Another fintech investment to chalk up in London: Passion Capital has invested an undisclosed level of seed funding in Ravelin, an early stage online fraud prevention startup formed by a team of ex-Hailo employees in January this year. The startup is currently working out of Passion’s co-working space, White Bear Yard.
Ravelin’s founders had been involved in fraud prevention at the taxi app, among other roles, where they came up with the original idea to spin out their own business. The seed funding will be used to get their platform to market later this year.
Passion’s average seed round investment — last we crunched the numbers — was around £187,300. TechCrunch understands that in Ravelin’s case the VC is investing above that average. Albeit the sums involved here are clearly modest at this nascent stage. Ravelin is currently working with a small group of beta customers as it develops the platform for a full launch, pegged for Q4.
“We’re starting beta development with our four confirmed beta partners. We have a long list which we’re introducing gradually so as not to swamp our development team. And we expect by the end of summer to have a working product that’s in production with those guys, and then we’re going to go through a hardening process with the actual product and build up our marketing and sales team before we take it to market at the end of the year,” says co-founder and CEO Martin Sweeney.
Ravelin is competing with well-financed online fraud prevention startups, such as U.S.-based Sift Science, which has pulled in some $23 million in VC funding thus far. However it’s not just a less flush U.K. clone. The spin here is it’s not taking a pure-play machine learning route to tackle online card fraud. Rather it’s setting itself up as hybrid platform that blends human agency and machine learning smarts — looping in some algorithmic fraud detection but also not requiring businesses to abandon existing human-oversight and processes. So it’s a stepping stone sales pitch, for those not ready to fully embrace robotic overlords.
“If you look around in the market you’ve got the established [fraud prevention] players, who’ve been around for 10 or 15 years, who’ve all been bought by big payment processors like AmEx or Visa, and then at the other end of the scale you’ve got new entrants who are doing very much pure-play machine learning. And what we’ve found by talking to people — real world companies — is that neither are particularly serving their needs,” Sweeney tells TechCrunch.
“It’s very hard to go into a company and say ‘look guys, all of that expertise you’ve built up over your lifetime as a company is no longer relevant and your team’s experience is defunct because this new technology replaces it. But also it’s quite hard to go in and say the same tools you’ve been using for the past 10 years are going to be up to date with the changing face of fraud.”
Omnichannel retailing means mobile commerce is proliferating, and more and more browsers and payment methods are being looped into the mix. It’s generally an increasingly complex landscape to monitor when it comes to fraud prevention. So that means human oversight is being strained — but that does not mean it should be replaced entirely, argues Sweeney.
“The old stuff doesn’t work, but the new stuff doesn’t address that fact that these people know what they are doing. So we’re aiming to sit directly in between those two positions,” he adds.
What does that mean in practice? Ravelin will offer access to an aggregated dashboard for fraud analysts that’s pitched as easier to use and more comprehensive than existing offerings, and which also leverages machine learning and social graph information to aid fraud detection — but as a supplementary signal which the humans who oversee fraud prevention in-house can choose how they respond to.
“We’re really trying to increase the efficiency and empower the human beings who are the core of the fraud prevention set up in any established company,” he continues. “What we try to do is to use new technology, including machine learning and social graph and graph network analysis to make their lives easier and make their jobs more efficient.
“These guys already have dashboards that they use, but they tend to be quite hard to use. And they have disparate sources of information. So we’re combining all of that information into once place in the dashboard for the analysts to use. We also provide ways for usually the manager of those analysts, the actual head of fraud, to keep on top of the numbers. To understand exactly what’s happening. To have a lot more control over what you would in the industry call a false positive.”
Sweeney says this approach allows for businesses to set their own risk appetite and continue making their own decisions on whether they want to review all transactions or let them all go through automatically.
“They can still define the rules themselves for what makes fraud, rather than relying on the machine learning — which does the easy stuff but doesn’t take into account all of their expertise. So it’s combining machine learning with some of the more traditional fraud techniques that these people are used to,” he adds.
“People don’t really trust [AI]. It’s not being around long enough to say I trust my job and my responsibility to this computer. They still want some oversight and what we’re doing is we’re aiming to give them all the benefits of these new technologies but without making it seem and feel too risky. We don’t want to make them feel like their jobs are at risk, and they’re putting their teams out of business. We’re just making their jobs more efficient.”
Ravelin will be targeting the platform at SMEs — or “large scale companies who are tech savvy”, as Sweeney puts it. Its business model is software as a service, following the established industry model of charging a per transaction fee, with the exact level based on volumes, albeit aiming to slightly undercut the industry average.
It’s using the new seed financing to fund its beta development process, likely to last six months, shaping the platform in conjunction with its first set of customers to ensure it hits a spectrum of ecommerce businesses’ needs.
“Having come from the taxi industry, we know everything there is about fraud in the taxi industry but not so much about for instance gaming, gambling, retail and all the other different arenas that ecommerce touches these days,” adds Sweeney. “Not just people who sell goods, but people who sell services as well. Because that’s an increasing area that fraud is touching.
“People who sell services are just as exposed these days because of the prevalence of stolen credit card data. And the quality of identity information you can buy on the dark net. Because of all the leaks that have happened, and all the hacking attacks — even the ones you don’t hear about. There’s a huge amount of high quality information out there which is just really hard to track.”