Mindful B2B Marketing | Business Growth and Social Impact (Former: Forward Launch Your SaaS)

S2E8: How to leverage AI in B2B Marketing -- ft. Rich Edwards, CEO of Mindspan Systems

• Keirra Woodard • Season 2 • Episode 8

Share episode

MAIN INSIGHT: Leverage data-driven strategies and artificial intelligence to transform and optimize marketing efforts.

GUEST BIO: Rich Edwards is the CEO of Mindspan Systems, a company specializing in transforming community financial institutions through data-driven strategies and innovative technologies. With a career spanning over a decade at IBM, Edwards has significant experience in product strategy, data center automation, and artificial intelligence, including a pivotal role in the launch of the IBM Watson Developer Cloud. 

RESULTS: 

  1. Increased personalization in marketing, allowing companies to cater specifically to individual customer preferences and needs.
  2. Enhanced efficiency and effectiveness in marketing operations, as AI can automate and optimize many routine tasks and processes.
  3. Improved customer engagement and retention through targeted and relevant marketing communications.
  4. Greater competitive advantage by utilizing data insights to make informed strategic decisions.
  5. Cost savings from automation and the ability to produce more with less manual effort.


KEY TAKEAWAYS: 

Leverage First-Party Data: Utilize first-party data to gain a competitive advantage by delivering highly personalized and relevant marketing content. First-party data, being directly collected from customers, is crucial for creating accurate and impactful marketing strategies that resonate with the target audience.

Ethical Data Practices: Adopt transparent data collection methods and manage data prudently to uphold customer trust. Rich Edwards stressed the importance of treating customer data with the same care as money, considering the potential risks and implications of data breaches or misuse.

Data Safeguarding and Compliance: Implement robust systems to protect customer data, ensuring that data management practices comply with the latest regulations. This includes thoughtful interactions with customers about how their data is used and ensuring that data sharing and processing are done only when necessary and with proper customer consent.

Navigating Data Governance Challenges: Be vigilant about the dependencies that can develop from using platforms like Facebook for marketing, which may lead to potential constraints and costs. Marketers should be aware of the terms and conditions of data usage by third-party companies and strive for agreements that protect their interests and those of their customers.

LINKS TO DATA GOVERNANCE RESOURCES:

  • Here’s a good, easy-to-read introduction to data governance from HubSpot: HubSpot Data Governance.
  • One of the more general industry governance organizations is the Data Management Association (DAMA), offering good resources and certification standards: DAMA.
  • Microsoft’s policy on data usage with AI services can be found here: Microsoft Data Privacy.
  • An earlier interview with Microsoft CEO Satya Nadella specifically addressing their values on privacy is available on YouTube: Satya Nadella Interview.
  • Also, as discussed, Trust Insights provides insights, and here's an example of Chris Penn’s blog:

Give feedback on this episode by sending the host a text message.

SIGN UP FOR THE EMAIL NEWSLETTER at forwardlaunchdigital.com/podcast

[0:32] All right. Today, I am sitting down with Rich Edwards. He is CEO of Mindspan Systems, which helps community financial institutions transform themselves with data-driven strategies and technologies. Rich has worked in the banking, insurance, and financial services clients worldwide throughout his career. And he's led product strategy in data center automation information and artificial intelligence. And he was involved in the launch of IBM Watson Developer Cloud. So Rich, I'm super excited to chat with you and get to know a little bit about your story. Yeah, no, thanks for having me, Kira. I'm glad to be here. Yeah. So tell me a little bit like why you got into doing what you're doing now with Mindspan and like what kind of led up to this point in your career. Yeah, yeah. It was a very unlikely path that I took. I was a long-time IBMer. I was there for 13 years overall and spent a lot of time in software product management. So it was...

[1:39] That kind of space between business requirements and like technically what you're going to do from a roadmap standpoint. So it was all about like solving problems. And I worked in part of the business that was in big data center, big enterprise level deployments and, and, and automation at that level and spent a lot of time with government, but more financial services, Services, banks, insurance companies.

[2:07] Pseudo-government organizations that do healthcare administration, things like that, where there was volumes and volumes of information, and it was primarily a data and transactional-based business. And I did that for a long time and worked with a lot of clients, obviously a lot of banking clients. And in 2013, I got tapped to go be part of this new business unit they were creating that turned into IBM Watson around artificial intelligence. And frankly, I didn't have any business being there from a technology standpoint, but I was kind of like the guy that had been around and understood how to essentially bring products to market within a giant organization like IBM. And so I jumped into it with both feet and ended up spending four years just drinking thinking from the biggest fire hose I've ever seen, from understanding where this was going.

[3:05] Why machine learning had hit this inflection point at that point. And again, we're talking 10 years ago, right? So it's since then, even more so. But then also how it was really beginning to transform business. And because I had been the bank guy in the last job, I became the bank AI guy that went around to Wall Street and government organizations and a lot of financial institutions throughout the world to talk about this is where it's at and what we're doing and what are some interesting use cases. And so it was a great experience to kind of have this front row seat to.

