Paula Chiocchi: B2B Influence Spotlight on Industry Game Changers. I'm your host, Paula Chiocchi C CEO and founder of Outward Media. Today we are joined by John Kosturos from SpringDB. John is a business leader and entrepreneur with extensive experience in data, technology, and revenue growth. John has focused his career on turning data into meaningful business outcomes with expertise, spanning digital transformation, data quality, and emerging technologies. So today we'll explore how businesses can unlock greater value from their data, navigate the rise of AI, which we're all focusing on, and build smarter strategies for sustainable growth. Welcome, John.
John Kosturos: Thank you. Yeah, you usually when I introduce myself to people not in our space,
Paula Chiocchi: Yes.
John Kosturos: I'm like, yeah, I work with computers. Exactly. You know, for for my
Paula Chiocchi: That makes it this big.
John Kosturos: mom and all her friends that works really well 'cause she wasn't in tech,
Paula Chiocchi: Yeah.
John Kosturos: so you know.
Paula Chiocchi: Yeah, 'cause and then you try to drill down and then the eyes glaze over. So so
John Kosturos: Yeah.
Paula Chiocchi: so you spent much of your career at the intersection of data technology and business growth. What originally drew you into the industry and how has your perspective evolved over time?
John Kosturos: Yeah, so interesting path into data. When I was younger I always thought I was gonna be a professional athlete and you know how that works
Paula Chiocchi: Okay.
John Kosturos: out. So went from that to kind of entrepreneurialism and joined a a family office that we would at the time we were purchasing e commerce dropship businesses. and we were building out the the tech and the sales and the marketing. So you need data for all that. Lucky for us, you know, that's when Salesforce dot com started to
Paula Chiocchi: Okay.
John Kosturos: To take over. And I was young enough at that time where I really loved the idea of being able to work in the cloud from anywhere. And you know, so
Paula Chiocchi: Okay. Okay.
John Kosturos: we would inst install Salesforce and build out e-commerce catalogs and you know sales, marketing. At the time there wasn't a lot of marketing automation. It was like Genius, I think, was the main one then.
Paula Chiocchi: Okay.
John Kosturos: but loved Salesforce, loved CRM and Saw that that market was really exploding and they were building a great community around their partners. So we went to look for businesses to acquire in that space and found a company called Ringlead, which
Paula Chiocchi: Okay.
John Kosturos: at the time was a deduplication and duplicate prevention technology for Salesforce. long story from there, I mean I had some of the best operators in the world basically educating me. in my discovery calls and in my sales calls and customer success and so lucky enough to just kind of be educated by that RevOps slash data science persona for now close to two decades.
Paula Chiocchi: Okay. Well that's a that's an evolution though, right? To go from Salesforce to today. Big evolution. you've described your work as building companies that turn data into revenue and product outcomes. What does that mean in practice and what separates businesses that succeed at that and from those that actually struggle to do that?
John Kosturos: Yeah, so you know, I think that that question can go in a lot of ways, but I'll answer it in a way where we've kind of built our philosophy. you know, there's probably around five to six thousand companies in in our space right now selling real s real data products. But what I think is over the next five to ten years, as AI really starts to take over, that companies who today don't know that they have valuable data that can be monetized in a major way will potentially be joining us all in this data market. And that may become their biggest asset. So I think it it starts at sourcing data. What do you have as a first-party data asset? What are the third party data assets that can enhance and optimize, validate the information so it's refined and usable? and that's the first step. So it's it's sourcing whether it's first party or third party app data, and then being able to interact with it. And that's why we built our technology called InstaDB. Most companies, once data gets big, it gets very hard to manage and interact with. And that's really what our marketplace does is help you to source data, interact with it, and after you're able to interact with it, it's all about integration and then monetization. And that's kind of the wheel that goes you know, I guess to to break it down a little simpler. Every company has what we call a data bank. A data bank. So
Paula Chiocchi: A data bin? Okay. bank. Okay. Yep.
John Kosturos: Just like you have a bank account today, you make deposits in your b your bank account, or you you're pulling money out of your bank account, right? but the deposits in a data bank are like, you do some cleansing and remove duplicates or normalize data or you're getting data in from your customers or you're buying it from third parties. Like everything you do to enhance your data is just another investment into your accounts. And it As you make more and more investment into a bank account, and the balance gets higher and higher, what you can do at that point is you can start lending on the balance, right? And that's how we feel like we can help companies. A lot of the times we come in and it's I need to do better, get better marketing data or sales data, or I need better analytics for my team. And those are all great things to do, and those are great ways to invest in your bank. But ultimately, once you've made enough investment. investment, that's when you can start lending on that balance and productizing your data and monetizing it through this kind of circular motion.
Paula Chiocchi: Well, I think to your point at the beginning of your answer is that a lot of people don't even realize that their data has value. Right?
John Kosturos: Yeah.
Paula Chiocchi: And that that's a big, big point.
John Kosturos: Yeah. I just got an ad on my Facebook today and I've heard a lot about this, but it the AI labs are now reaching out to companies Saying that they'll pay them $250,000 to $4 plus million dollars for their internal workforce data. So hand
Paula Chiocchi: Really?
John Kosturos: over your Google Drive and your Google, your Gmail, and your Slack conversations and your your GitHub portal.
Paula Chiocchi: I haven't seen that. Wow. Wow. Ooh, really gonna be big brother, right?
John Kosturos: Yeah.
Paula Chiocchi: So so looking back at the companies and teams you've you've worked with over the years, what's one leadership lesson that has had the greatest impact on your approach to business today?
John Kosturos: that's a that's a good question. you know i it when we we're in the the kind of private equity family office space, you you purchase a company that you know has product market fit But it requires a bit of rigor and process to be built in to get to the scale that it deserves and to get to the exit that you want. And so I had a lot of I had a lot of kind of mentors brought into that process and and one of them, it's funny, it's like we we basically had to come in in a a dress shirt and a tie and look the part, you know, feel the part. you know, that's he inst instilled a lot of rigor. So metrics around calls, emails, opportunities, pipeline, you know, close one deals. but one of the things that he was really adamant about was you've got twenty meetings on your calendar this week. why not send out a personalized notification twenty four or forty hours before the meeting that has you and the you and your team who's gonna be on the call their pictures and an agenda. And when you get to the call, you have to have a personalized presentation. And you know, you pitch the store before and then the pain and then the product and the solution. and then a recap goes out out after every single call. And you know, just simple little processes that
Paula Chiocchi: Mm-hmm.
John Kosturos: made you stand out in a sales cycle thing like build a business case for somebody who's gonna invest in your solution. But those simple processes, you know, the buyer actually really appreciates. And at the end of the day, when you are trying to sell something of value to a client, they do have to put together a business case. And guess what? None of them like to do it.
Paula Chiocchi: Mm-hmm.
John Kosturos: So if you're the vendor that from start to fet finish has shown that process and at the end of the day produce material that will take time and work away from the the buyer, they're more likely to select you whether or not your product is superior, right? The the process is superior. And luckily enough for us now, all that stuff can be done with AI agents
Paula Chiocchi: Right.
John Kosturos: and we don't have to stay up till midnight the night before. Yeah.
Paula Chiocchi: Midnight every night. It really has changed that, hasn't it? I mean it's it's it's phenomenally just like that. So and and and probably just gonna get bigger and bigger, I would think. So you you gotta hang on to our little seats here because here w that's where we're going.
John Kosturos: Absolutely.
Paula Chiocchi: You've written about the importance of data quality and digital transformation. Why do you think so many organizations still struggle with data accuracy despite having more sophisticated technologies ever as we've talked about, just been talking about AI. Why why do you think they just aren't educated on it or what do you think the problem is?
John Kosturos: Yeah, I think that it's just this issue of data is not the sexy thing that the leaders of a company are thinking about investing in. You know. I've done projects with companies where we're going in to to fix their CRM data and to, you know, build out their total addressable market and get their ABM account strategy in place. And you know, you're in there for 90 days and the only thing that the CEO and the CFO can ask is like, where are the leads? Where are the leads? And
Paula Chiocchi: Yeah.
John Kosturos: they're out there They're like investing in email syndication and outsource teledialing and appointment setting. But so they're willing to to basically funnel a huge amount of investment into salespeople, into marketing, into ads, and the third-party syndication and appointment setting. but Ultimately they're not thinking about the root cause of their issue. And so I think fundamentally it's just where is focus of the executive team who ultimately make the purchasing decision. if you look at the RevOps team, the marketing ops team, the sales ops team, even the data science a little bit, but those operators are typically the most underappreciated individuals in the entire go-to-mark market organization. While probably knowing the most about the data, the most about the life cycle of the customer, and having really great ideas on how to solve strategic problems. And they're typically the ones that are instigating data purchases. But if you look at the budgets given to an a a RevOps professional these days, it's like they want you to cut costs. They want you to go get better analytics. now one thing that is changing though. Is that everyone in their in in their business is trying to figure out how to implement artificial intelligence, AI, you know, create agentic workflows, automation. And that is making data a bit sexier these days. It is
Paula Chiocchi: Mm-hmm.
