Case Study - Peer-to-Peer Marketplace

About Them

In a peer-to-peer platform, onboarding is everything. It’s the moment you welcome new users. But it’s also the moment fraud can sneak in. This mobile resale marketplace saw increasing attempts to abuse the signup process using disposable or masked email addresses. Some were bots. Others were setting up accounts with the intent to exploit features meant to protect real users.

They needed a way to spot risk early, and email turned out to be their most reliable clue.

The Challenge

Fraud wasn’t always obvious. In many cases, fake accounts appeared legitimate until they began exploiting protections or creating multiple profiles with lookalike emails. Key issues included:

• Lookalike emails with slight tweaks or patterns like added characters or numbers
• Brand-new email addresses with no history
• Aliases routing to the same inbox
• An overwhelming burden on manual review teams

With minimal signals at signup, the team needed a better way to distinguish trusted users from bad actors in real-time.

The Solution

The marketplace implemented AtData’s Fraud Prevention API, feeding in just an email address to find high-risk indicators instantly. What made it effective?

• Email Age & Longevity: Email addresses seen long ago are more trustworthy. Those created recently were seen to be 3x more likely to be fraudulent.
• Email Velocity & Popularity: AtData processes billions of activity events per month making it possible to assess the normalcy of email activity patterns.
• Tumbling Detection: Fraudsters often manipulate email formatting (e.g., j.o.h.n@, john+promo@) to create many accounts. AtData’s tumbling check identified these variations for quick resolution.
• Email Validation: The platform was able to block invalid, risky, or toxic emails in real time.
• Domain Risk Profiling: Recently registered or low-traffic domains are more likely to be used for fraud. Domain age, popularity, and activity velocity were evaluated for each signup.
• Risk Scoring: Powered by machine learning and AtData’s vast network, the score helped automate decisioning without needing access to a full identity graph.

The Results

While exact figures remain confidential, results were backed by well-documented risk patterns and AtData’s internal benchmarks:

• Newly created email addresses were flagged and deprioritized during onboarding
• They spotted lookalike emails that slipped past basic checks like the same inbox disguised with dots or extra characters.
• Fraudulent behavior was detected using metadata tied to billions of active and historical email addresses.
• Integration was seamless and fast, with API access that required only an email address but optionally supports additional data for more complex use cases and future scaling.

And importantly:

• Friction stayed low for trusted users thanks to lightweight, real-time checks.
• Manual review requirements dropped, freeing up the Trust & Safety team to focus on more nuanced cases.

Conclusion

Email is a behavioral fingerprint. With billions of data points behind each address, this marketplace could finally distinguish real users from bad ones at the point of entry.

By layering signals like first-seen date, popularity, domain history, and tumbling detection, AtData delivered fraud insights that weren’t visible anywhere else, no cookies or third-party trackers required.

Looking to Reduce Risk at Onboarding?

AtData helps platforms prevent fraud using signals you already collect, starting with the email address. With 99% global recognition and billions of behavioral events processed monthly, our network helps you spot fraud sooner and cost effectively to act with confidence.