated with its November 2 promotion. The retailer's affiliate report shows 27 approved orders. The partner wants a higher commission to compensate for the difference.

The manager asks which dates, statuses and identifiers support the publisher's total. The first useful finding is that 40 is the publisher's initial estimate, while 27 is the retailer's approved count after its review process. Those totals answer different questions. The remaining work is to compare like with like, then investigate any unresolved orders.

A completed internal note might read:

• Placement: November 2 gift newsletter, approved tea sampler offer.
• Publisher claim: 40 estimated orders, November 2 through November 8.
• Program report: 27 approved orders in the same purchase-date window.
• Known differences: Five orders are still pending review; three were cancelled; two were outside the agreed offer conditions.
• Unresolved: Three claimed orders have not yet been matched to a program record.
• Owner and next update: Affiliate manager collects permitted references; tracking specialist checks the unmatched cases; partner receives an update November 11.

The arithmetic accounts for 40 claimed orders: 27 approved, five pending, three cancelled, two outside the conditions and three unresolved. It does not establish that every classification is correct. The manager still needs the evidence and the applicable agreement behind each decision.

The team now has five pending cases to finish and three unmatched cases to investigate before discussing a blanket increase for future orders. The partner gets a concrete update rather than a vague assurance that tracking is being checked.

Agree on what each number means

Compare the same purchase-date window and time zone. Ask whether a report uses purchase date, recording date or approval date. Confirm whether it includes pending transactions, cancellations, refunds, taxes, shipping or other amounts. Use the definitions in your actual platform and agreement; don't assume every report or network treats these fields alike.

Then separate counts from money. A difference in commission can come from the commission base or an applied rate even when the order count agrees. Record the order amount, eligible commission base, rate and status separately where your authorized tools allow it.

Keep an unchanged export or dated snapshot of the reports used in the review. If a pending transaction becomes approved tomorrow, you should be able to explain why today's count differs from yesterday's. Don't overwrite the original claim and lose the history of the conversation.

Investigate a few cases before making a program-wide change

For the three unmatched tea orders, start with the evidence you can legitimately access: the relevant promotion, destination, partner link, permitted transaction reference and the recorded time. Ask the tracking specialist to follow those particular cases through the actual setup.

A useful request is specific: “Please check these three permitted references against the November 2 newsletter placement and report which step can be verified, which cannot and what evidence is missing.” That is easier to resolve than “Affiliate tracking is broken.”

Test the current path through the approved testing procedure. Verify the actual link and destination, then check the events or records the implementation is supposed to produce. A working test today does not settle what happened to an earlier purchase. Keep the historical investigation and the current test as separate findings.

Use an authorized private channel for order-level evidence. Don't ask the publisher to post customer details in a shared campaign spreadsheet. Share only the references needed for the investigation, with the people authorized to handle them.

Give the partner a useful interim answer

The partner should know what has been resolved and when the next answer is coming. Here's a sample message for the fictional review:

Thanks for flagging the difference. We've aligned the November 2 through November 8 purchase window. Our current record has 27 approved orders, five pending review, three cancellations and two outside the agreed offer conditions. Three references remain unmatched and are with our tracking specialist. We'll update you by November 11 with the remaining findings and any correction required under the program terms. If your 40-order estimate uses a different window or status, please send that definition so we can compare the same basis.

Only send a message like that after checking the numbers and classifications. If the team hasn't established why an order was excluded, describe it as under review. A deadline for an update is something the manager can own; a guaranteed outcome before the investigation is finished is a different commitment.

Separate tracking corrections from future commission terms

If the review finds a documented error, follow the program's authorized correction process and record the affected transactions, reason and approval. If evidence remains incomplete, say exactly what remains unresolved. Don't close a case simply because a dashboard total changed.

Discuss future commission terms on their own merits: the partner's planned work, customer value, expected costs and what the business can support. An unexplained historical gap is a poor foundation for a permanent rate decision. A rate increase also doesn't repair a faulty link or an incorrect commission base.

For a paid placement, use the same evidence discipline in your placement test plan. Later, use the renewal review to decide whether the next booking makes commercial sense.

Finish with a record somebody else can follow

Close the review with the original claim, agreed definitions, resolved cases, unresolved cases, any approved adjustment and the next owner. Include the date and the evidence location. Another manager should be able to understand the decision without reconstructing a week of chat messages.

Start with one current discrepancy and complete that record before expanding the exercise. If the same issue appears across partners, that pattern gives the technical team a stronger investigation brief.

Experience Advertising helps businesses manage affiliate programs, including partner communication and performance review. Contact us if your team needs help turning recurring reporting disagreements into a manageable review process. For finding prospective partners, AffiliateFinders supports discovery and contact research; it doesn't resolve transaction attribution. Evan Weber founded both businesses.