[3:44] At least in my lifetime, was like a big transformation in this thing that we were doing and the whole world was going to be different after that. So it was a great experience And I did that for a little over four years. And I kind of got to this point where I wasn't in my career of, you know, I always kind of wanted to do the entrepreneur thing and kind of run my own show. And, you know, for me, I was kind of getting into my mid forties and it was like, look, you're only going to get so many bites of the apple here. Like you kind of need to, you know, do it or don't at this point in your life. So, so I did, I, I, I moved on and I was looking for, you know, a small company to either join or potentially start. And, you know, my criteria, what I was looking for was saying, okay.

[4:27] Data is going to be a really big part of what's coming and how, as machine learning and artificial intelligence matures even more and there's more mainstream adoption, the data part is going to be the really interesting part and the part where there's going to be a lot of value. So I went looking for a company that worked in that space and that eventually led me to Mindspan Systems where they had had like, at that point, I think about 16 or 17 years of doing data engineering, data science work, but particularly working with companies where data was really valuable, regulated industries, healthcare,

[5:06] financial services, banking, some government work. And had this core capability, core competence around dealing with valuable information, particularly where compliance is a really big deal. You talk to a bank and how they handle data is very different than how organizations, even of the same size and complexity, handle data because they have this heavy, heavy stick from a regulatory standpoint. So that's kind of what led me to Mindspan Systems and joining the company.

[5:41] And, you know, we've kind of been like riding that wave ever since. So, okay. Well, yeah, that's very interesting. So what would you say is...

[5:55] About ai so it as it relates to like b2b marketing how do you believe it's like changing the landscape or like what's your b2b marketers be aware of yeah no no absolutely you know i mean the headline answer is well it's generative ai it's chat gpt it's anthropic it's companies like that that are making you know text and image generation you know things like dolly mid journey, all these things that are like, you know, from a marketer standpoint, asset creation, the friction from that, you know, went way down to like zero. Yeah. Very, very easy, easy to use. And, and that's one very clear way. And, and one, you know, certainly in the past year that that's taken off.

[6:39] And, you know, I would say from a marketer standpoint.

[6:44] I mean, this is, this is This is very similar to what you saw in the 80s where desktop publishing really changed the way graphic arts and design was done. And you saw this ability of a design ethos suddenly got applied to a way larger surface area than it was before. Way more companies could afford to have professionally designed brochures and materials and logos and it just kind of like generally upped the game for everybody because it became so accessible and you know obviously that the macintosh was a big part of that which just hit i think the 50th in is that right i think was it 40th a big big milestone it just hit like this suite and and then a lot of like programs i'm dating myself kind of remembering quark express and kind of a lot of the products that eventually turned into like the adobe like creative suite that you know today like the nascent version of that and it it just made this like it democratized right instead of having to have a full-time school trained graphic designer on staff you now had this box that anybody with a little bit of training and a little bit of work could get something out of. I mean, quality would vary, but that was it. And so we're kind of at that.

[8:10] Place right now where what generative AI is, is making available to you. It's, it's suddenly, you know, what, even at a relatively low cost, like stock photography, how, how many more use cases are there where you could have a very relevant Apple book image as part of what you're doing, whether it's a blog post or email or your marketing collateral or your presentations, like that game is suddenly getting a lot better across the board. Same thing for, for copywriting. You're seeing that diffuse itself into many places where it was hacky or very amateurish. That game is getting lifted up more. All of this is to say that the expectations from the market about what you're going to do, how you're going to present your brand or your company, there's going to be a higher expectation of that because all these tools are broadly available. And just that reaping expectation that customers, particularly consumers, are going to have, you're going to have to meet that. So from a certain extent, the job got easier, but the expectations for what you're going to deliver have gotten higher. Hmm. Okay.

[9:27] Well, in terms of what would you recommend companies actually do now that the AI landscape looks like this and all these new tools are available? How can you use that to gain a competitive advantage using AI? Yeah. No, no, no. That's a great question. And one everyone is struggling with, right? So, I would say there's two frames of reference to think about here, right? One is...

[10:00] I kind of want it like that elevating expectation and having access to more tools. There's an element of, you kind of have to keep up with the Joneses on that, right? You're, you're, you're going to have to meet the rising expectations and probably take advantage of some cost savings along the way. And that's just like, get used to like using a new tool and probably doing business a little bit different, but you're not fundamentally doing anything differently, right? Things get nicer. you're able to do it a little faster. It's going to be cheaper, but it's all kind of incremental better versions of what you were already doing.

[10:36] And that's kind of this vein where you start to get this sense of the fear that's out there of, you know, is AI going to take my job? And, you know, kind of the pat answer is no, AI isn't going to take your job yet, but certainly someone using AI to get those efficiencies to be able to provide that material cheaper and faster, that's the bigger threat to what you're doing, right?

[11:05] So you kind of have to play the game on that. But that's an incremental view, right? What I'm doing today, a little bit better, a little bit nicer, a little bit faster, a little bit cheap, right? That's winning. The other one is, what aren't you doing today because practically it doesn't make sense or it's too expensive or it's too slow, right? So one area, particularly in marketing, you kind of think about is personalization.

