Olin Hyde is the CEO of LeadCrunch, Inc. Based in San Diego, LeadCrunch offers an artificial intelligence platform for generating business-to-business (B2B) leads. Olin received a masters degree from UCSD in advanced studies on systems engineering and has been active in applying artificial intelligence for solving real-world problems for over a decade. I had the opportunity to catch up with Olin, who’s always on the move, and talk about what’s new with his business and opportunities to leverage artificial intelligence in business development.
DF: We’ve known each other for several years now, and to say it upfront, I’m on the Board of Advisors for your company, LeadCrunch, and it’s been a pleasure working with you.
OH: Thanks Dave. As you know, I’m a huge fan of your books, research, and ideas.
DF: So let’s start with an obligatory question: What’s LeadCrunch all about?
OH: LeadCrunch is an intelligent targeted demand generation platform that accelerates business-to-business (B2B) sales. It combines the power of what artificial intelligence does best — pattern recognition with speed, scale, and accuracy — with what humans do best — which is make emotional meaningful connections with each other. Our system uses artificial intelligence to find prospects that look like your best customers, that is, it recognizes the patterns that make a best customer. It finds the leads that wouldn’t be described by a simple industry code or size of a company. LeadCrunch specializes in discovering hard-to-find customers — ones that close fast and are ideal fits to your company.
DF: OK, so when you find that prospect, how do you communicate with them?
OH: When we find your ideal prospects, we also model how those prospects will respond to you. We know what has worked for tens of thousands of companies – so we can repeat this success with you. Our models tell us what form of communication will generate the best responses for you: content marketing or telemarketing or email marketing. Then we model what type of message this particular buyer is most interested in. So when we hand you a lead, you know that they are the right person, at the right company, who is interested in your solution. This makes sales much easier and faster.
DF: It’s interesting that there are consistent patterns across preferred communications for various businesses.
OH: Yes, exactly. It’s stunning how predictable people are. Recognizing who fits into what pattern allows us to deliver relevant information. This allows your customers to complete a lot of their journey of discovery long before your sales people call them. So the sales cycles is extraordinarily short and your salespeople are very effective.
DF: And how does the LeadCrunch make money?
OH: We make money by cost per lead. This allows our customers to accurately measure their return on investment and compare the quality of our service to others.
DF: And who are your ideal customers?
OH: We look for customers who are actively running demand generation campaigns. We really want people who currently use the big vendors such as IDG, Ziff-Davis, eMedia — because we typically outperform them by 300 percent. We require that our customers measure success. We look for customers that use Salesforce, Marketo, and other CRM and marketing automation technologies. Our best customers are analytics companies. We only sell to those who we know will be successful with our technology. We are in the customer success business – we are not in the business of training people who do not know how to market.
DF: Do you like talking about the specific AI algorithms you use?
OH: Well, when AI works people call it software and when it doesn’t work people talk a lot about artificial intelligence technologies. [Smiling] Let’s talk about our software. We adopt a hyper-signal processing approach where we’re not looking at any one signal but rather processing a spectrum of different types of signals. We can literally look at many millions of attributes for any given prospect. Our secret sauce is how we reduce those millions down to something that a human can understand, which usually is fewer than seven.
DF: OK, I expect many readers will be getting their heads around processing millions of attributes for lead generation. What types of data are we talking about?
OH: Think of that as looking at the unique attributes of every single person in every single company. There are millions of ways to describe each person and even more for an entire company. Marketers need to be specific to be effective. What we do is replace the generalizations like industry codes with specifics like what equipment they use. Just like we don’t use stereotypes to understand our family, our friends, or our spouse there is no reason to use stereotypes, or generalizations, to understand customers. Rewarding relationships require that we look deeper than the generalization. I didn’t pick my wife based on a stereotype. I don’t want to get to know my customer through a stereotype. Going back to the way things have been done in this business until now, it’s been through stereotypes.
DF: A lot of statistical analysis is like that of course.
OH: Yes, right. What industry are you in? How big is your company? If you know a customer sells only to companies with one particular software vendor – like Salesforce – it might be fine if you’re marketing a Salesforce plug-in. Great. That’s a pretty easy customer to find. I think a more challenging and more interesting and rewarding problem to solve is “How do we find new areas of opportunity that are outside of our imagination?”
