How to Cancel Affiliate Commissions for Fraud

Fraudulent affiliates cost brands between 5% and 15% of their total affiliate programme revenue in our experience. Cancelling those commissions is how you recover it. But most affiliate managers either don't know the process or hold back out of fear of damaging relationships.

This article covers how to build a solid evidence case for a reversal, how to handle pushback from networks and sub-networks, and why being hesitant usually makes the problem worse.

Why Most Brands Don't Cancel (And Why That's the Wrong Call)

The most common reason brands don't cancel fraudulent commissions isn't that they can't. It's that they won't. They're worried about upsetting affiliates, creating disputes, or losing good partners from their programme.

That hesitation is understandable; relationships are key in the affiliate industry. It's also what keeps fraud happening.

When fraudulent affiliates see that a brand responds to complaints with accommodation rather than firm action, they continue. And they often escalate. The brands running the cleanest programmes are the ones that act quickly and build a reputation for not tolerating rule-breaking. The fraudsters move on to softer targets.

There's a practical time constraint that makes hesitation expensive, too. Most networks hold commissions for 30 to 60 days before paying out. After that comes "action locking," the point at which reversals are no longer possible. This typically happens one month after the end of the month in which the sale occurred. Miss that window, and the commission is gone permanently.

How to Build the Evidence Case

You can't reverse a commission without evidence. Networks won't often accept "we believe this affiliate is fraudulent" with no other evidence. You need to show them the data.

The most reliable signals come from your transaction data. Download your last 30 days of conversion reports from your affiliate platform and look for these patterns:

Identical click timestamps. Fraudsters generate click IDs in batches, at a specific time, on a specific device. Real customers don't all click at the exact same second. If you're seeing dozens of conversions where the click-through time matches to the second, that's not a coincidence. It's a tell.

Click device vs transaction device mismatches. The click IDs fraudsters generate originate from one machine, typically a PC. But the actual buyers are on iPhones, Androids, iPads. A pattern where every click device shows as "PC Windows" while transaction devices are all mobile is a strong signal of manufactured attribution.

Duplicate click IDs. A legitimate click ID appears once. If the same ID appears across multiple transactions, it's been generated and attached to multiple users.

Duplicate sale times. Clusters of transactions at identical timestamps don't happen organically. They suggest automated generation.

Unbelievable conversion rates. Often, fraudulent activity converts a conversion rate that simply isn't believable based on the type of activity that was supposedly being run. E.g. an influencer converting at ten times the rate of any other influencer activity. These conversion rates are a massive red flag and can help build evidence  

Evidence within a tracking tool. One of Marcode's main jobs is to collect evidence so that our clients can save commission. When you use a tracking platform, it will provide the definitive proof needed to show that the publisher was running fraudulent activity. It is still worth doing all of the above because a tracking tool like Marcode might have a subset of fraud, but not all of it

Sort your export by affiliate ID and timestamp. The patterns become obvious quickly. Once documented, you have the foundation for a reversal request.

The Sub-Network Problem

Here's where it gets complicated, and where most guides stop.

If the fraudulent affiliate is operating through a sub-network, you're partially relying on that sub-network to honestly report which transactions were fraudulently generated. That's a problem.

Sub-networks often have a financial incentive to minimise the scale of the issue. We see this regularly: the sub-network says 3% of transactions were affected. Based on the data, it's often significantly more.

The root cause is a structural gap in attribution. Some networks, like Awin with their click ref system, require transactions to carry an identifier that ties back to the specific affiliate within the sub-network. Many don't. Without that identifier, you can't independently verify which transactions came from the fraudulent source. You're dependent on the sub-network's self-reporting.

When we're sceptical about their numbers, we look at the broader picture:

• Timeline of the fraudulent activity. Identify when it started and stopped, and assess the total affected period against what the sub-network is claiming.
• Revenue behaviour when the activity stopped. If programme revenue dropped sharply, or went close to zero, when the fraudulent affiliate was removed, that tells you something about the real scale.
• Transaction anomalies across the disputed period. Duplicate click IDs and device mismatches are consistent and verifiable. They don't change based on who's reporting.

This approach won't always recover every fraudulent commission. But it builds a far stronger case than accepting the sub-network's version of events at face value.

How to Submit the Reversal

Once you have your evidence, the process varies by network.

Most affiliate platforms let you request commission reversals through their dashboard. You submit the order IDs, the reason for reversal, and supporting documentation. The network handles the technical reversal and notifies the affiliate. Some have dispute processes where affiliates can challenge reversals, which is exactly why your documentation needs to be solid before you submit.

For networks with well-built APIs, such as Partnerize and Rewardful, it's possible to push reversals programmatically rather than through manual submissions. This makes a difference at scale, particularly when dealing with a large volume of affected transactions.

What to include in your reversal submission:

• The specific order IDs you're reversing
• The fraud pattern, with evidence (duplicate click times, device mismatches, duplicate click IDs)
• The date range of the fraudulent activity
• Any prior communication with the affiliate

Keep everything documented. If the affiliate disputes the reversal, you'll need a clear paper trail.

When Affiliates Push Back

Some affiliates accept the reversal. Others won't, particularly when the amounts are significant.

The ones most likely to push back are often the ones who know exactly what they've been doing. Apply appropriate scepticism to their explanations. "There was a technical error with our tracking" is a standard response when patterns show hundreds of identical click timestamps. Technical errors don't produce data that precise.

For cases where the amounts are large or the dispute becomes aggressive, Marcode can generate legal documentation supporting the reversal claim. A formal document outlining the fraudulent activity and the evidential basis for the reversal changes the conversation considerably.

The key is not letting the pressure of a dispute cause you to back down from a legitimate reversal. Evidence-backed cancellations are defensible. Letting fraud continue because you want to avoid confrontation is not a sustainable programme management strategy.

What We See at Marcode

The pattern we see across clients is consistent: the brands losing the most to fraud are the ones who've been too cautious about taking action.

One client came to us after years of being reluctant to cancel commissions. They were worried about network relationships and affiliate disputes. By the time we audited the programme, they were losing over £500,000 a year to fraudulent commissions. That level of fraud had persisted for so long because the fraudsters knew the brand wouldn't push back hard.

The turnaround comes when affiliate managers shift their approach. The most effective ones treat their programme with a healthy degree of scepticism, act quickly when patterns emerge, and communicate reversals clearly and professionally. They end up with programmes full of good affiliates, partly because the bad ones know they'll be found and removed.

Fraud doesn't go away when you ignore it. It gets more confident.

Key Takeaways

• Fraudulent commissions typically represent 5% to 15% of programme revenue; acting quickly is the only way to recover them before action locking windows close
• The strongest evidence comes from transaction data: duplicate click timestamps, click/transaction device mismatches, and repeated click IDs
• Sub-networks often underreport the scale of fraud; use timeline data and revenue behaviour to build an independent picture
• The affiliate managers running the cleanest programmes take a firm hand, act on patterns early, and don't back down from evidence-based reversals
• Hesitating to cancel fraudulent commissions doesn't protect affiliate relationships. It prolongs fraud

If this is a problem in your programme, Marcode can help identify affected transactions and manage the reversal process.