John Kosturos: driving, you know, there is a dialogue that I've seen a lot saying like you can't build AI on this shaky foundation. Like, if you do, it will fall. But if you build the A AI on a solid foundation of data, the sky's the limit. And so I I feel like over the next two to three years, the executive committee at both small and large businesses are going to finally make the investment deserved in the data quality.
Paula Chiocchi: Mm. So that would be kinda like music to people's ears like ours, right? Finally read by data quality.
John Kosturos: Yeah. I mean the the one challenge I think that is still gonna be faced is it's also a talent issue. Like cleaning data is not easy and you really need to find good people to do it. so I am also hopeful that you know that talent issue is not just gonna go away, but with AI
Paula Chiocchi: They get better. Yeah, yeah.
John Kosturos: Maybe we can supplement the talent into helping at a at a larger scale. Yeah.
Paula Chiocchi: So so there's job security and cleaning data, it sounds like. So
John Kosturos: Yeah.
Paula Chiocchi: not to be afraid of AI. So
John Kosturos: Yeah.
Paula Chiocchi: so so with companies are collecting enormous amounts of information, you and I both know that, but collecting data and creating business value are two very different things. Where do you see the biggest disconnect?
John Kosturos: Yeah, I think the collection of data, it's it's becoming more and more valuable regardless, I guess, again, because AI can do a decent job of navigating through it if if it has the right access. But you know, I think the what really needs to be done at a at a in a better way and at a across more businesses is just the organization of the data, right? my brother is a commercial real estate investor and for many years he would tell me it's like he's older and he would say you could sell your business and if it's if you don't have folders in your email and your your drive and and you don't create like what what we now call that AI brain where it's tagged, it's it's properly organized, you'll end up having to be at that business for many years after you sell it, and you won't be able to sell it for as much. But an
Paula Chiocchi: Okay.
John Kosturos: organized business is sold at a higher rate, and it's also less likely that you're gonna have to stick Around for a long period of time if you don't want to. And so I think that's the same with just data hoarding. It's like you need to make sure that you have processes to tag to know create folder structure and create sense around your data assets. I think Notion and like Airtable have been amazing technologies for the organization of data that will ultimately feed into AI. use cases. so those are two really great resources that we use. And then like make.com can interact with that and you can build really cool stuff off of it. but there's also you know companies now on the web like there's a company called Wint Labs. They're literally scraping the entire web and licensing that data to AI AI labs and that's become their entire business, right? So, you know, there's different levels to hoarding and building data sets, but I think for the average company it's all about like you need to have it labeled, you need to have it tagged, you need to be able to create governance and access control around it so that you can utilize all that information.
Paula Chiocchi: Well, I'm learning a lot today. I this this this last one you just talked about scraping everything and giving it to AI. I haven't heard about that.
John Kosturos: Yeah, no, it's crazy. But they have
Paula Chiocchi: Is yep.
John Kosturos: probably hundreds of petabytes of data, which is just more than we could imagine.
Paula Chiocchi: The computing power. So
John Kosturos: Exactly.
Paula Chiocchi: how should organizations evaluate the true value of their data investments beyond traditional metrics such as things that you and I have heard for years, volume coverage, cost?
John Kosturos: Yeah, I think we do a lot of valuation on data and I think it starts with scoring the the cleanliness of the information. And we'll I guess we'll talk about traditional contact company opportunity data, right? And then we can get into other areas. But it's like, okay, you're looking at acquiring a a business, maybe it's a strategic acquisition. your goal is to to to get new customers or new revenue through the acquisition So the first thing that I would look at is okay, how many accounts do they have in their system? How many contacts do they have in their system? And one of the things I'll tell you right now is that at Ringlead, we probably worked with 10,000 Salesforce orgs from very small high-growth companies to enterprises that are had thousands of users. One thing that I could tell you about the companies that were doing the most revenue the fastest, and this could be like a company that started two years ago and they're just crushing it, is they had millions of contacts in their B2B databases. literally
Paula Chiocchi: Okay.
John Kosturos: anywhere from two to fifty million contacts in their database. And I'm not saying that every company should go out and just acquire hundreds of thousands or millions of contacts, but What I saw that was that that number did make a lot of sense. And maybe they had a really great PLG technology where they could build a database that quickly. But others built community, they built content. there were different ways that they were building out these data sets. So I think, you know, the the the size of the database, the overlap of You know what what if you're going to acquire a business, what do they have that you don't have? Where
Paula Chiocchi: Okay.
John Kosturos: is the delta, the overlap? and then just ultimately there's little things like how many duplicates are in the database? You know, what is the fill rate of the fields in the information? serious decisions, now no longer serious decisions. I
Paula Chiocchi: Right.
John Kosturos: think they're you know, they've been acquired, but they used to say the cost of dirty data on demand gen between a best practice organization with data and a normal organization is about a 60% swing in revenue. And the real the reason behind that is that let's just say you compared two contact databases that both had a million records. Quality to mean means that do there's not a duplicate. So hey, if there's 20% duplicates, you got 800,000 records in one.
Paula Chiocchi: Right.
John Kosturos: hey, job titles and states and countries have a a convention, they're and they're populated with data. So let's say the the bad database is missing 20% of their job titles. And then the other and then there's 20% that just have poor standards. And really building a sales and marketing campaign stems from reporting into your data. Okay. I wanna find people in California with these job titles. That have a business email, right? So if you ran that query over here on that best practice database with pristine information versus the one over here that had slight variations, they hadn't been filling in their data bank, well, you're gonna
Paula Chiocchi: Mm-hmm.
John Kosturos: get a a larger number of participants in every single call campaign, email campaign, advertising campaign. It really does matter.
Paula Chiocchi: Mhm. Yeah. Well, and I think I think people are starting to figure that out a little bit better than they used to. 'Cause w
John Kosturos: Yeah.
Paula Chiocchi: y in terms of like filling out their data records and their C R platforms and all, more so than they did even a few years ago, I think.
John Kosturos: Yeah, there's great enrichment providers and and now workflow enrichment providers where you can combine AI enrichment with multiple vendors and so I think they are getting better at it for sure.
Paula Chiocchi: So it's like like there's no excuse anymore. It's it's a lot easier. Yeah, right. Right.
John Kosturos: Yeah, exactly. The only excuse is the budget, right? Do we have the money to pay for the data?
Paula Chiocchi: And then you have to make the case to the management to get that budget. So
John Kosturos: That's right.
Paula Chiocchi: bad data in, bad data out, right? So when a when a company discovers that its data isn't producing the expected business results, where do you think leadership should begin looking for the answers? Is it going back to the data, the the data the quality of the data in and out or do you s see other things?
John Kosturos: Yeah, I I I think it does start at going back to the, you know, I would say the revenue operations organization, getting some reporting on you know, duplicate percentages, missing data values, valid emails, valid phone numbers, identifying areas where there is missing information. And then there's also just it that's on your first party data, right? Then there's this whole third party data world where if you evaluated with your data science team what percentage of your total addressable market was actually in your system, in your marketing automation, in your nurture campaigns, you'd probably find that there's a higher percentage of non-TAM records in your CRM and marketing system than there are TAM records. and that's a really important next step. It's like what what do you have that's clean? But then now we do some market sizing and figure out what is the universe. and a lot of data vendors will license you their whole what they call data cubes. You take those cubes, you overlay them on your data, you run some simple data warehouse views, and you'll identify like what is the white space of data that I don't have in my campaigns today, I don't have in my system. And then there's w ways to activate that. You don't need to go out and do illegal email spamming and stuff like that. You could work with
Paula Chiocchi: Mm, mm, mm.
John Kosturos: you know, there's a a ton of syndication providers that do have a lot of those records already opted in and they'll send
Paula Chiocchi: Right.
John Kosturos: emails, they'll drive people to your landing page. It's and it's about you. Do you have a community for them to interact? Do you have good content for you know, these these this TAM audience to be driven to? Are you creating any media and funny videos or in informational videos that you can go run ads to those people on their LinkedIn and try to get them to follow you and and and subscribe. So I I do think where a lot of the leaders like if you went to a company right now and a CEO, nine out of ten if you ask them, tell me the data attributes about your ICP. Tell me the size of that market and then tell me how much of that data how many of those people are actually in your system today. They would have
Paula Chiocchi: They don't know.
John Kosturos: no idea. Yeah.
Paula Chiocchi: I even wonder if the i i the people that are managing the data really understand it all either.
John Kosturos: Yeah.
Paula Chiocchi: it's just like they pile a bunch of data in and sometimes they don't even know what they have, right? And that's one of the things
John Kosturos: Exactly.
Paula Chiocchi: that you do is help them sift through that and and
John Kosturos: Yeah.
Paula Chiocchi: and get get get get the better part of the data that they would need. So 'cause it being TAM, being a total addressable market for maybe the few people that don't know what it means. Just wanna put that in there.