DF: Thinking outside the black box, as it were.
OH: Exactly. What is the unknown unknown where we could really accelerate the growth of our company? And now we have story after story where we’ve helped our customers find completely green fields for growth. One is a point-of-sale (PoS) vendor who sells PoS systems to restaurants and bars and they couldn’t figure out how to profile their customers. It seemed random as to which restaurants and bars were buying their system. But in our system, without any training, we automatically detected that their best customers served Bloody Mary’s at 10 a.m. on Sunday mornings. How did it know that? It went and started reading menus. It found a very strong signal there, and it didn’t make a lot of sense to us right away. We went to their vice president of marketing and said “Look, does this make sense to you?” And it did. Those businesses have a completely different inventory management system, they have a different business model, and it was a perfect fit for them.
DF: How repeatable is this?
OH: The nuggets of insight vary by every customer, but we typically find this sort of nugget in almost every case. Having these insights allow our customers to quickly close qualifying deals.
DF: How long does it take between the time someone engages with you and the time they have their first new business leads?
OH: Typically, the time between someone engaging with us and seeing our first delivery is within three days. So it’s quick. We provide the first targets to go through and they call a few of them and let us know what they think of them and score them. That’s gives us some calibration. Our AI techniques are a combination of unsupervised learning using deep neural nets and supervised learning where the human is actually giving feedback. The combination of unsupervised and supervised learning together creates what we call a system of intelligence where the machine has trained very specifically to act as an agent on behalf of our customer. And this is a real competitive advantage for our company because once you get a system trained that’s specific to you — wow — it gets smarter and smarter and smarter. You get what’s called a first-mover advantage.
DF: How’s your rebooking rate?
OH: Incredibly good. We went to market in September 2016. After nine months we reached a $1.9 million annual run rate. In May, 100 percent of our customers rebooked within 90 days. And that rate is increasing over time. We were over 133 percent rebooking for our May cohort. [Smiling] The only way we get above 100 percent is people are buying larger orders on their second order because they are so happy with the first.
DF: Do you operate an on-demand model then?
OH: Right now, yes, it’s an on-demand model. Because that is what the market wants. We are in alpha testing of a subscription product that provides insights that you can’t find anywhere else. We expect to be in beta testing by August. Even before we’ve released it, we’ve had two customers who’ve stepped forward and said they want to sign up.
DF: Are you looking for more people to get on the alpha?
OH: Absolutely. But we’re selective on who we get. We’re looking for companies with more than 150 employees that have a director or manager of demand generation. That’s our primary persona. Secondarily, we do like the little guys too. The mission and vision of our company is to democratize the power of AI to enable main street USA businesses to compete against the enterprise companies. Everyone likes the little guy and oftentimes the little guy doesn’t have the opportunity to get the same advanced technologies at the enterprise level that we want to give to them. So for the companies with fewer than 150 employees we’re looking for vice presidents of marketing or vice presidents of sales and marketing who are responsible for lead generation for their companies. We want people who’ve had experience with demand generation campaigns and who have a baseline to measure us against. In addition, when it comes out of private trial and we start charging for it, we’ll give them a 50 percent discount. So there’s a good incentive there.
DF: Quantitative measurement of success or failure in lead generation must be central to your business model.
OH: Yes. We seek customers that are analytically driven. This enables integrity. We need to know the conversion rate for our leads. What’s the return on investment on our service? Why are we objectively better than our competitors? We are all about delivering truth that is predictable, repeatable, and transparent.
DF: Let’s talk about your team. As I mentioned at the start, I’m on your Board of Advisors, but let’s focus on the people who are working every day full-time to make LeadCrunch a success.
OH: Our operational team has experience at more than 20 startups where they have transformed several industries and realized large exits. My partner, Sanjit Singh, and I actually met as angel investors in another startup. We were investors in an online dating company. We like to say we met online dating — [Laughing] Don’t worry, our wives know about it. We decided to start LeadCrunch together because we liked how we solved problems together. We started with a very different use case and we’ve had a couple of pivots along the way. But once it was clear we were going to do LeadCrunch, we built this team specifically for this market. Our chief scientist Steven Biafore is co-inventor of FICO’s Falcon algorithm and is also founder of Global Analytics, Inc., Zebit, and SoftStack Factory. If you have a credit card, chances are his algorithm has protected you from fraud. Steve’s work got the attention of FICO, and one of our largest investors is the inventor of the FICO algorithm. Our advisory board and board of directors have realized more than $20 billion in exits from early stage investments. So we are focused on growing a very big, profitable business.
