The Future of Data and Economic Development
Hey everyone, welcome to a super special edition of America’s Jobs Team podcasts. This is Faye Davis. I’m super excited to share with you this awesome interview that Ron Kitchens did with Steve Jast of Gazelle.ai which is a cutting-edge AI-powered business intelligence growth platform. This interview happened live on stage at ECONOMIX this past Thursday, and I hope you learn as much as I did. Enjoy.
All right, you ready for this? I’m ready. Let’s do it. Alright, so first tell us who you are as a leader, then you can tell us about the company but I think unless we know who people are as a leader, I’m not sure the value of what they do matters as much right? I think that everybody has a certain set of skills.
And I think as a leader, it’s very important to surround yourself with people who are really good at the things that you’re not good at. And so it’s important to do some self-reflecting and to understand where you need help and then to surround yourself with those kinds of people. And then let them guide you and let them help you and let them become part of your team and listen to them. And I’ve never been the kind of leader who would want “Yes people” around me and agree with me.
And possess the same kind of skills as I do for fear that I might become insecure that somebody knows more than I do. I want people that know more things than I do and that helps with the creative process. So that’s the way I’ve always tried to be and that’s the way I’ve always tried to surround people around me with. So, what did you do before you started ROI? I was a site selection consultant.
My very first job at University was working at an advertising agency. We did some posters and some work for them and then a gentleman who had just sold his site selection consulting practice was working as a director of government relations and saw these posters and asked who made them. They told him that it was the person representing that company and we got the contract to write his brochures as he was going to start up a new site selection consulting practice.
And as I was writing the copy for those brochures, I said this sounds amazing! I didn’t even know this profession existed and I asked him if he would hire me and so he did. And I spent the next 12 years going from Junior Consultant to Senior Consultant, Director of Consulting. Ultimately 12 years later I became a partner in that firm and it was inside that firm that we started ROI. And in 2002, I went to him and I said, you know, I’ve got this entrepreneurial itch and I would really love to take ROI and spin it out and see if I can build a company from there.
So we approached lead generation and the foundation of ROI with some of that inside knowledge that we possess as being site selection consultants and seeing how the whole process worked out. So when we were approaching companies to talk to them about their growth and their potential growth plans, we could talk to them from a position of knowledge. So that was our value proposition amongst other lead generation companies.
So let’s fast forward. Gazelle.ai started three years ago? So Gazelle.ai started probably closer to 5 years ago. We started our R&D on the product about three years before we launched it. And tell us what it is and then I want to go back and tell us why you thought you should do it.
So Gazelle is a sales intelligence platform very similar to other sales intelligence platforms that are out there in one way. We provide contact information, company listings for about eight million companies around the world. So there’s nothing really amazing about that other than our efforts to bring good data into the platform. The sort of difference that Gazelle brings to the table is the artificial intelligence and some of the proprietary modelings that we’ve done for some of our heat maps.
So we wanted to see if we could bring data together. Big mass amounts of data. We’re talking about billions of data nuggets and see if we can uncover the signals that are prognosticative of corporate expansion in the 18 months before a company expands. That way Economic Development professionals, B2B professionals, deal professionals can get out in front of companies while they’re in those planning stages and try to get them as clients or influence their decisions to come to their jurisdictions. So why I thought I should do that. Well a lot of the same reasons, you know, Ron that you talked about when you first came out.
Um, I had a good friend of mine, Claude Theoret. He’s a Ph.D. astrophysicist and he been telling me for years “you know, Steve you’re curating so much data by all the phone calls and the service that you’re doing, you need to do something with it”. But I was a service guy, I didn’t know anything about technology. I didn’t know anything about artificial intelligence. For 15 years, he kept on me, you know, “this is gold, you’ve got do something with it”.
So I decided that I also didn’t want to become a dinosaur. I didn’t want to be scared of mastering something new, let the old man in. So we decided to embark on this journey and my good buddy Claude Theoret did an initial study on whether or not algorithms could detect these kinds of signals. He thought it was possible. And then the fun started.
Steve who had no experience in technology was now managing a Ph.D. mathematician, Ph.D. physicist, economist. So that could be scary when you first get into that because they’re talking a language that, at the time, I knew nothing about. So fast forward to five years, you’re doing the prognosticative data analysis. Is it working? And is it working at scale? So as a company, I think there are two parts to that question. So we’re trying to scale the business.
So from that perspective, we have about 700 users since we launched our data product in January 2017. We launched the 1st full version in September 2017. So we brought on those users in the past two years. And it’s been exciting because our main goal and our main vertical that we were going after was obviously Economic Development professionals to try and help them better target companies that could be growing and expanding, and interested in their jurisdiction.
So instead of just looking for the old signals that people were looking for a change in c-suite and which may or may not mean a company’s expanding. We wanted to bring a little more science and some method to the madness to try and build lists of companies that really could be in expansion mode. So the company itself has scaled. We brought on a lot more technology people. We have a full-stack Dev team. We’re selling licenses now to institutions of higher education, utilities, B2B marketers.
