Affiliate Reporting: Why Averaging Approval Rates Can Mislead

Two affiliate sources can each report an order approval rate, yet the arithmetic average of those percentages may not describe their combined traffic. A small source and a large source should not automatically carry equal weight in an order-level total.

For performance marketing media buying, this distinction matters when reviewing ecommerce CPS campaigns and digital-product offers. The calculation below is an analytical method with hypothetical numbers, not a BlueFriday performance benchmark or a description of any network's contractual reporting rules.

Define the denominator before comparing sources

Specify which records count as submitted orders, which statuses qualify as approved, and how pending, duplicate, canceled, or rejected records are handled. Use the same definition for every source in the comparison. Record the conversion period, status cutoff time, and offer scope.

Two useful measures answer different questions. Approved orders divided by all submitted orders describes the approved share of the submitted cohort at the chosen cutoff. Approved orders divided by decided orders describes the approval share among records with a final decision. If pending orders exist, show their count and do not give these two measures the same label.

Add the counts before dividing

Suppose Source A has 20 approved orders from 25 submitted orders, while Source B has 30 approved orders from 75 submitted orders. Assume all 100 orders have final decisions and use the same eligibility rules. These figures are entirely hypothetical.

• Source A approval rate: 20 / 25 = 80%.
• Source B approval rate: 30 / 75 = 40%.
• Combined order approval rate: (20 + 30) / (25 + 75) = 50%.
• Simple average of source percentages: (80% + 40%) / 2 = 60%.

The 60% figure gives each source equal weight, regardless of order count. It may answer a question about the average source-level rate, but it does not represent the approved share of all submitted orders. Label the measure according to the question it answers.

Keep counts and value on separate lines

An order-count approval rate does not reveal how much commission was approved. If commission values differ, build a separate value-based calculation with a defined currency, commission basis, and status cutoff. Do not weight an order-count measure by spend or commission and continue calling it an order approval rate.

For a reporting sheet, keep source, submitted count, approved count, rejected count, pending count, and cutoff timestamp in separate columns. Add commission fields only when their definitions are clear. Reconcile the status counts to the included order population before calculating a total.

Check maturity and composition

A recent source may have more undecided orders than an older source. Show equal-age cohorts where possible and distinguish preliminary from mature results. Where offers or markets have different rules, inspect those segments before interpreting a combined movement.

A changing source mix can move the total even when each source's own rate stays constant. Preserve both the source-level rates and their denominators so a reviewer can distinguish a composition change from a change within a source. Neither pattern alone proves fraud, traffic quality, or tracking failure.

Use the measure as a diagnostic input

Before changing budgets, examine conversion definitions, pending decisions, rejection explanations, approved commission, and acquisition cost together. Approval rate is one part of the evidence, not a standalone instruction to scale or stop an affiliate source.

Teams working across media buying and advertiser operations can share a report containing the raw counts, exact formula, cutoff, and unresolved questions. That makes the calculation reproducible and keeps a dashboard percentage from carrying more meaning than the underlying records support.