Buyer to seller. Driver to rider. Courier to consumer. Multi-sided platforms like digital marketplaces and gig economy apps are a part of everyday life, especially after the pandemic. These platforms have the power to facilitate connections between different types of people in a way that benefits everyone—connecting someone who needs a good or service with someone who can provide it. When things go well, everyone benefits: the consumer side, the seller or worker side, and the platform itself.
Unfortunately, wherever thriving industries go, fraudsters follow. Fraudsters and other bad actors throw the balance of multi-sided platforms off by taking advantage of one or more parties. A bad actor on a food delivery app might use a consumer-side account to abuse promotions. A bad host on a vacation rental platform might defraud consumers by falsely advertising. And so on.
These types of users are bad for the platform and all of its good users. Platforms maintain the right to ban or suspend users to protect their product’s integrity, but unfortunately, banning a fraudster doesn’t always mean you’ve actually seen the end of them.
You can’t always predict when someone is going to defraud your platform for the first time. Sometimes, like with fake marketplace listings, the fraudster’s journey mimics the user’s journey perfectly until the fraud happens. Other times, like with some courier social engineering scams we’ve seen, the fraudulent interaction happens off of the app, making it hard for the app to predict or police.
As a fraud prevention stakeholder, what you do have control over is whether fraud and abuse becomes systematic.
Fraudsters don’t make significant money off of a single instance of fraud or a single scam. Their real bread and butter is in repeat offenses. When a bad actor can build a system that allows fraud over and over again, their profits—and their negative effects on the platform—can grow exponentially.
Ban evasion is what allows fraud on multi-sided platforms to reach that kind of scale. Obviously, a bad seller-side account will be reported and banned if it commits multiple infractions. But when fraudsters can successfully make a new account and ban evade, they can keep the scheme going indefinitely.
Ban evasion on multi-sided platforms means:
Incognia is no stranger to the challenges of ban evasion and multi-accounting. Our solution relies on taking away fraudsters’ abilities to mask their identity when they return to a platform.
For example, our device fingerprinting solution recognizes return users based on the device they use. If fraudsters try to factory reset or switch devices, we also rely on a location intelligence layer to assess the risk that a seemingly new device is actually a user we’ve seen before.
As Andre described in the panel, this layered approach is how one of our clients was able to achieve a 97% reduction in a specific courier scam involving a tampered Point-of-Sale machine and some social engineering.
Using a persistent device fingerprint, the platform was able to keep the bad couriers blocked permanently, ending the ban evasion that enabled the scam. “It was a very difficult challenge because the food delivery platform didn't have visibility about what was going on. For them, the only thing they were seeing was that the order got canceled by the courier, and they had to refund the customer, but everything else was happening offline. The platform wasn't seeing what was going on. This is a classic ban evasion issue, because if that issue is reported by the customer, you're probably going to investigate and then eventually ban the courier, but they may come back, right? Because they're making a lot of money. So they're going to find ways to rent someone else's account or purchase an account or open a new account using fake credentials.”
Using a persistent device fingerprint, the platform was able to keep the bad couriers blocked permanently, ending the ban evasion that enabled the scam.
Fraudsters do what they do to make a profit. As Garrett shared during the panel, using transaction monitoring and geolocation together to follow the money can lead to powerful insights about who’s committing fraud on your platform and how:
“Ultimately on the courier side, where we're predominantly seeing any type of ban evasion or multiple accounts is from the financial trail. So we spend an exorbitant amount of time monitoring the transactions and the frequency. From a geolocation perspective, where are these anomalies popping up?”
From there, it’s a manual investigation. Addressing each case manually helps Wolt gain a keener understanding of the unique ban evasion challenge in each country where they operate.
Understanding fraudster behavior and using that to inform strategy is another keen way to limit ban evasion. At Vinted, for instance, Aurelija mentioned that they have eyes on different Telegram groups used by fraudsters, so they have a better understanding of how they operate and what exactly they know.
Knowing that fraudsters test for vulnerabilities constantly and act immediately when they find them, Vinted uses statistical analysis of new registration numbers of one of its tells that fraudsters might be trying something new. As she explains,
“That's where we're investing our effort at the moment. The way we measure it is we calculate abnormal amounts of registrations, meaning if you look at it not from a single account perspective, but from a mass statistical perspective, there is an inflow that you can't explain. So the current focus is exactly on that.”
Fraudsters don’t just give up and go away after a ban, and they’re not likely to give up just because we’ve found successful anti-ban evasion tactics.
Like other areas of fraud prevention, it’s a cat and mouse game, and fraudsters will be hard at work looking for new ways to compromise our signals and circumvent their bans.
That’s why constant innovation—and learning from the rest of the industry—is so important. When we share knowledge and insights like those above with each other, we see more of the full picture of ban evasion today.
For more information about the anti-ban evasion cases Incognia has seen in the past, read our food delivery case study here.