Sramana Mitra: I actually have a BMW 3 Series that I categorically lease. I change them every three years.
Dave Elkington: It’s uncanny. Every time I give the example, there’s a 90% overlap. People cluster with other people who behave similarly. What I needed to build was a massive engine that collected all of those profiles of people. The epiphany came to me on how to collect this much data. It’s not a little data. The concept is I needed to know everything about everybody. If I know massive amounts about most people, I should still be able to predict the people I don’t know about and how they’ll behave in similar situations.
I was watching a lecture by a guy who invented the CAPTCHA. When you fill out a web form, there’s usually an image of text that comes up to verify that you’re not a robot. The guy who invented it was a professor at Princeton at that time. He explained how he was faced with this problem. Today, it seems commonplace but at that time, it was very innovative. He explains, “There are hundreds of millions of web forms using this CAPTCHA.” He realized that he missed his window of opportunity to make some money.
He then decided to start a firm called reCAPTCHA, which has the same concept as CAPTCHA. This time, rather than just using random letters and words, he’s using OCR technology to capture words in a scanned or PDF book that needs to be transcribed. There’s this passive crowdsourcing initiative that he came up with. There are hundreds of millions of people filling out these web forms. He’s using their effort to transcribe books for companies. People were paying him to transcribe these books. I had an epiphany that the way I can collect data to build this learning engine needs to be done passively so that people aren’t asked to do things. I decided to build an application and most particularly, an enterprise application. I had to provide enough value like he did with the CAPTCHA where people would use the solution and there’s enough productivity value that I could passively collect the data until I could then begin to provide value around the prediction.
I looked at the customer life cycle. In the enterprise life cycle, you’ve got marketing management. Below, you’ve got sales management and ultimately, customer management. I needed massive amounts of data. Up at the very top of the funnel, there was huge breadth of data with very little depth. Josh James, CEO of Omniture, was a friend of mine. I looked at his model and he had massive breadth without depth. You go to the very bottom of the funnel. There were companies who had huge depth of data without breadth of data. What I was looking for was somewhere in the middle where I could get both breadth and depth. I decided to build a lead management platform that gave me access both above and below into sales and customer management. We built this platform that companies were buying for the pure productivity value but they were agreeing to give us access to their data. These are not just usage data but anonymized inputted data. If you’ve ever bought from any of our customers, you’re anonymous but you’re profiled in our system. We built a communication toolset—things that automated the InsideSales web and more importantly, their activities.
Sramana Mitra: You have some idea about what you wanted to do. One thing we stress hugely is validation. What was happening while this was going on in your head? Were you doing it by the side on your own or were there actual customers that you were talking to who were part of the process?
Dave Elkington: You’re exactly right. We were bootstrapped of course. There were times like in the summer of 2004 when I was starting to transition the professional services activity down into what was increasingly the delivery of the product. With my CFO at that time, we toyed with this concept of just-in-time cash flow. What we would do is we would manage our cash flow very aggressively. We even used some regression models around that. Then, I would hire and would spin the money almost exactly as I got the money. I knew I was going to get some money next month. I would begin the hiring process knowing I wouldn’t have to pay them until next month.
There were two times when I was too aggressive. I overspent. I did that a few times. There was a period that I became significantly too aggressive. Rather than laying everybody off, I went with my wife and got a job cleaning a doctor’s office at night and used the cash that we earned from being janitors to pay employees. You’ve talked to enough of these bootstrapped entrepreneurs where it’s managed tough. In hindsight, you’d do it again.
You have to deliver products to customers. At the end of the day, you can’t do this in your head. I think there are some entrepreneurs who think that they know better than the users. If you’re barely surviving, all that matters is what customers want and more importantly, what they’re willing to pay for. I created this approach of what I thought the solution was. Then, we would go and sell the approach to a customer or partner. I’d say, “If I build this, would you pay for it?” That’s really the first four or five years. It was completely following the customer demand. We would deliver a solution and we would go to our customers and say, “What do you think about this?” More times than not, a customer would say, “This is really great, but what we’d really love is in addition to what you provided us.”
Sramana Mitra: Who were your first five customers?
Dave Elkington: It’s funny you asked. They were all my vendors. These are the guys who did my payroll, insurance, technology. It was mostly my vendors. At the end of 2003, I had the vision. We built this thing without any customers. but we talked to a bunch of customers. We then released it in early 2014. We got a couple of people on and began iterating. That’s one of our cultural definers. We iterate fast. We make very quick decisions. We sell fast and we fix fast. We went to our vendors and said, “Would you be willing to come on? We’ll give it to you free for the first two or three months. Give us lots of feedback and as long as there’s value at the end of that, you can continue to use it for a fee.” Then we finally get enough revenue where we think we’re close enough to where we can cover our monthly burn with these customers.