John Kosturos: Yeah.
Paula Chiocchi: so you've you've pursued executive education AI and listening to you so far today, sounds like you're really up on what's going on in business strategies. So what's one thing you've learned about AI that you think more business leaders need to understand?
John Kosturos: It's a good question. you know, we can start with the small business leader. number one is I I I think it's it's very important to start to invest in the technologies in the back end of your business that are able to connect downstream. Right. Like for me, you know, whether it's an internal model hosted on my machine, very secure for for exercises where I don't want my data and my clients data to go into my Claude you know, I've built that out. Whereas like other tasks and data in my organization, I'm I'm very okay with going into Claude and but it won't go there unless you you have the the technology in your stack that will actually interact, integrate. and I just can't tell you how how powerful fully integrated internal data is to building out Agentic use cases that will ultimately save the the business owner, the business leaders, and the end users a lot of time and make them more efficient. I think for the enterprise though, that's another conversation where they should start thinking about, and and many of them are. about using open source models, about hosting them, about licensing their own GPU access and training models on their first party data, because the amount of information that these types of enterprises have for their industry is is highly valuable. And if they harness that the right way and they train their own models, then tomorrow their product may be entirely different and more scalable. But it is important to do some investigation like building your own cloud storage infrastructure, licensing your own GPU access, and powering your own open source models so that you're not just sharing your information openly with some of these foundation models. So that's kind of where I I think things are going.
Paula Chiocchi: So the GPU allows you to do GPU that's a new term for me. So that allows you to do that?
John Kosturos: Absolutely. Yeah. So there are you can license GPUs from AWS and and Azure and and Google, but there's an another market that people don't realize is very large called Neo Cloud, where you can license GPUs and really high performance storage at a fraction of the rate than you would at some of those larger players and and that makes it more affordable. You're getting the same gear, you're getting the same access, but you can use some of those savings to invest in the people, the data and and everything else that you need to to build your models. But yeah, I think the GPUs are really essential to that process.
Paula Chiocchi: It sounds like it goes all the way down to even the small operator then, the small business person. It that Yeah.
John Kosturos: Absolutely.
Paula Chiocchi: So so it kinda allows you to then think in terms of being able to compete, with the larger enterprise.
John Kosturos: Yeah, or at least have your own little vertical of it, right? I used I I followed a a public stock for many years called Constellation Software.
Paula Chiocchi: Okay.
John Kosturos: it's pretty much a legendary stock portfolio of small SaaS vertical businesses. And they I think acquired about 600 or so of these vertical SaaS businesses. So think about the software that manages the tolls on toll highways. You're not gonna
Paula Chiocchi: Okay. Okay.
John Kosturos: like replace the your toll software for 30 years maybe. So the annual
Paula Chiocchi: Okay.
John Kosturos: recurring revenue on a vertical SaaS business in the right vertical can be very valuable over time. Like they've acquired I think over 600 companies. They've only sold one or two of them ever. And their fund has grown at over 30% a year for like 20 or 30 years. And so I see the same opportunity. Like, we don't need to compete with Chat GPT and Claude. They're gonna be amazing. They're like the resources they have behind them are so ridiculous that like we probably can't compete with them, but we can take our knowledge of the world and our knowledge from our business and create what I would call a vertical AI solution. Solution that can either be fed into something that's more foundational or just licensed on demand from consumers, right?
Paula Chiocchi: Well, considering the amount of money that's invested to get Claude and Chat G P T to where it is, why why re why why reinvent the wheel? Right. So
John Kosturos: Absolutely. Yeah.
Paula Chiocchi: okay.
John Kosturos: And that's really the main fight back on something I said earlier that companies should be training their own models. If they're like, well, why should I do that if if they're just gonna beat me anyway? But I I do still think that people don't give their own businesses enough credit. of what data they have that can be valuable. And so you really gotta think about like what do we have? And the more companies that give these AI labs their data for the 250 to 400 4 million, the less proprietary all of our data is gonna get. So you know
Paula Chiocchi: Yeah. Yeah.
John Kosturos: I think that that's kind of a an anomaly. I can't even believe that that's happening, you know?
Paula Chiocchi: I'm I'm glad I'm talking to you today. That's a big thing to learn. I mean figured it was coming, I just didn't know when, you know.
John Kosturos: Yeah.
Paula Chiocchi: so so speaking of AI as it becomes more embedded in the sales and marketing business operations, how important is the quality of the underlying data to its success?
John Kosturos: Immensely important, I would say. Yeah.
Paula Chiocchi: Yeah. Yeah.
John Kosturos: and then it's, you know, like w in our solution, right, we host a lot of rows and columns of data. So that could be a lot of different things. People, businesses, signals, you name it, rows and columns of data. And you know, when we first initially built out the MCP on top of that data. it didn't have a lot of the skills. Even though it was good data, you still have to create skills on top of that data to do specific operations that you might want to to to pull up from the the data set. So Yeah, it really starts with the quality of the information fed into the model. but it you don't stop there. You know, you can add skills to refine any task that you want to basically complete on top of that information. And I think for the novice user, like that's where they really get caught up is like they're not using projects with memory. They're not using skills to refine the model and give it more access to different things. And so they're gonna get, you know. maybe the the clawed deck that doesn't look like your brand and doesn't have cool styling and stuff. but that those are all things that you'll learn as you kind of move forward with your journey.
Paula Chiocchi: So I'm wondering if there use pe people out there that are claiming to be the consultancy for that type of thing, for for AI c then why do they think that they're able to do that when it's all so new for everybody, you know? Yeah.
John Kosturos: Yeah.
Paula Chiocchi: So it's some of it I think is also probably some trial and error, I would think. Right.
John Kosturos: Absolutely.
Paula Chiocchi: So so looking ahead, do you believe AI will fundamentally change how companies acquire, manage and monetize their data? I mean we kinda touched on this, but I th it seems like
John Kosturos: Yeah, yeah I do.
Paula Chiocchi: I mean, that's what you do. I mean that's what your company does. Is help help comp companies monetize
John Kosturos: Yeah, I I think it's
Paula Chiocchi: their data.
John Kosturos: yeah, it's start with aquire right? We're already seeing the data acquisitions start to to change with tools like Zoom Info's GTM Studio or their GTM One or Clay.com, where you can basically combine web scraping, AI research, and third-party data assets to acquire into your system or your ownership. And combining those three things has made data enrichment and list building just so much more powerful. You know, you'd be surprised you'd have a list of a thousand names. Maybe some of them have emails and company associations and others have addresses and phone numbers. And if you just run that through an enrichment tool in their API, you're gonna get like a 20% match rate because it needs certain data elements to match. But if you send that first through like, go Look this person up on LinkedIn and try to find them on the web and then fill in any additional information that I don't have. So that now that at 90% of your file has like first name, last name, company name, a website, maybe not an email yet. You want to enrich that, but they've got you know your job title and your a link to your LinkedIn profile, your meta account. Now try to feed that through the enrichment provider. You're gonna get like a 70, 80% match rate because. It has the input values that are required to actually enrich a piece of data. So the acquisition of data I think has been really powerful. And we're all thinking maybe the next step is that you don't need to enrich data, you just ask the the AI, like for whatever data you're looking for, and we'll just feed it to you, right? the management of data, Somebody I know very closely just created a tool called Live CRM dot ai.
Paula Chiocchi: Okay.
John Kosturos: and so like things deduplication, data normalization, when job people change jobs at your com at a company, like all those operations used to be rules we would set up and tools to try to like a clean the duplicate. Well, now it's like generative AI, learning how
Paula Chiocchi: Mm-hmm.
John Kosturos: to find the duplicate, make sure you choose the right master record, make sure you choose the right field value. When you're merging like 10 of the same people into one record, you can make a lot of mistakes in a rule-based system. But like with an an artificial intelligence that learns more and more how a human would have merged those records, that's so powerful, right? data normalization, we used to create l pre built libraries of NAICS codes to general industries. And like that took so
Paula Chiocchi: Mm-hmm.
John Kosturos: much time. And then you'd have to like workflow that into like a flow step and Salesforce to actually make it work. Now it's like you just tell AI which how you want nascodes and sit codes to map to your own general industry codes, and it's done. It's crazy, right? So a lot of cool stuff.
Paula Chiocchi: It long as you have the right prompts. Right. But but then you also learn from the mess up, so but but
John Kosturos: Yeah, exactly. You could probably mess it up pretty h pretty easily too. That's right.
Paula Chiocchi: but but you mentioned web scraping and it's always been in our industry considered kind of a bad word. Do you think that's changing?
John Kosturos: I think it's definitely changing. it it's still if you say maybe if the the term web scraping still is looked down on, but
Paula Chiocchi: Yeah that
John Kosturos: but that's literally what Google, Microsoft, Meta Open open AI is is Microsoft, but that's how they build these things is they literally scraped the entire web, fed the data in, and trained their models. And so, you know, if the top companies in the world have built businesses on that, you know, it's just part of yeah, it's part of life.
