Geva Solomonovich at Palo Alto BB November 4

Francis Adanza interviewed Geva Solomonovich, an early employee at Fraud Sciences which was acquired by Paypal for 170M.

Francis noted:

Geva joined the company while it was still a startup and played a significant role in helping the company grow. I thought the Bootstrappers Breakfast community might benefit from Geva’s lessons learned. In addition to this Q&A blog post, Geva will be joining the round table discussion as a featured guest speaker on Friday, November 4 at the Palo Alto Bootstrappers Breakfast. Come join us and ask Geva your own questions.

BB: Hi Geva – Thanks for taking the time to share your entrepreneurial insight with the Bootstrappers Breakfast community.

Geva: Hi Francis, glad to be here.

BB: Prior to Fraud Sciences, what were you doing?

Geva: Before Fraud Sciences I worked in several other startups. One was SofaWare which built small office Firewall and VPN appliances. SofaWare was partnered with CheckPoint and the initial code base was derived from CheckPoint’s Firewall code. Another company was Savantis – a pioneer in Database Virtualization systems. Savantis closed down after 2.5 years. I learned a valuable lesson there – it’s almost impossible to sell datacenter critical solutions to large enterprises when you are a small startup. Duh. Even though the startup didn’t succeed, it was a great experience, and I was even able to get an algorithm I built filed for patent.

BB: What did Fraud Sciences do?

Geva: Fraud Sciences did e-commerce credit card fraud analysis. E-commerce merchants would send us the details buyers provided in the checkout page (name, address, credit card number, etc.), we would analyze the transaction information, and reply to the merchant with a green or red light. The interesting part of course was how we were able to make good decisions… this was a combination of outstanding technology and outstanding fraud analysts who understood and defined legitimate and fraud behavioral patterns. Let me give an example of what I mean by outstanding technology: in a typical e-commerce situation, an American credit card is presented at a checkout page of an American commerce site like amazon.com. A quick geo location checkup shows that the buyer’s IP address is actually in Romania. Most fraud systems would automatically decline such a transaction as the IP geo mismatch here is severe. Our technology enabled us to automatically analyze the buyer to see (and confirm) he works in an international corporation which has a branch in Romania. All of a sudden, the story looks completely different and some of the red lights dim.

Overall, our technology was so great we offered 100% fraud chargeback guarantee. If we “approved” a transaction and it was later chargedback, we covered 100% of the item cost. It was that simple. The value prop for merchants was outstanding – they could now sell to new risky markets they were avoiding before. Think about the ability to ship a $10k diamond ring to Russia without having to worry.

BB: How big was the company when you joined?

Geva: I joined the company when we were still sharing half a floor of a small building with 3 other startups. I was the fifth employee at the time with one other developer, an analyst and the two founders. Seeing the company grew from this early stage to being acquired has been really inspirational.

BB: What were some the early sales/marketing challenges you faced during the first couple of years in business?

Geva: The biggest challenge is getting through the first door. Online merchants are pretty busy people and they get a lot of offers from startups to fix/change/improve/solve/expand their business. Then comes the question of trust:  who is this small startup that wants us to share our checkout information? How can I trust their recommendations to approve/decline a transaction? “I have my own fraud management system…”

BB: What did the company do to overcome these challenges?

Geva: To overcome these challenges we started selling to very eccentric and non-mainstream merchants – people who really have high-fraud and funky businesses. Some of our initial merchants were selling anonymous web-browsing services. As you can image that attracts a lot of fraudsters who want to be anonymous on the web. We actually saw that fraudsters are using the anonymizer services to fraudulently buy more subscriptions from them! Another genre of merchants we integrated with were e-gold brokers. E-gold is like a virtual currency – where there is money there is always fraud. After we had a proven record with these types of merchants we continuously worked to revisit and improve our offering. We simplified the integration process to a 10 minute exercise. We simplified the feedback to the merchants to a simple Yes/No. We slowly grew to the 100% fraud chargeback guarantee. In parallel, the sales force continued improving their sales pitch, and we worked very hard to get supportive feedback from top-notch merchants (like www.ice.com) that we used in order to reduce new merchants objections.

BB: As the company grew, what were some of the operational challenges you faced as you gained traction?

Geva: One of the biggest challenges we had was scale. For a long time our analysis process was 100% manual human review. Technology played a big role by scanning through 1000s of data sources, scraping the web, running sophisticated algorithms and aggregating data to make the decision process easier for the human analysts. But it was still a human analyst who made the final decision. Hiring these human analysts wasn’t scalable either – these were top notch, super intelligent, high SAT, highly analytical kind of people. There is no standard job posting nor school education that you can look for.

By far, my biggest contribution to the company was in building the automated decision platform. My team and I spent hours interviewing our human analysts to try to quantify and frame their thought process into something we can let a computer do. We studied machine learning algorithms, data mining processes, statistics etc. and then started implementing. In 3 months we had the first version out. This version outperformed most of our human analysts, and of course worked much faster. It was a great success…

BB: How long was the company in business before you raised venture capital?

Geva: The founders worked 3-4 years by themselves without raising venture capital. They then raised a small series A funding before they started recruiting employees.

BB: What risk reducing milestones did the company achieve that merited risk capital investment?

Geva: On the business side, we worked very hard to have a repeatable and predictable sales cycle. That was of uttermost importance. New merchant acquisition had to be streamlined and efficient, while trying to reduce the paperwork as much as possible. On the technology side, we needed to show that we can grow our maximum transaction processing volume by x10, and then by another x10, and then to a point where to scale would only mean buying more servers.