While retailers of late have worked hard to broaden the range of experiences they offer, there are some things that no reasonable person would expect merchants to do. The general expectation is that retailers will be experts in selling goods — and hopefully experts on the things they sell. But no one goes to a merchant with a serious expectation of finding an expert on taxes or payroll — even though paying taxes and employees are both things merchants do — because those kinds of functions are outsourced to experts.
But when it comes to transaction security, the situation most retailers are facing is a bit different.
“Merchants being liable for transaction fraud is just not appropriate,” Signifyd’s Vice President of Partnerships Skye Spear told PYMNTS in a recent conversation. “When it comes to the cost of fraud, merchants have pretty much just had to eat it.”
When merchants are held liable for fraud, it presents two significant challenges: (1) merchants did not go into business to become fraud experts, and (2) they face a formidable and persistent adversary in a global army of digital fraudsters.
But, Spear noted, merchants being required to become fraud experts is common practice given a traditional approach to fighting fraud:
- Merchants purchase a fraud-fighting tool from a vendor.
- They integrate it into their systems.
- They use the tool and hope it’s worth the time and money they invested.
- They remain completely liable for all fraud and chargebacks.
“If you buy a traditional fraud tool, there’s a cost associated with that. [Merchants] get that. Using the tool and running it also has a cost. But despite those costs there are still chargebacks and you’re losing legitimate orders you turned away due to the fear of chargebacks. For a merchant, that’s the total cost of fraud — everything that is spent and lost getting to that final point.”
Signifyd was convinced there should be a better way — one that decouples the merchant from the liability of fraud while eliminating the ever-fluctuating costs of fraud on the business.
“When a merchant uses Signifyd we will be held responsible for our decisions. Chargebacks go off their books and on to ours,” Spear said.
But how exactly can Signifyd eliminate chargebacks on behalf of merchants?
Only Paid When Payments Work; On The Hook When It Fails
Reducing chargebacks down to zero, Spear noted, is actually quite easy. But what makes Signifyd’s approach viable for the majority of their 5,000 merchants is the ability to eliminate chargebacks without impacting legitimate customers.
“As long as you have something people want, you have something people want to steal. That’s why we work in all verticals and all industries,” Spear said, noting that “some categories tend to see much more fraud than others, and these are the categories one might expect: electronics, apparel, high-end jewelry, watches.”
But what has been interesting to watch is the degree to which players of all scales and scopes are embracing the idea that they can outsource responsibility for fraud and chargebacks.
“Over the last 18 to 24 months, we’ve seen a move up-market. Larger companies that have been using big custom-built solutions are now realizing this doesn’t need to be a core part of their business.”
Some of the world’s largest retailers have deployed Signifyd, including some “ubiquitous goods sellers” that deal in everything from high-def TVs to toilet paper. With such sizeable companies adopting Guaranteed Fraud Protection, momentum is building towards moving these kinds of decisions to specialists and out of the core business function.
And in Signifyd’s case, these specialists are particularly incentivized to act in the merchant's interests.
“We only charge merchants for the payments we help them accept, we don’t charge anything on what we help them decline,” said Spear. “The only way for us to make money is to take as many orders as possible and keep the acceptance rate at a very high level. And the only way for us to stay in business and keep that money is to correctly identify and turn away fraud across all our merchants, because otherwise we’d be eating a lot of losses that can add up quickly.”
Responding To A Dynamic Market
Fraudsters don’t give up. When one type of fraud is identified, they either tweak it slightly to make it work again or simply try something else. Giving up just doesn’t seem to be in the cybercriminal’s playbook.
Given modern automation that fraudsters can leverage, this dynamic nature for fraud is a big problem for rule-based systems, said Spear, which require a tremendous amount of maintenance to stay current on new variations and iterations of fraud.
Signifyd’s software doesn’t work that way — instead, it uses artificial intelligence to “rewrite the rules in real time to create a dynamic tool for a dynamic threat. We’ve removed the static conditions that fraudsters rely on to know what will work for any particular merchant.”
This approach, Spear noted, is greatly aided by the fact that Signifyd pays merchants every time it gets a decision wrong.
“It’s expensive, sure, but I’m ‘paying’ for excellent data every time the system fails. In contrast, any organization that claims to be data-driven and says they’re leveraging machine learning without assuming liability for their decisions is going to have a very difficult time getting appropriate, real-time feedback for their models,” Spear said.
When things go wrong for a customer using a standard fraud tool, the merchant is still on the hook for the chargeback, so their motivation to report these errors to the tool maker is limited. The merchant will almost always pay their chargeback and move on. And if any one tool lets too many chargebacks through, the merchant provides “feedback” by simply pulling the plug.
“But since Signifyd promises to pay a merchant back every time we’re wrong, we are the first ones to be notified when there’s a chargeback,” Spear noted. “Merchants tell us right away because we’re paying them back for it. That sounds expensive, but we are ‘buying’ good data and our system is getting immediately smarter because we have a perfect feedback loop that lets our machine learning get sharper and sharper very quickly.”
Merchants aren’t fraud experts — nor should they have to be. Likewise, according to Spear, Signifyd has a lot of fraud experts who would not do well selling household goods or widgets online.
And since no one asks their fraud experts to sell widgets, Signifyd is working hard to return the favor — by getting merchants out of the business of being liable for fraud.