Paula Chiocchi: Why can't everybody else? Yeah. So maybe that's changing 'cause it really was a bad connotation for years, right? and you know,
John Kosturos: Yeah, absolutely.
Paula Chiocchi: even clients would ask you, Do you do web scraping? 'Cause we don't we don't want any of that, you know. And so
John Kosturos: Yeah.
Paula Chiocchi: but to your point, I mean that is what Google and the rest of are doing, so so
John Kosturos: Yeah, I I remember there was a tool at Ringleed, we we acquired a a business called Broadlook Technologies. And Broadlook had a technology that would basically go to a website and find the people that were on the Meet the Team page and then put them into a little spreadsheet. And so it was a form of scraping, but going to business websites and trying to build a list of people and universities and stuff. And back then that data was it was cool. But like, hey, it didn't come with their email addresses or their phone. Like
Paula Chiocchi: Mm-hmm, mm-hmm.
John Kosturos: that wasn't as valuable as today if you could do the same thing and pull back the lists of people that were on the business and then have AI go figure out
Paula Chiocchi: Right.
John Kosturos: what their job titles were and then enrich them with a third party application like that could potentially be a really valuable asset. So it's been around for a long time. It's just now that with AI and other operations you can wrap around the scraping of information that it just become so necessary and a part of what we do you know almost constantly.
Paula Chiocchi: I do agree with that. So let's talk about identity resolution. What do you see for identity resolution and connected customer data as businesses try to reach d decision makers across all the channels? Because that's one of the things we talk about at OMI all the time is being able to reach everybody in the in the the channels that you desire to be in.
John Kosturos: Yeah, I think you know it's funny when I built Ringlead we did a lot of CRM cleansing, marketing automation cleansing and you know, a lot of the use case was what I'd call offline. So like, I need an email, I need a phone number, or you know, I wanna Company data and contact data, and marketers and sellers would do email campaigns and cold calling and LinkedIn campaigns. But when I left ZoomInfo and started the data marketplace for products, we went, we we approached identity graph solutions first. So the online data, right, which is you've got personal identified information, whether it's B2B or B2C. Well, depending on the source you're working with, that could be a couple hundred million US people, or it could be a billion people around the globe. In my world, a couple of hundred million people or a billion people is not that big of a database. But when you link those people to their mobile identifier mobile IDs or their hashed emails, hems or their IP addresses, we could be talking about linkage of tens of billions of records now. And then when
Paula Chiocchi: Mm-hmm. Mm-hmm.
John Kosturos: you think about linking that to their intent data assets that refresh every single day. That could be 30 billion signals every single day linked to the 12 billion devices linked to the 300 million people, right? And that's where the solution we built for infrastructure helps to manage the the identity graph processing. Now w finding the right identity provider is huge for any sales and marketing team. because what it does is it gives you the ability to again total addressable market identified ad budget deployed towards maybe ABM for really high demand type marketing, other brand budget applied to branding exercises on the the rest of the total addressable market. And now we go into display ads, we go into connected TV, we go into maybe Meta or LinkedIn. Ads, maybe we syndicate and email those audiences, right? It's very important at that point to work with the right identity graph to help you to onboard data into those ecosystems and do it at a high rate. But also, many of them have pixels that allow you to resolve identities of the people that are being served the ad or making it to your website, and ultimately being able to track the journey and the attribution across all those channels. Is all done with identity resolution. And that can either be a really powerful process that you get a lot of value from, or if you choose the wrong provider, it can be a tremendous disaster, right? And you could have a lot of mistakes happen from it.
Paula Chiocchi: Well, because I think at least in our experiences a lot have been more probabilistic versus deterministic on the identity side. And we've had experiences in delivering data for programmatic execution where they come back and said, you didn't match anything because the data was like unemployed students and we don't even have that in our files. So they how did that
John Kosturos: Yeah.
Paula Chiocchi: happen? And that's frustrating.
John Kosturos: Absolutely.
Paula Chiocchi: Right. So so mm what is your experience in that probabilistic versus deterministic world?
John Kosturos: Yeah, I I err on the side of deterministic resolution. that that would be my
Paula Chiocchi: Me too.
John Kosturos: first choice. How however, right? there could potentially be a good mix and that's why in my world we built a marketplace and and data agnostic solution where, maybe you start at the top here with the deterministic and you get what you can from it, right? 'Cause not every data vendor's got everything, right? And so deterministic is very important. It's also pretty rare. Right? Like There's you could probably count on one hand how many identity solutions truly have deterministic resolution. They may say they do, but then you've also got the cookie versus the IP resolution. So the the two choices I think that you have to think about when evaluating an identity solution is: is it deterministic? or is it probabilistic? And is it cookie based or is it IP based? And a lot of people are moving towards IP based simply because they think the cookie is going away. I'm not really sure what's gonna happen there. Yeah, I I don't I don't think it's going away. I don't I No.
Paula Chiocchi: Hadn't happened yet. Yeah. They've been saying that for years and still hasn't. Yeah. Yeah. Doesn't seem like it. Yeah. So so coming here to the end, when you look at the next three to five years, what changes in data, AI, or business technology do you think will have the greatest impact on how companies compete? That's a big question.
John Kosturos: Yeah, that's a tough one. You know, I think the big question is, for me at least, is third-party data going to continue to create the same amount of value that it always has? And I I do think that I personally think it will because some companies are are actually they actually have proprietary collection mechanisms, contributory networks, you know, unique ways that they're actually aggregating that data that you know OpenAI can't just scrape off the web. But there's other companies that all they're doing is licensing another company's data and resell repackaging it and reselling it. And
Paula Chiocchi: Mm-hmm.
John Kosturos: so I think like the companies that have the process for building really unique data sets are gonna survive over the next three to five years.
Paula Chiocchi: Mm-hmm. Survived. Okay.
John Kosturos: And then I think that there's gonna be a lot of new businesses that come into play that are prop that pr are providing data that never did before. Right? just think about, if you owned a trucking company, a logistics company that had a route between the northwest of the United States to like northeast of Canada and you're you had trucks going on all sorts of different routes all year long. And now you start putting sensors and GPS in those trucks and you know you're whether you're managing like what is the weather like during this part of the year. Is it gonna be wet? Is it gonna you know are there gonna be trees likely
Paula Chiocchi: Mm-hmm.
John Kosturos: in the road? Like there's just so much data that you collect. That, okay, now you can potentially license that to Tesla and Waymo and all these robotaxi companies that are coming up. And that pot that data could potentially be more valuable than the trucking routes themselves over time. So I do think there's just going to be this wave of companies that are actually getting into data that never used to be in data because there are just way more ways to monetize it now.
Paula Chiocchi: To build it and monetize it, right?
John Kosturos: Yeah.
Paula Chiocchi: Yeah. A lot easier and faster. So final question for you, John. This has been very, very enlightening. If you could give one piece of advice to business leaders trying to build more data driven organizations, what would it be?
John Kosturos: I think it would be Take a little bit of a break investing in your next, I've seen companies, they'll they'll they'll they'll have a hundred SDRs and then they'll go out and get fifteen more. And look that's a million and a half dollar investment or something like that. But they're not investing in RevOps, marketing ops, data science. They're not licensing full data assets from the Dun and Bradstreets, the Zoom Infos, like the the the the big data players that you can actually license full data sets from and build analytics around and incorporate into your CRM and understand your TAM and your SAM and your SOM I'm giving all these acronyms out. Like set up a nice ABM
Paula Chiocchi: Mm.
John Kosturos: work stream and put your ad dollars to use on people that are actually good for your business and So I think it's all about take pull a little bit back from this heavy growth now. We need growth now. and I I understand that can be looked at as really negative, especially if you're part of a public company, but I'm not saying forever. So I pull some of that investment back, put it into RevOps, put it into marketing ops, put it into sales ops. Trust those people. They're smart. They're probably smarter than 95% of the people in the organization. invest. Invest in data technologies, build the foundation, understand your market. And if you do that, what's funny is when you understand your market, that data is in your system, you've refine it down to the targets that you really want to go after. You can align executive decisions to product development, to sales, to marketing, and you can create that curve that everybody's looking for. But a lot of business leaders today, all they want is like that. short-term growth, they're gonna throw their next investment into more sales, more marketing, and they're doing it before they have the information they need to make logical decisions all the way through the
Paula Chiocchi: Information. Yeah.
John Kosturos: organization.
Paula Chiocchi: I've talked to many consultants over the years that that's their primary objective is going there and teach all how to w work together and play in the sandbox together and be on the same trajectory that you just mentioned. So and and it's still a big need. So John, this has really been very enlightening. I learned a lot today and I appreciate your time and for our audience I would imagine they could get a hold of you in LinkedIn and then your company name is Spring Db dot com, right? Diaz and David.
John Kosturos: Spring D B dot Io.
Paula Chiocchi: Oh.io, I'm so sorry. I'm so sorry. Yeah. Yeah.
