Sequel's UI Tweak Slashes Fraud by 30%. The Lesson? Stop Teaching Scammers.

A dating platform discovered its anti-fraud systems were accidentally running a training programme for scammers. By simply removing the notification that told users their accounts had been suspended, Sequel cut fraudulent registrations by over 30% overnight. The finding reveals an uncomfortable truth: transparency designed to help legitimate users can arm the exact actors it's meant to deter.

This wasn't a case of insufficient detection or inadequate moderation resources. Sequel's systems were catching bad actors effectively. The problem was what happened next—a suspended account message that gave scammers immediate confirmation they'd tripped a detection rule, paired with tacit instruction to adjust tactics and try again.

When transparency becomes a training manual

The mechanics are straightforward. Sequel's moderation systems were flagging suspect accounts and applying restrictions. When those accounts attempted to access the platform, they received a suspension notice on the preview screen. For a legitimate user mistakenly flagged, that's useful feedback. For a scammer running multiple registration attempts, it's something else entirely: immediate confirmation that the account tripped a detection rule.

Create a free account

Unlock unlimited access and get the weekly briefing delivered to your inbox.

Register free

No spam. No password. We'll send a one-time link to confirm your email.

According to Sequel, this is precisely what happened. The company's trust team observed fraudulent accounts returning with iterative improvements—tweaked profile copy, different photos, altered registration patterns. The suspended account message wasn't deterring fraud. It was calibrating it.

Removing the notification severed that feedback loop. Scammers no longer receive instant confirmation that their accounts have been detected.

They're left guessing whether a profile was rejected, whether it's under review, or whether it simply failed to load. That uncertainty, Sequel's data suggests, is worth a 30% reduction in fraudulent registrations on its own.

The broader implication here cuts across the industry. Dating platforms have spent years investing in transparency for members—clearer community guidelines, more visible reporting tools, better communication about moderation decisions. That's appropriate for legitimate users. But when applied indiscriminately to all accounts, transparency can arm the exact actors it's meant to deter.

Hardware authentication and risk scoring at scale

The UI change was paired with more technical interventions. Sequel has deployed what it describes as a multi-layered hardware authentication system designed to differentiate genuine mobile devices from virtual machines, emulators, and bot networks. The company disclosed plans to add network-based checks and mobile carrier validation to flag location spoofing.

The platform has also expanded its dynamic risk scoring, analysing factors including network behaviour, referral sources, and uploaded media to surface suspicious patterns in real time. Duplicate profile photos, recycled biographies, and other shared characteristics are now cross-referenced against known scam networks, allowing the platform to block coordinated spam campaigns faster. Sequel claims its systems can also detect AI-generated content used to evade profile verification, though it did not specify the technical methodology or accuracy rates.

These are familiar tools in the broader trust and safety toolkit—risk scoring, device fingerprinting, media hashing—but their effectiveness depends on execution and scale. Sequel positions itself as a niche offering focused on "intentional dating", which raises the question of whether a smaller, more homogenous user base makes these techniques more effective than they might be on larger, more diverse platforms.

That doesn't diminish the finding about UI feedback loops—that's platform-agnostic. But it does matter when assessing whether Sequel's broader technical approach offers a blueprint for operators at Match Group scale or whether it's more applicable to emerging platforms still defining their moderation posture.

What this means for the wider fraud fight

Romance scams continue to intensify across the industry. Figures from the US Federal Trade Commission show romance scam losses reached $1.14B in 2023, the highest of any fraud category tracked. Scammers are increasingly deploying automation, AI-generated imagery, and coordinated bot networks to scale operations.

The dating industry's response has largely focused on detection volume—more AI, more moderators, faster flagging. Sequel's case study suggests a parallel track: reducing the information advantage platforms inadvertently hand to bad actors.

If a single suspension notice can function as a training signal, what about password reset flows that confirm whether an email is registered? Or reporting mechanisms that reveal how many reports trigger action?

The shift from reactive moderation—ban the account—to preventative design—don't confirm the ban—represents a maturation in how platforms think about adversarial users. It assumes bad actors are adaptive, not static. It treats moderation systems as potential attack surfaces, not just enforcement tools.

Whether other operators adopt similar approaches will depend partly on member experience trade-offs. Legitimate users value transparency about moderation decisions, particularly when they're the subject of them. Stripping out all confirmation messages risks frustrating real members caught in false positives. The design challenge is building systems that withhold information from bad actors without abandoning clarity for everyone else.

Sequel's 30% reduction is self-reported and covers a short initial timeframe—the company disclosed the drop occurred "the following day" and has been sustained since, but didn't specify how long "since" represents or whether scammers have adapted. Still, the finding is specific enough to be falsifiable and the logic is sound. If nothing else, it's a reminder that the industry's fraud problem isn't just about better AI or more moderators. Sometimes it's about shutting up and not telling scammers what they got wrong.