Even some site selection consultants are on the platform now. So we scaled it from that perspective. From the perspective of does it work? We use machine learning principles, which means that we don’t hardwire artificial intelligence. We let the technology tell us what the signals are to look for. And then we let the technology look for those signals amongst all of our companies and then rate them based on those signals being there. Right? So, for example, if pension plan growth and defined contributions are a signal that most companies have that are expanding, well, that’s something that the algorithms told us.
And now they’re looking for those kinds of signals on their own. So we’ve scaled the business in the sense that we first started out with a training set of about 6,000 companies. Now, we have a training set of 120,000 companies. We started out with a data set that went back 10 years. Now, we’ve got data back going back 12 years. So we scaled the business from a data perspective.
We’re always trying to train the algorithms and the more data we’re pumping into the algorithms for a longer period of time, the better it gets. So to answer your question, we’ve definitely seen an up tick in our conversion rates from building lists that we used to build just based on a hunch, to using the technology and using companies index for growth.
So from my perspective the value of it really is twofold. So it’s this idea that there’s a machine out there constantly grinding corn and then point it pings and says, oh this corn goes to Ron and then we can see whether or not it really does is usable for us. Right? But the other one is I like the ability to go in and use it as a catalog. To be able to say of the companies that we’re targeting, here’s the deep data on them, and here’s a ranking on their ability to grow right.
So it even lets us be smarter even if we’re ahead of marketing to them when they’re maybe three years out from a decision, but we know that based on the data, there are good or bad opportunities to expand. Are other users using it that way? I think so from the very outset we wanted to try and figure out the sort of the gestation period if you will, between the time a company thinks they want to expand to the time that shovel goes in the ground. And that can vary from company to company. As a site selector, we’re talking about yesterday. Sometimes an economic development professional will send out an RFI and the company goes quiet for a while. So the assumption that we made is that a company will be in a planning stage for 18-24 months.
Some can be 36 months, some can be longer. So to your point, I think that the main difference that Gazelle has been able to afford to Economic Development professionals responsible for investment attraction is to get out in front of these companies who may be expanding in 18 to 24 months because that’s when the decisions are being made. So I hope they’re using it that way. That’s the way it was built and the way it was meant to be used. But we also have a lot of data as it relates to VC funding events, the AI scores, we also have live projects.
So users are able to cluster companies and discover target-rich areas for them for the industry that they’re going after. Rather than just a scattergun approach, they can really hone in on a location where we see revenue growth in the industry, establishment growth in the industry, average salaries raising, perhaps lots of VC funding, and then lots of highly rated companies for growth based on the AI.
So these to me are far richer zones to troll in, as opposed to you know, who’s changed their CFO, who changed their CEO, which is what we used to do 10-15 years ago before we had this kind of data.
So think about using it in going deeper on even your existing companies and getting that data that maybe your local plant manager has no idea about from a corporate standpoint. How are people using that or are they using it to manage their own portfolio? That’s an interesting question. So like a manufacturer type of organization? Yes, like somebody is in your community that’s in your database, and getting that information, how many people do you have? And what’s the chance of growing?
Well, I think that from a BR&E perspective the system definitely affords the ability to see what kind of companies are in your jurisdiction. Are they subsidiaries? Are they headquarters? Are they likely to grow? We have some utility users who like to see what the possibilities are for companies to grow in their grid and to see which particular industries may be growing in their grid. What the system does not deal with. However, is it does not prognosticate bankruptcy or failure. It’s not a credit risk tool.
We have experimented with putting critic ratings into the platform to see if that would alter the scoring of the algorithms. But from a business retention and expansion perspective, I think it’s effective. Now as it relates to a company manufacturer there’s a use case there. I mean, there’s certainly a B2B use case for folks who want to be bringing in business. We buy a lot of data from a lot of great partners and one of our partners is EMSI and we buy the industrial buying and selling ecosystems from them and the main difference between us and EMSI is then we populate a lot of companies in those buying and selling ecosystems.
So for example, if you’re running a company and you want to see where some potential suppliers are. If you’re an EDO and you’ve got cluster leakage the system works well for that as well, because it can find your addressable market on the sell-side and it can find your potential suppliers on the buy-side. So if an economic development professional has a lead and the lead says, “well, I need to see where my suppliers would be located within a 1h or one day drive or what my addressable Market is between in a 1h to one day drive away”.
They drive we can show them that as well. So I hope that answers your question. Okay, yes it does. What do you think the future for artificial intelligence data gathering to be applicable to men and women in this room? The future.
I mean if you would have told me five years ago that we’d be using artificial intelligence in this way, I would say that that would be amazing. We still need to play with the algorithms. We still need to see over time how effective they really are. It’s going to take a few years for us to continue to validate and see to what extent they’re effective. We see certainly an uptick in conversion rates by using highly rated AI scores. I think in the future what we like to see are different kinds of AI scores: sustainability scores, industry growth scores.
Right now for industry growth we basically see linear regressions. I think we can do better than that. We’d also like to see M&A scores that would then allow B2B or business folks and economic development professionals to get different kinds of signals to better prepare and better formulate their strategies. It’s interesting to see the environmental score, but I can also see some variation of Glassdoor scores, where you know, we only want companies that hold these standards.