John Kosturos: It's all good. But yeah, you find me on LinkedIn, that's a great place to connect.
Paula Chiocchi: If they want to continue the conversation. So I greatly appreciate your time, John.
John Kosturos: Absolutely, Paula. I had a great time.
John Kosturos: Thank you. Yeah, you usually when I introduce myself to people not in our space,
Paula Chiocchi: Yes.
John Kosturos: I'm like, yeah, I work with computers. Exactly. You know, for for my
Paula Chiocchi: That makes it this big.
John Kosturos: mom and all her friends that works really well 'cause she wasn't in tech,
Paula Chiocchi: Yeah.
John Kosturos: so you know.
Paula Chiocchi: Yeah, 'cause and then you try to drill down and then the eyes glaze over. So so
John Kosturos: Yeah.
Paula Chiocchi: so you spent much of your career at the intersection of data technology and business growth. What originally drew you into the industry and how has your perspective evolved over time?
John Kosturos: Yeah, so interesting path into data. When I was younger I always thought I was gonna be a professional athlete and you know how that works
Paula Chiocchi: Okay.
John Kosturos: out. So went from that to kind of entrepreneurialism and joined a a family office that we would at the time we were purchasing e commerce dropship businesses. and we were building out the the tech and the sales and the marketing. So you need data for all that. Lucky for us, you know, that's when Salesforce dot com started to
Paula Chiocchi: Okay.
John Kosturos: To take over. And I was young enough at that time where I really loved the idea of being able to work in the cloud from anywhere. And you know, so
Paula Chiocchi: Okay. Okay.
John Kosturos: we would inst install Salesforce and build out e-commerce catalogs and you know sales, marketing. At the time there wasn't a lot of marketing automation. It was like Genius, I think, was the main one then.
Paula Chiocchi: Okay.
John Kosturos: but loved Salesforce, loved CRM and Saw that that market was really exploding and they were building a great community around their partners. So we went to look for businesses to acquire in that space and found a company called Ringlead, which
Paula Chiocchi: Okay.
John Kosturos: at the time was a deduplication and duplicate prevention technology for Salesforce. long story from there, I mean I had some of the best operators in the world basically educating me. in my discovery calls and in my sales calls and customer success and so lucky enough to just kind of be educated by that RevOps slash data science persona for now close to two decades.
Paula Chiocchi: Okay. Well that's a that's an evolution though, right? To go from Salesforce to today. Big evolution. you've described your work as building companies that turn data into revenue and product outcomes. What does that mean in practice and what separates businesses that succeed at that and from those that actually struggle to do that?
John Kosturos: Yeah, so you know, I think that that question can go in a lot of ways, but I'll answer it in a way where we've kind of built our philosophy. you know, there's probably around five to six thousand companies in in our space right now selling real s real data products. But what I think is over the next five to ten years, as AI really starts to take over, that companies who today don't know that they have valuable data that can be monetized in a major way will potentially be joining us all in this data market. And that may become their biggest asset. So I think it it starts at sourcing data. What do you have as a first-party data asset? What are the third party data assets that can enhance and optimize, validate the information so it's refined and usable? and that's the first step. So it's it's sourcing whether it's first party or third party app data, and then being able to interact with it. And that's why we built our technology called InstaDB. Most companies, once data gets big, it gets very hard to manage and interact with. And that's really what our marketplace does is help you to source data, interact with it, and after you're able to interact with it, it's all about integration and then monetization. And that's kind of the wheel that goes you know, I guess to to break it down a little simpler. Every company has what we call a data bank. A data bank. So
Paula Chiocchi: A data bin? Okay. bank. Okay. Yep.
John Kosturos: Just like you have a bank account today, you make deposits in your b your bank account, or you you're pulling money out of your bank account, right? but the deposits in a data bank are like, you do some cleansing and remove duplicates or normalize data or you're getting data in from your customers or you're buying it from third parties. Like everything you do to enhance your data is just another investment into your accounts. And it As you make more and more investment into a bank account, and the balance gets higher and higher, what you can do at that point is you can start lending on the balance, right? And that's how we feel like we can help companies. A lot of the times we come in and it's I need to do better, get better marketing data or sales data, or I need better analytics for my team. And those are all great things to do, and those are great ways to invest in your bank. But ultimately, once you've made enough investment. investment, that's when you can start lending on that balance and productizing your data and monetizing it through this kind of circular motion.
Paula Chiocchi: Well, I think to your point at the beginning of your answer is that a lot of people don't even realize that their data has value. Right?
John Kosturos: Yeah.
Paula Chiocchi: And that that's a big, big point.
John Kosturos: Yeah. I just got an ad on my Facebook today and I've heard a lot about this, but it the AI labs are now reaching out to companies Saying that they'll pay them $250,000 to $4 plus million dollars for their internal workforce data. So hand
Paula Chiocchi: Really?
John Kosturos: over your Google Drive and your Google, your Gmail, and your Slack conversations and your your GitHub portal.
Paula Chiocchi: I haven't seen that. Wow. Wow. Ooh, really gonna be big brother, right?
John Kosturos: Yeah.
Paula Chiocchi: So so looking back at the companies and teams you've you've worked with over the years, what's one leadership lesson that has had the greatest impact on your approach to business today?
John Kosturos: that's a that's a good question. you know i it when we we're in the the kind of private equity family office space, you you purchase a company that you know has product market fit But it requires a bit of rigor and process to be built in to get to the scale that it deserves and to get to the exit that you want. And so I had a lot of I had a lot of kind of mentors brought into that process and and one of them, it's funny, it's like we we basically had to come in in a a dress shirt and a tie and look the part, you know, feel the part. you know, that's he inst instilled a lot of rigor. So metrics around calls, emails, opportunities, pipeline, you know, close one deals. but one of the things that he was really adamant about was you've got twenty meetings on your calendar this week. why not send out a personalized notification twenty four or forty hours before the meeting that has you and the you and your team who's gonna be on the call their pictures and an agenda. And when you get to the call, you have to have a personalized presentation. And you know, you pitch the store before and then the pain and then the product and the solution. and then a recap goes out out after every single call. And you know, just simple little processes that
Paula Chiocchi: Mm-hmm.
John Kosturos: made you stand out in a sales cycle thing like build a business case for somebody who's gonna invest in your solution. But those simple processes, you know, the buyer actually really appreciates. And at the end of the day, when you are trying to sell something of value to a client, they do have to put together a business case. And guess what? None of them like to do it.
Paula Chiocchi: Mm-hmm.
John Kosturos: So if you're the vendor that from start to fet finish has shown that process and at the end of the day produce material that will take time and work away from the the buyer, they're more likely to select you whether or not your product is superior, right? The the process is superior. And luckily enough for us now, all that stuff can be done with AI agents
Paula Chiocchi: Right.
John Kosturos: and we don't have to stay up till midnight the night before. Yeah.
Paula Chiocchi: Midnight every night. It really has changed that, hasn't it? I mean it's it's it's phenomenally just like that. So and and and probably just gonna get bigger and bigger, I would think. So you you gotta hang on to our little seats here because here w that's where we're going.
John Kosturos: Absolutely.
Paula Chiocchi: You've written about the importance of data quality and digital transformation. Why do you think so many organizations still struggle with data accuracy despite having more sophisticated technologies ever as we've talked about, just been talking about AI. Why why do you think they just aren't educated on it or what do you think the problem is?
John Kosturos: Yeah, I think that it's just this issue of data is not the sexy thing that the leaders of a company are thinking about investing in. You know. I've done projects with companies where we're going in to to fix their CRM data and to, you know, build out their total addressable market and get their ABM account strategy in place. And you know, you're in there for 90 days and the only thing that the CEO and the CFO can ask is like, where are the leads? Where are the leads? And
Paula Chiocchi: Yeah.
John Kosturos: they're out there They're like investing in email syndication and outsource teledialing and appointment setting. But so they're willing to to basically funnel a huge amount of investment into salespeople, into marketing, into ads, and the third-party syndication and appointment setting. but Ultimately they're not thinking about the root cause of their issue. And so I think fundamentally it's just where is focus of the executive team who ultimately make the purchasing decision. if you look at the RevOps team, the marketing ops team, the sales ops team, even the data science a little bit, but those operators are typically the most underappreciated individuals in the entire go-to-mark market organization. While probably knowing the most about the data, the most about the life cycle of the customer, and having really great ideas on how to solve strategic problems. And they're typically the ones that are instigating data purchases. But if you look at the budgets given to an a a RevOps professional these days, it's like they want you to cut costs. They want you to go get better analytics. now one thing that is changing though. Is that everyone in their in in their business is trying to figure out how to implement artificial intelligence, AI, you know, create agentic workflows, automation. And that is making data a bit sexier these days. It is
Paula Chiocchi: Mm-hmm.
