A conversation has been making the rounds in consumer rewards and channel incentive circles, and it goes like this: How much protection are you willing to trade for how much friction you’re willing to accept? The implication is that these two things exist on a scale, that more security means more disruption for participants or members and that the job of a program operator is to find a livable point somewhere in the middle.
That framing deserves scrutiny. Because if it’s true, it has real consequences for how programs get built and how participants get treated. And if it isn’t, the industry deserves a better conversation than the one it’s been having.
The position worth taking is that it’s a false choice. Not because fraud is easy to manage, but because the tradeoff isn’t inherent to the security. It’s a product of how security is designed.
Where the Framing Comes From
The critics of redemption-layer security aren’t wrong about the symptom. Blunt controls do create friction. A single detection layer calibrated to flag anything outside a narrow normal range will catch some bad actors and delay or deny many legitimate ones. A top-performing channel partner or loyal customer whose first seasonal redemption gets held without explanation notices. When that happens, a fraud control damages the relationship the program exists to build. That’s a real failure, and it deserves to be named as one.
It’s also largely avoidable. A participant who understands that some orders require a brief review, and who receives clear communication about what they can expect and when, has a fundamentally different experience than one left wondering what happened. The security measure itself isn’t the problem. The silence around it is.
But the cause of that failure isn’t the presence of security at the redemption layer. It’s the absence of enough layers. When the work of detection falls on a single control, that control has to be set broadly enough to catch what it might miss. Broad thresholds create friction. That’s not a security problem. That’s a design problem.
What Changes When Detection is Layered
Layering changes the math. When multiple controls work together, no single one has to cover everything, which means no single one has to be set broadly enough to cause damage.
Authentication stops most bad actors before they get anywhere near a reward. Redemption-layer scoring picks up what authentication never sees, the velocity patterns, address anomalies, device mismatches, behavioral signals that only appear when someone actually tries to redeem. And when a transaction lands in the middle range, trained reviewers step in where automation runs out of confidence. The patterns that score just below a threshold but carry enough softer signals to warrant a second look. That judgment call is where a lot of fraud gets stopped, and where a lot of legitimate participants get protected.
No single layer has to do all the work. Which means no single layer has to be set broadly enough to cause collateral damage to the participants or members your program was built around.
For the participants who aren’t doing anything wrong, which is the overwhelming majority, fraud scoring happens fast enough that there’s no perceptible delay. The protection runs in the background. The experience is unchanged. That’s not an aspiration. That’s what the right design produces.
The System Has to Stay Current
Good design at launch isn’t enough. Fraud tactics shift, and they shift faster than most programs update their controls. What looks like a novel attack pattern today becomes a common one within months. A detection system that isn’t actively adapting isn’t really protecting anything. It’s running on assumptions that may no longer be true.
The programs that stay ahead of this treat their fraud rules the way a good editorial team treats a publication: something that gets updated based on what’s actually happening, not set once and left alone. Rules change based on what reviewers are seeing. Signals that stop predicting get recalibrated. New patterns get incorporated before they become expensive. The system earns its keep by getting sharper over time, not by staying static.
The scale of what’s at stake becomes clearer when you look at what well-designed layered detection catches. Across years of redemption data and multiple program types, it stops millions of dollars in fraudulent orders, not because the controls are maximally aggressive, but because they’re layered, current and calibrated to each program’s specific risk profile rather than applied from a generic template.
What the Goal Is
Zero fraud is the wrong target. A zero-fraud rate is a sign that controls are set too aggressively. Tight enough to stop bad actors, but also tight enough to stop good ones. That’s not success. That’s the experience problem in a different form.
The goal is a fraud rate within the range the industry typically cites, between two and six percent, not because the program isn’t being targeted, but because the controls are working across the full participant journey without announcing themselves to the participants they’re protecting.
What’s at stake is the relationship between a brand and a participant who earned something. That relationship is what fraud at the redemption layer threatens. It’s also what well-designed protection preserves.
The Question That Is Always Worth Asking
Programs that get this right don’t talk about the tradeoff. They outgrew that conversation already.
The conversation about how much protection to trade for how much friction assumes the two move together. They don’t have to. What moves together is the quality of the design and the quality of both outcomes.
The question isn’t protection versus experience. It’s always whether the protection is designed well enough to deliver both.