DF: There are obvious connections in pattern recognition and pattern discovery between those problem areas and finding new business leads.
OH: Right. Our team has a lot of experience in applied AI for credit analysis, fraud detection, and cybersecurity and we do see the problems in those areas as being very similar to demand generation. One of our advisors is Dr. Marvin Langston, who is the former ISO director of DARPA’s AI projects and went on to be the chief information officer of the Navy. Marv looked at this technology and knew it was really solid for target analysis and verification. We actually did a defense project for Lockheed where we beat IBM Watson and Palantir to win a research and development project. But we’re not defense people and we want something that’s going to scale really quickly. That’s wasn’t going to be the defense sector.
DF: I recall you once mentioned to me that you were surprised to find few companies using advanced analytics for their lead generation.
OH: It’s true. While we were developing our product we had more than 1000 companies sign up for either a free or paid trial. We talked to all of them and we learned that 96 percent were not using analytics to find their customers. We found the real problem was that the data to really understand their customers were fragmented across many different platforms. There was no way for these companies to holistically optimize across those platforms. Our platform helps fill that need.
DF: What is the investment landscape in this space?
OH: To date, there’s been about $700 million invested in demand generation technologies. That investment has only generated about $150 million in revenue. Not so good. And why is there this incredible disparity so far? We believe that a lot of the venture-backed companies have been solving the wrong problem. They’re solving the problem as if one approach can do the whole thing. We think that’s wrong. The system of intelligence that we’re building is really the enablement of people to do superhuman tasks with AI.
DF: Augmenting the domain expert rather than replacing him or her?
OH: Right, and that gets back to that recursive learning and the combination of supervised and unsupervised learning. Our software gets trained specifically for you. It becomes an extension of you and that doesn’t happen if you’re just doing a pure AI play and all you’re doing is scoring predictions. That actually doesn’t add a lot of value.
DF: I’ve found intelligence augmentation is really a key in many applications.
OH: I believe that’s true, and that’s one of the reasons we’re such a big fan of your work. I know you’ve written about it many times.
DF: Right, for example, in Blondie24, where we evolved a deep learning unsupervised convolutional neural network for playing checkers, we found that even though the net was evaluated ultimately at the level of human experts, if you combine the neural network with domain-specific knowledge about endgame positions, it’s very easy to generate performance at the higher level of human masters.
OH: I believe that there’s been a lot of misinformation about what AI is and is not. It’s important to build everything around the person. Human in-the-loop is insufficient. The human is the loop. Humans drive adoption and use. I think one of the failures many companies have is that they get lost in the algorithm or the technique. Today, everybody really has access to the same mathematics. Some of the best technologies right now are actually open source and it’s not really about the algorithm, it’s about the data and it’s about how do you apply algorithms to a specific set of data to get a specific outcome. And that tuning and data curation process is really hard. I would say the greatest advantage that we have technically is the way we annotate and combine data to get unique insights. It’s really exciting but it’s not something that you can really go out and tell the customer, “Hey this is what we’re doing deep under the hood.” Honestly, they don’t care. All they care about is “Is this lead converting? Am I getting a faster, more high-quality relationships?” And so that’s how we measure our success.
DF: What’s your long-term goal for LeadCrunch?
OH: We’re growing as quickly as we can to be a $100 million business. We believe this business has the capability of being very large and very profitable very quickly. We’ve doubled in size every two months since we released our product in September 2016. We see our growth rate will still be double digits month over month for the foreseeable future. We are aiming to be a $20 million business within two years. We’d love to be the first San Diego Unicorn – a company worth over $1 billion. However, we also recognize that we have to have a national presence. Our sales team is out of Chicago. We have an office in San Francisco. There’s a lot of pressure on us to relocate our headquarters to San Francisco. And my co-founder and I have the attitude we’ll do whatever it takes to make our customers successful because that’s the shortest path for us to be successful.
To contact Olin Hyde, email: firstname.lastname@example.org
To contact David Fogel, email: email@example.com
© David Fogel, 2017