John Kosturos: driving, you know, there is a dialogue that I've seen a lot saying like you can't build AI on this shaky foundation. Like, if you do, it will fall. But if you build the A AI on a solid foundation of data, the sky's the limit. And so I I feel like over the next two to three years, the executive committee at both small and large businesses are going to finally make the investment deserved in the data quality.
Paula Chiocchi: Mm. So that would be kinda like music to people's ears like ours, right? Finally read by data quality.
John Kosturos: Yeah. I mean the the one challenge I think that is still gonna be faced is it's also a talent issue. Like cleaning data is not easy and you really need to find good people to do it. so I am also hopeful that you know that talent issue is not just gonna go away, but with AI
Paula Chiocchi: They get better. Yeah, yeah.
John Kosturos: Maybe we can supplement the talent into helping at a at a larger scale. Yeah.
Paula Chiocchi: So so there's job security and cleaning data, it sounds like. So
John Kosturos: Yeah.
Paula Chiocchi: not to be afraid of AI. So
John Kosturos: Yeah.
Paula Chiocchi: so so with companies are collecting enormous amounts of information, you and I both know that, but collecting data and creating business value are two very different things. Where do you see the biggest disconnect?
John Kosturos: Yeah, I think the collection of data, it's it's becoming more and more valuable regardless, I guess, again, because AI can do a decent job of navigating through it if if it has the right access. But you know, I think the what really needs to be done at a at a in a better way and at a across more businesses is just the organization of the data, right? my brother is a commercial real estate investor and for many years he would tell me it's like he's older and he would say you could sell your business and if it's if you don't have folders in your email and your your drive and and you don't create like what what we now call that AI brain where it's tagged, it's it's properly organized, you'll end up having to be at that business for many years after you sell it, and you won't be able to sell it for as much. But an
Paula Chiocchi: Okay.
John Kosturos: organized business is sold at a higher rate, and it's also less likely that you're gonna have to stick Around for a long period of time if you don't want to. And so I think that's the same with just data hoarding. It's like you need to make sure that you have processes to tag to know create folder structure and create sense around your data assets. I think Notion and like Airtable have been amazing technologies for the organization of data that will ultimately feed into AI. use cases. so those are two really great resources that we use. And then like make.com can interact with that and you can build really cool stuff off of it. but there's also you know companies now on the web like there's a company called Wint Labs. They're literally scraping the entire web and licensing that data to AI AI labs and that's become their entire business, right? So, you know, there's different levels to hoarding and building data sets, but I think for the average company it's all about like you need to have it labeled, you need to have it tagged, you need to be able to create governance and access control around it so that you can utilize all that information.
Paula Chiocchi: Well, I'm learning a lot today. I this this this last one you just talked about scraping everything and giving it to AI. I haven't heard about that.
John Kosturos: Yeah, no, it's crazy. But they have
Paula Chiocchi: Is yep.
John Kosturos: probably hundreds of petabytes of data, which is just more than we could imagine.
Paula Chiocchi: The computing power. So
John Kosturos: Exactly.
Paula Chiocchi: how should organizations evaluate the true value of their data investments beyond traditional metrics such as things that you and I have heard for years, volume coverage, cost?
John Kosturos: Yeah, I think we do a lot of valuation on data and I think it starts with scoring the the cleanliness of the information. And we'll I guess we'll talk about traditional contact company opportunity data, right? And then we can get into other areas. But it's like, okay, you're looking at acquiring a a business, maybe it's a strategic acquisition. your goal is to to to get new customers or new revenue through the acquisition So the first thing that I would look at is okay, how many accounts do they have in their system? How many contacts do they have in their system? And one of the things I'll tell you right now is that at Ringlead, we probably worked with 10,000 Salesforce orgs from very small high-growth companies to enterprises that are had thousands of users. One thing that I could tell you about the companies that were doing the most revenue the fastest, and this could be like a company that started two years ago and they're just crushing it, is they had millions of contacts in their B2B databases. literally
Paula Chiocchi: Okay.
John Kosturos: anywhere from two to fifty million contacts in their database. And I'm not saying that every company should go out and just acquire hundreds of thousands or millions of contacts, but What I saw that was that that number did make a lot of sense. And maybe they had a really great PLG technology where they could build a database that quickly. But others built community, they built content. there were different ways that they were building out these data sets. So I think, you know, the the the size of the database, the overlap of You know what what if you're going to acquire a business, what do they have that you don't have? Where
Paula Chiocchi: Okay.
John Kosturos: is the delta, the overlap? and then just ultimately there's little things like how many duplicates are in the database? You know, what is the fill rate of the fields in the information? serious decisions, now no longer serious decisions. I
Paula Chiocchi: Right.
John Kosturos: think they're you know, they've been acquired, but they used to say the cost of dirty data on demand gen between a best practice organization with data and a normal organization is about a 60% swing in revenue. And the real the reason behind that is that let's just say you compared two contact databases that both had a million records. Quality to mean means that do there's not a duplicate. So hey, if there's 20% duplicates, you got 800,000 records in one.
Paula Chiocchi: Right.
John Kosturos: hey, job titles and states and countries have a a convention, they're and they're populated with data. So let's say the the bad database is missing 20% of their job titles. And then the other and then there's 20% that just have poor standards. And really building a sales and marketing campaign stems from reporting into your data. Okay. I wanna find people in California with these job titles. That have a business email, right? So if you ran that query over here on that best practice database with pristine information versus the one over here that had slight variations, they hadn't been filling in their data bank, well, you're gonna
Paula Chiocchi: Mm-hmm.
John Kosturos: get a a larger number of participants in every single call campaign, email campaign, advertising campaign. It really does matter.
Paula Chiocchi: Mhm. Yeah. Well, and I think I think people are starting to figure that out a little bit better than they used to. 'Cause w
John Kosturos: Yeah.
Paula Chiocchi: y in terms of like filling out their data records and their C R platforms and all, more so than they did even a few years ago, I think.
John Kosturos: Yeah, there's great enrichment providers and and now workflow enrichment providers where you can combine AI enrichment with multiple vendors and so I think they are getting better at it for sure.
Paula Chiocchi: So it's like like there's no excuse anymore. It's it's a lot easier. Yeah, right. Right.
John Kosturos: Yeah, exactly. The only excuse is the budget, right? Do we have the money to pay for the data?
Paula Chiocchi: And then you have to make the case to the management to get that budget. So
John Kosturos: That's right.
Paula Chiocchi: bad data in, bad data out, right? So when a when a company discovers that its data isn't producing the expected business results, where do you think leadership should begin looking for the answers? Is it going back to the data, the the data the quality of the data in and out or do you s see other things?
John Kosturos: Yeah, I I I think it does start at going back to the, you know, I would say the revenue operations organization, getting some reporting on you know, duplicate percentages, missing data values, valid emails, valid phone numbers, identifying areas where there is missing information. And then there's also just it that's on your first party data, right? Then there's this whole third party data world where if you evaluated with your data science team what percentage of your total addressable market was actually in your system, in your marketing automation, in your nurture campaigns, you'd probably find that there's a higher percentage of non-TAM records in your CRM and marketing system than there are TAM records. and that's a really important next step. It's like what what do you have that's clean? But then now we do some market sizing and figure out what is the universe. and a lot of data vendors will license you their whole what they call data cubes. You take those cubes, you overlay them on your data, you run some simple data warehouse views, and you'll identify like what is the white space of data that I don't have in my campaigns today, I don't have in my system. And then there's w ways to activate that. You don't need to go out and do illegal email spamming and stuff like that. You could work with
Paula Chiocchi: Mm, mm, mm.
John Kosturos: you know, there's a a ton of syndication providers that do have a lot of those records already opted in and they'll send
Paula Chiocchi: Right.
John Kosturos: emails, they'll drive people to your landing page. It's and it's about you. Do you have a community for them to interact? Do you have good content for you know, these these this TAM audience to be driven to? Are you creating any media and funny videos or in informational videos that you can go run ads to those people on their LinkedIn and try to get them to follow you and and and subscribe. So I I do think where a lot of the leaders like if you went to a company right now and a CEO, nine out of ten if you ask them, tell me the data attributes about your ICP. Tell me the size of that market and then tell me how much of that data how many of those people are actually in your system today. They would have
Paula Chiocchi: They don't know.
John Kosturos: no idea. Yeah.
Paula Chiocchi: I even wonder if the i i the people that are managing the data really understand it all either.
John Kosturos: Yeah.
Paula Chiocchi: it's just like they pile a bunch of data in and sometimes they don't even know what they have, right? And that's one of the things
John Kosturos: Exactly.
Paula Chiocchi: that you do is help them sift through that and and
John Kosturos: Yeah.
Paula Chiocchi: and get get get get the better part of the data that they would need. So 'cause it being TAM, being a total addressable market for maybe the few people that don't know what it means. Just wanna put that in there.
John Kosturos: Yeah.
Paula Chiocchi: so you've you've pursued executive education AI and listening to you so far today, sounds like you're really up on what's going on in business strategies. So what's one thing you've learned about AI that you think more business leaders need to understand?
