OPW INTERVIEW - Oct 11 - This is a new strand of interviews called Service Spotlights for companies that serve the dating industry. We're pleased to start off with the CEO of Intro Analytics, Nick Tsinonis. - Mark Brooks
What does Intro Analytics do?
It’s a data intelligence matching company designed for dating sites and social media. We look at the behavior people have online and we match them based on that.
Back in 2004 when I was working with Cupid.com, I had the opportunity to visit Professor Dan Ariely. The key takeaway I had from that meeting was that peoples behaviors give you a better indication of who they are and what they are looking for. That’s why I’ve been keen to speak to you.
Who’s the brains of Intro Analytics and what’s your founding story?
The brains are myself and Gavin Potter. I started a website in 2006 called YesNoMayB. I was going to revolutionize the online dating industry by matching and analyzing your taste without you having to do much. I realized that it was really hard to have the science to match people like that. So I approached Gavin. He had his own techniques for data analysis as he came from a mathematical and psychology background and he also had strong computer science. Within 24 hours he astonished me with what he did with my data. He said ignore NO’s and just concentrate on the YES and MAYBE’s.
Why ignore the NO’s?
He said there’s a lot of noise in the NO’s. He said think more about signals. A clear signal would be a match so get rid of dirty signals. We launched Intro Analytics at iDate ’09. It’s great to have the science/data crunching. To turn that into a product is easy. But for a customer to send you data needs a lot of operation and product management.
A lot of data for you to choose from. What data do you prefer? What do you draw from primarily to establish behaviors you want to match?
We look at implicit data, which is behavior. We are looking at what a person is clicking on and who he is messaging to. We look at the instances of that event and when it happened. Some are stronger signals than others i.e. a message is stronger than a view but even a view is important. When you look at all the views of a person you can personalize the site for those users in a 1-1 way, create a completely different window. People can branch off into area they want to.
One thing I’ve heard consistently is the computer power needed to do this kind of matchmaking is pretty intense. Do you provide computing power?
Yes, it’s part of our magic. Computing has become a lot more powerful but if you know how to utilize hardware for processing data then we can speed it up 100 times more than a database. We don’t use databases, this would take us days to crunch.
Does this happen daily? Do your client’s give you the data feed daily and then you crunch?
Yes, it could happen hourly if you want. But in terms of being able to do a crunch it has to be done in a batch. We need to go back at least 6 months.
You say that you help Internet dating companies make 30 % more money. How?
Amazon says that 35% of its sales come from recommendations, why? Every single page of content on Amazon has a section just for recommendations for you. We look at usual activity, we can show you more of what you like, keep your attention for longer, and convert higher. People need to trust that you have a real science doing real things.
You have a wonderful vantage point in the online dating industry. You can observe a lot of behaviors’ from users. What would you say is the most counter intuitive behavior that you have observed?
We’ve noticed that users 50+ are a lot more loyal, stick around longer. We can actually create stories from looking at data. If someone wants us to look at something specific such as: When people go to bed? We can say that males stay up later whereas females are usually logged off by 2am beauty sleep.
Let’s talk about money. Let’s say I have a dating site with 100,000 paying members and several hundred thousand registrants at any one time. But the paywall is set up so that only the 100,000 subscribers are getting the matches. How much is that going to cost me to use Intro Analytics?
That’s a pretty decent sized site. The ideal price is $3,000 - $10,000 range. We try to squeeze as much analysis as we can into the service. We can offer in-house servers just for you so you can take some of the cost and for smaller clients that can save you about 20%.