Exactly. And yeah, and I can see particular States like California where you know, we want to target companies that meet the state’s regulatory strategy. Exactly. So that’s what we’re looking into now building different kinds of scores. Of course, it takes time. We need to find the right data. Sometimes data sources are free. Sometimes they’re not free and when they’re not free that creates a certain amount of risk in the R&D environment.
You spend a lot of money buying data, you spend a lot of money experimenting with that data and you don’t really know what the outcome is going to be. But truly that’s what I think the future is going to be. Artificial intelligence, in my opinion, is never going to replace humans in a lot of ways. In some ways. It might I mean, for example, ABS brakes are used by, you know, there are algorithms in brakes. Algorithms have been around for a very long time. I don’t know.
I mean, I was educated on AI, I wouldn’t have know any of this five years ago, but it’s not new. Algorithms have been around for 25 years longer. What’s new, as far as our company’s concern, is the proliferation of the availability of data that you can plug into these algorithms. So I don’t remember what the question is. But yeah, but I can keep talking if you like. So speaking of that. Do we have questions from the audience?
Okay, I’m a data a guy and my question is do you have a category for startups and incubator type companies because we are starting to mine firms. Those are the guys who you really want to be looking at as well because those are the future larger companies that people are going to be wanting to surround themselves with. Yes, startups are important for a number of reasons. So we could have a coffee after because this could be a long conversation.
So we do have startups. There are nine million corporate establishments in the gazelle platform. And I really don’t want this to become an infomercial on Gazelle. But since you asked the question, I’ll tell you how we approach this. We wanted to populate the database product with companies that would be relevant to Economic Development professionals. So for us, startups are relevant in two ways. Did they receive a pre-seed or seed funding, which means they have the possibility to grow.
We’re not really populating the database with let’s say incubators companies. They need to have gone sort of through that phase at least and received some level of funding. So if they’ve been funded there in the database. And that’s important for two reasons. Some of them can be what we call “Gazelle companies”, which is kind of what we named the platform after. These are companies that have a combined annual growth rate of 20%, five years in a row, doubling in size. And these really are the catalysts and the generators of about 70% of net new jobs in any economy.
So they are important. From the other perspective. I think what’s becoming a really interesting strategy for economic development professionals, is to spend a little less time on the big multinationals capable of these transformational projects, as we heard from the site selection Consultants. These are really not a dime a dozen they’re hard to find but if you have a technology innovation ecosystem that is fueling innovation, large multinationals will find them.
They will want to invest in them and they will then come to your community because they may set up a JV, they may buy the company. So a great way to attract the large companies capable of transformational projects is to keep an eye on these really innovative tech companies. So to answer your short question with a very long answer, we do have those companies in the platform so long as they’ve been pre-seed or seed funded and we also have SBR funding companies in the platform as well.
Companies that are smaller than that, it’s tough to really want to build a list of those. They’re not really a company, they are an idea. There are other database products that I can mention to you that focus exclusively on startups all over the world.
Other questions. clearly the AI requires all the data you can get. Drives live on data. Do you see going forward, will these datasets be more robust or will it be things about business reporting that will make it more difficult. Just kind of curious. You’re looking out, what kind of opportunities are you looking for? It’ll make your product even more robust. From a data perspective? Yes.
The simple answer is yes. So when we first started the platform we had a training set of about 6,000 companies. A training set of a hundred twenty thousand companies. We probably started out with 50 or 60 data points. We know 900 data points. So we have two Ph.D. economists and folks that specialize in cognitive studies and neural networks and machine learning and their job really is to constantly be looking for data that might impact the algorithms. So this is ongoing all the time.
So I think this is the last question. We have more data available to us now than any humans that have ever lived on the planet have access to and most of it’s on our phones. What we don’t know is what to do with the data. Is Gazelle the answer to that? Of taking this data and giving it in a way that we can immediately see a way for us to use it to benefit our communities? I hope so. I hope that’s the experience that the users are taking away from it. Part of the frustration that we had when we built Gazelle was, as a lead generation company, we were using about 20 different database products.
And then we had to cross match and look for data that was interesting about a company on a different platform and we couldn’t find them. It was very frustrating, and it creates a lot of inefficiencies for us. So Gazelle, number one created efficiencies for us. And then it would just be completely unrealistic and impossible. We have over a billion data points that are turning on a daily basis. So if you are an investment attraction professional, it would just be simply impossible to process that much data to make your life a bit easier.
In as much as finding companies that could possibly be growing and that’s what we want right as an investment attraction professional look for growing companies that can come into your community and continue to grow and thrive there. Awesome.
Well, thanks for coming out and talking about it, and as part of my idea of Mastery I told you about, we’re going to subscribe to this but I’m going to be the first person on the team to go through it because, one I’m a data geek, but two I want to master it. So when I actually have earned this certificate now. Steve, thanks. This has been great. Yeah. Thank you. Thanks, everybody.
Other Interesting Content