John Kosturos: It's a good question. you know, we can start with the small business leader. number one is I I I think it's it's very important to start to invest in the technologies in the back end of your business that are able to connect downstream. Right. Like for me, you know, whether it's an internal model hosted on my machine, very secure for for exercises where I don't want my data and my clients data to go into my Claude you know, I've built that out. Whereas like other tasks and data in my organization, I'm I'm very okay with going into Claude and but it won't go there unless you you have the the technology in your stack that will actually interact, integrate. and I just can't tell you how how powerful fully integrated internal data is to building out Agentic use cases that will ultimately save the the business owner, the business leaders, and the end users a lot of time and make them more efficient. I think for the enterprise though, that's another conversation where they should start thinking about, and and many of them are. about using open source models, about hosting them, about licensing their own GPU access and training models on their first party data, because the amount of information that these types of enterprises have for their industry is is highly valuable. And if they harness that the right way and they train their own models, then tomorrow their product may be entirely different and more scalable. But it is important to do some investigation like building your own cloud storage infrastructure, licensing your own GPU access, and powering your own open source models so that you're not just sharing your information openly with some of these foundation models. So that's kind of where I I think things are going.
Paula Chiocchi: So the GPU allows you to do GPU that's a new term for me. So that allows you to do that?
John Kosturos: Absolutely. Yeah. So there are you can license GPUs from AWS and and Azure and and Google, but there's an another market that people don't realize is very large called Neo Cloud, where you can license GPUs and really high performance storage at a fraction of the rate than you would at some of those larger players and and that makes it more affordable. You're getting the same gear, you're getting the same access, but you can use some of those savings to invest in the people, the data and and everything else that you need to to build your models. But yeah, I think the GPUs are really essential to that process.
Paula Chiocchi: It sounds like it goes all the way down to even the small operator then, the small business person. It that Yeah.
John Kosturos: Absolutely.
Paula Chiocchi: So so it kinda allows you to then think in terms of being able to compete, with the larger enterprise.
John Kosturos: Yeah, or at least have your own little vertical of it, right? I used I I followed a a public stock for many years called Constellation Software.
Paula Chiocchi: Okay.
John Kosturos: it's pretty much a legendary stock portfolio of small SaaS vertical businesses. And they I think acquired about 600 or so of these vertical SaaS businesses. So think about the software that manages the tolls on toll highways. You're not gonna
Paula Chiocchi: Okay. Okay.
John Kosturos: like replace the your toll software for 30 years maybe. So the annual
Paula Chiocchi: Okay.
John Kosturos: recurring revenue on a vertical SaaS business in the right vertical can be very valuable over time. Like they've acquired I think over 600 companies. They've only sold one or two of them ever. And their fund has grown at over 30% a year for like 20 or 30 years. And so I see the same opportunity. Like, we don't need to compete with Chat GPT and Claude. They're gonna be amazing. They're like the resources they have behind them are so ridiculous that like we probably can't compete with them, but we can take our knowledge of the world and our knowledge from our business and create what I would call a vertical AI solution. Solution that can either be fed into something that's more foundational or just licensed on demand from consumers, right?
Paula Chiocchi: Well, considering the amount of money that's invested to get Claude and Chat G P T to where it is, why why re why why reinvent the wheel? Right. So
John Kosturos: Absolutely. Yeah.
Paula Chiocchi: okay.
John Kosturos: And that's really the main fight back on something I said earlier that companies should be training their own models. If they're like, well, why should I do that if if they're just gonna beat me anyway? But I I do still think that people don't give their own businesses enough credit. of what data they have that can be valuable. And so you really gotta think about like what do we have? And the more companies that give these AI labs their data for the 250 to 400 4 million, the less proprietary all of our data is gonna get. So you know
Paula Chiocchi: Yeah. Yeah.
John Kosturos: I think that that's kind of a an anomaly. I can't even believe that that's happening, you know?
Paula Chiocchi: I'm I'm glad I'm talking to you today. That's a big thing to learn. I mean figured it was coming, I just didn't know when, you know.
John Kosturos: Yeah.
Paula Chiocchi: so so speaking of AI as it becomes more embedded in the sales and marketing business operations, how important is the quality of the underlying data to its success?
John Kosturos: Immensely important, I would say. Yeah.
Paula Chiocchi: Yeah. Yeah.
John Kosturos: and then it's, you know, like w in our solution, right, we host a lot of rows and columns of data. So that could be a lot of different things. People, businesses, signals, you name it, rows and columns of data. And you know, when we first initially built out the MCP on top of that data. it didn't have a lot of the skills. Even though it was good data, you still have to create skills on top of that data to do specific operations that you might want to to to pull up from the the data set. So Yeah, it really starts with the quality of the information fed into the model. but it you don't stop there. You know, you can add skills to refine any task that you want to basically complete on top of that information. And I think for the novice user, like that's where they really get caught up is like they're not using projects with memory. They're not using skills to refine the model and give it more access to different things. And so they're gonna get, you know. maybe the the clawed deck that doesn't look like your brand and doesn't have cool styling and stuff. but that those are all things that you'll learn as you kind of move forward with your journey.
Paula Chiocchi: So I'm wondering if there use pe people out there that are claiming to be the consultancy for that type of thing, for for AI c then why do they think that they're able to do that when it's all so new for everybody, you know? Yeah.
John Kosturos: Yeah.
Paula Chiocchi: So it's some of it I think is also probably some trial and error, I would think. Right.
John Kosturos: Absolutely.
Paula Chiocchi: So so looking ahead, do you believe AI will fundamentally change how companies acquire, manage and monetize their data? I mean we kinda touched on this, but I th it seems like
John Kosturos: Yeah, yeah I do.
Paula Chiocchi: I mean, that's what you do. I mean that's what your company does. Is help help comp companies monetize
John Kosturos: Yeah, I I think it's
Paula Chiocchi: their data.
John Kosturos: yeah, it's start with aquire right? We're already seeing the data acquisitions start to to change with tools like Zoom Info's GTM Studio or their GTM One or Clay.com, where you can basically combine web scraping, AI research, and third-party data assets to acquire into your system or your ownership. And combining those three things has made data enrichment and list building just so much more powerful. You know, you'd be surprised you'd have a list of a thousand names. Maybe some of them have emails and company associations and others have addresses and phone numbers. And if you just run that through an enrichment tool in their API, you're gonna get like a 20% match rate because it needs certain data elements to match. But if you send that first through like, go Look this person up on LinkedIn and try to find them on the web and then fill in any additional information that I don't have. So that now that at 90% of your file has like first name, last name, company name, a website, maybe not an email yet. You want to enrich that, but they've got you know your job title and your a link to your LinkedIn profile, your meta account. Now try to feed that through the enrichment provider. You're gonna get like a 70, 80% match rate because. It has the input values that are required to actually enrich a piece of data. So the acquisition of data I think has been really powerful. And we're all thinking maybe the next step is that you don't need to enrich data, you just ask the the AI, like for whatever data you're looking for, and we'll just feed it to you, right? the management of data, Somebody I know very closely just created a tool called Live CRM dot ai.
Paula Chiocchi: Okay.
John Kosturos: and so like things deduplication, data normalization, when job people change jobs at your com at a company, like all those operations used to be rules we would set up and tools to try to like a clean the duplicate. Well, now it's like generative AI, learning how
Paula Chiocchi: Mm-hmm.
John Kosturos: to find the duplicate, make sure you choose the right master record, make sure you choose the right field value. When you're merging like 10 of the same people into one record, you can make a lot of mistakes in a rule-based system. But like with an an artificial intelligence that learns more and more how a human would have merged those records, that's so powerful, right? data normalization, we used to create l pre built libraries of NAICS codes to general industries. And like that took so
Paula Chiocchi: Mm-hmm.
John Kosturos: much time. And then you'd have to like workflow that into like a flow step and Salesforce to actually make it work. Now it's like you just tell AI which how you want nascodes and sit codes to map to your own general industry codes, and it's done. It's crazy, right? So a lot of cool stuff.
Paula Chiocchi: It long as you have the right prompts. Right. But but then you also learn from the mess up, so but but
John Kosturos: Yeah, exactly. You could probably mess it up pretty h pretty easily too. That's right.
Paula Chiocchi: but but you mentioned web scraping and it's always been in our industry considered kind of a bad word. Do you think that's changing?
John Kosturos: I think it's definitely changing. it it's still if you say maybe if the the term web scraping still is looked down on, but
Paula Chiocchi: Yeah that
John Kosturos: but that's literally what Google, Microsoft, Meta Open open AI is is Microsoft, but that's how they build these things is they literally scraped the entire web, fed the data in, and trained their models. And so, you know, if the top companies in the world have built businesses on that, you know, it's just part of yeah, it's part of life.
Paula Chiocchi: Why can't everybody else? Yeah. So maybe that's changing 'cause it really was a bad connotation for years, right? and you know,
John Kosturos: Yeah, absolutely.
Paula Chiocchi: even clients would ask you, Do you do web scraping? 'Cause we don't we don't want any of that, you know. And so
John Kosturos: Yeah.
Paula Chiocchi: but to your point, I mean that is what Google and the rest of are doing, so so
John Kosturos: Yeah, I I remember there was a tool at Ringleed, we we acquired a a business called Broadlook Technologies. And Broadlook had a technology that would basically go to a website and find the people that were on the Meet the Team page and then put them into a little spreadsheet. And so it was a form of scraping, but going to business websites and trying to build a list of people and universities and stuff. And back then that data was it was cool. But like, hey, it didn't come with their email addresses or their phone. Like
Paula Chiocchi: Mm-hmm, mm-hmm.
John Kosturos: that wasn't as valuable as today if you could do the same thing and pull back the lists of people that were on the business and then have AI go figure out
Paula Chiocchi: Right.
John Kosturos: what their job titles were and then enrich them with a third party application like that could potentially be a really valuable asset. So it's been around for a long time. It's just now that with AI and other operations you can wrap around the scraping of information that it just become so necessary and a part of what we do you know almost constantly.
Paula Chiocchi: I do agree with that. So let's talk about identity resolution. What do you see for identity resolution and connected customer data as businesses try to reach d decision makers across all the channels? Because that's one of the things we talk about at OMI all the time is being able to reach everybody in the in the the channels that you desire to be in.
John Kosturos: Yeah, I think you know it's funny when I built Ringlead we did a lot of CRM cleansing, marketing automation cleansing and you know, a lot of the use case was what I'd call offline. So like, I need an email, I need a phone number, or you know, I wanna Company data and contact data, and marketers and sellers would do email campaigns and cold calling and LinkedIn campaigns. But when I left ZoomInfo and started the data marketplace for products, we went, we we approached identity graph solutions first. So the online data, right, which is you've got personal identified information, whether it's B2B or B2C. Well, depending on the source you're working with, that could be a couple hundred million US people, or it could be a billion people around the globe. In my world, a couple of hundred million people or a billion people is not that big of a database. But when you link those people to their mobile identifier mobile IDs or their hashed emails, hems or their IP addresses, we could be talking about linkage of tens of billions of records now. And then when
Paula Chiocchi: Mm-hmm. Mm-hmm.
John Kosturos: you think about linking that to their intent data assets that refresh every single day. That could be 30 billion signals every single day linked to the 12 billion devices linked to the 300 million people, right? And that's where the solution we built for infrastructure helps to manage the the identity graph processing. Now w finding the right identity provider is huge for any sales and marketing team. because what it does is it gives you the ability to again total addressable market identified ad budget deployed towards maybe ABM for really high demand type marketing, other brand budget applied to branding exercises on the the rest of the total addressable market. And now we go into display ads, we go into connected TV, we go into maybe Meta or LinkedIn. Ads, maybe we syndicate and email those audiences, right? It's very important at that point to work with the right identity graph to help you to onboard data into those ecosystems and do it at a high rate. But also, many of them have pixels that allow you to resolve identities of the people that are being served the ad or making it to your website, and ultimately being able to track the journey and the attribution across all those channels. Is all done with identity resolution. And that can either be a really powerful process that you get a lot of value from, or if you choose the wrong provider, it can be a tremendous disaster, right? And you could have a lot of mistakes happen from it.
Paula Chiocchi: Well, because I think at least in our experiences a lot have been more probabilistic versus deterministic on the identity side. And we've had experiences in delivering data for programmatic execution where they come back and said, you didn't match anything because the data was like unemployed students and we don't even have that in our files. So they how did that
John Kosturos: Yeah.
Paula Chiocchi: happen? And that's frustrating.
John Kosturos: Absolutely.
Paula Chiocchi: Right. So so mm what is your experience in that probabilistic versus deterministic world?
John Kosturos: Yeah, I I err on the side of deterministic resolution. that that would be my
Paula Chiocchi: Me too.
John Kosturos: first choice. How however, right? there could potentially be a good mix and that's why in my world we built a marketplace and and data agnostic solution where, maybe you start at the top here with the deterministic and you get what you can from it, right? 'Cause not every data vendor's got everything, right? And so deterministic is very important. It's also pretty rare. Right? Like There's you could probably count on one hand how many identity solutions truly have deterministic resolution. They may say they do, but then you've also got the cookie versus the IP resolution. So the the two choices I think that you have to think about when evaluating an identity solution is: is it deterministic? or is it probabilistic? And is it cookie based or is it IP based? And a lot of people are moving towards IP based simply because they think the cookie is going away. I'm not really sure what's gonna happen there. Yeah, I I don't I don't think it's going away. I don't I No.
Paula Chiocchi: Hadn't happened yet. Yeah. They've been saying that for years and still hasn't. Yeah. Yeah. Doesn't seem like it. Yeah. So so coming here to the end, when you look at the next three to five years, what changes in data, AI, or business technology do you think will have the greatest impact on how companies compete? That's a big question.
John Kosturos: Yeah, that's a tough one. You know, I think the big question is, for me at least, is third-party data going to continue to create the same amount of value that it always has? And I I do think that I personally think it will because some companies are are actually they actually have proprietary collection mechanisms, contributory networks, you know, unique ways that they're actually aggregating that data that you know OpenAI can't just scrape off the web. But there's other companies that all they're doing is licensing another company's data and resell repackaging it and reselling it. And
Paula Chiocchi: Mm-hmm.
John Kosturos: so I think like the companies that have the process for building really unique data sets are gonna survive over the next three to five years.
Paula Chiocchi: Mm-hmm. Survived. Okay.
John Kosturos: And then I think that there's gonna be a lot of new businesses that come into play that are prop that pr are providing data that never did before. Right? just think about, if you owned a trucking company, a logistics company that had a route between the northwest of the United States to like northeast of Canada and you're you had trucks going on all sorts of different routes all year long. And now you start putting sensors and GPS in those trucks and you know you're whether you're managing like what is the weather like during this part of the year. Is it gonna be wet? Is it gonna you know are there gonna be trees likely
Paula Chiocchi: Mm-hmm.
John Kosturos: in the road? Like there's just so much data that you collect. That, okay, now you can potentially license that to Tesla and Waymo and all these robotaxi companies that are coming up. And that pot that data could potentially be more valuable than the trucking routes themselves over time. So I do think there's just going to be this wave of companies that are actually getting into data that never used to be in data because there are just way more ways to monetize it now.
Paula Chiocchi: To build it and monetize it, right?
John Kosturos: Yeah.
Paula Chiocchi: Yeah. A lot easier and faster. So final question for you, John. This has been very, very enlightening. If you could give one piece of advice to business leaders trying to build more data driven organizations, what would it be?
John Kosturos: I think it would be Take a little bit of a break investing in your next, I've seen companies, they'll they'll they'll they'll have a hundred SDRs and then they'll go out and get fifteen more. And look that's a million and a half dollar investment or something like that. But they're not investing in RevOps, marketing ops, data science. They're not licensing full data assets from the Dun and Bradstreets, the Zoom Infos, like the the the the big data players that you can actually license full data sets from and build analytics around and incorporate into your CRM and understand your TAM and your SAM and your SOM I'm giving all these acronyms out. Like set up a nice ABM
Paula Chiocchi: Mm.
John Kosturos: work stream and put your ad dollars to use on people that are actually good for your business and So I think it's all about take pull a little bit back from this heavy growth now. We need growth now. and I I understand that can be looked at as really negative, especially if you're part of a public company, but I'm not saying forever. So I pull some of that investment back, put it into RevOps, put it into marketing ops, put it into sales ops. Trust those people. They're smart. They're probably smarter than 95% of the people in the organization. invest. Invest in data technologies, build the foundation, understand your market. And if you do that, what's funny is when you understand your market, that data is in your system, you've refine it down to the targets that you really want to go after. You can align executive decisions to product development, to sales, to marketing, and you can create that curve that everybody's looking for. But a lot of business leaders today, all they want is like that. short-term growth, they're gonna throw their next investment into more sales, more marketing, and they're doing it before they have the information they need to make logical decisions all the way through the
Paula Chiocchi: Information. Yeah.
John Kosturos: organization.
Paula Chiocchi: I've talked to many consultants over the years that that's their primary objective is going there and teach all how to w work together and play in the sandbox together and be on the same trajectory that you just mentioned. So and and it's still a big need. So John, this has really been very enlightening. I learned a lot today and I appreciate your time and for our audience I would imagine they could get a hold of you in LinkedIn and then your company name is Spring Db dot com, right? Diaz and David.
John Kosturos: Spring D B dot Io.
Paula Chiocchi: Oh.io, I'm so sorry. I'm so sorry. Yeah. Yeah.
John Kosturos: It's all good. But yeah, you find me on LinkedIn, that's a great place to connect.
Paula Chiocchi: If they want to continue the conversation. So I greatly appreciate your time, John.
John Kosturos: Absolutely, Paula. I had a great time.