[11:34] How what you do for your customers or your prospects is specifically tailored to them. Now, most marketers will have a sense of segmentation. They'll have maybe a set of personas that they're selling to. This type of person, demographic information, where they live, what they like, stuff they like to read. And they'll just kind of generally tailor it to that, to one type of person. But that persona is really a stand-in for a much larger group. And that's a best practice and certainly something you'll see broadly used at every level of marketing. The individual small business all the way up to the enterprise level. Well, the question becomes, well, what if you had more segments? Because you know the market is more complicated than that. It's more nuanced.

[12:28] What if you had segments that were literally segments of one that you knew each individual person you were talking to, not just your customers, but your prospects as well? What do you know about them? Maybe not a lot, but what do you know about them? And how could you use that to provide a more tailored, more relevant, more personalized experience to them? Whether that's the email you're sending them or anything from a messaging standpoint to the very experience they have with your business or service. Come into the store talk to the associate once you know who that is you've identified them because you have a relationship with them or whatever how do you then change that experience in a way that's going to be better for them now for most companies that it's like okay great that kind of sounds very much like what you saw in the movie minority report right where they're like reading eyeballs balls and doing things, which most people found pretty creepy and rightfully so. But you begin to think about, well, if I have this amazing machine that can write very, very good copy, at what point do we get where anything that's written down and presented to someone is only written for them? It's a one of one.

[13:50] Every email, every message, every piece of collateral that gets printed out and sent to them in the mail is a single one-off piece of material that's sent to them. Now, that's an area that just beginning to see the edge of that now, just beginning to see like where those potentials are. But you think about that, and we're just talking about marketing. We're talking about like messaging and collateral on it. Start to think about the way that you actually deliver your business, how you treat them, what you do, the terms about how you deliver that, ways that you can just absolutely custom one-off do what you're doing for this particular client. And again, in a lot of ways, it's impractical today. It doesn't make sense. It may require too much overhead, too much risk to manage, whatever, whatever. A lot of those excuses about like, I can't do it because it's too hard. Those begin to shrink away when you begin to like apply some of these tools to that. And I will just say, I'm talking very broadly here. I'm not just narrowly talking about generative AI. There's a lot of other elements to this. It's, there's a big tent when you start talking about what machine learning and AI can do.

[15:04] Okay. So one thing that's like fascinating, but like one thing that kind of strikes me about it is like, on the one hand, people really want this because it would be great. Like if you had something like that one-on-one interaction that you get, if you're actually talking to another person and they get to know you over time, but on the other hand, And it coming from like a machine or like this data set that this company is collecting is where it kind of feels like off-putting or creepy to people. And like, you know, the data management behind that, like if you have so much information about a particular person, you know, that could be used for purposes that are not beneficial to that person. Yeah. No, no, absolutely. Yeah. This is like one of those things. I think there's two good points you made in here. Like one is.

[15:56] There's a good and a bad version of this. And I think you're absolutely right. And the way to think about it and the way that I think is missing from a lot of the discussion about how AI and machine learning should be regulated is the idea that it's inherently good or bad.

[16:13] And I would say that's probably not the right way to think about it because outside of some of the philosophical elements about will AI destroy us all and the existential risk, which I'm a little bearish on. It's not zero, but it also has this big thing and it has to be addressed and you have to think about it. But that begins to imply this is bad in and of itself. I think the correct way is to say it is neither good nor bad, but how it is used can be good or bad, the specific use cases. I'll give you one example that is forefront, particularly from a regulatory standpoint in banking is credit decisions that are made. Whether you're going to be offered a loan and at what rate and what the terms are, a lot of times is programmatic. There's an algorithm that figures that out. And as AI works its way into it, the thing that regulators rightfully are saying is, you can't have a black box approach. You can't say, well, well, I put all these factors in and then it did something behind the scenes and then told me yes or no, or told me it's 6% versus 5%, right? Because inherently you can't.

[17:32] Make sure you don't have inherent biases in what you're doing, particularly for something like that, credit decisions, access to money, that is so fundamentally important to how our economy and our society works. That has to have this level of transparency and trust from a societal standpoint, not just this transactional element between people. It has to be there that you can show, No, no, no, no. We don't have an inherent bias against certain class, certain socioeconomic, certain race, certain background, whatever you want to pick from a class standpoint or how they think about protected class standpoint. You don't want to have that suddenly behind this veil of you don't know what's going on, right? So there's an element of this is useful, it is helpful, but it has to be used in a way that's beneficial to everybody involved.

[18:27] So that's what? The white hat, black hat part, right? I think the other element from a how data gets used standpoint is really focusing the efforts and how companies and service providers look at their markets to have more empathy for their customers, for their market. They kind of understand them in a more specific way, in a way that helps them serve them better. So we'll go back to that example of the persona. You can kind of talk about who you serve for your services or your products, who you serve and who this is for in a very kind of generic way.

[19:21] But inherently, you understand you're taking shortcuts in that, right? There's things that you're assuming, you're trying to make some generalities. And probably for what you're doing, that's going to be good most of the time. But you're also going to be missing things. There's things you're leaving out for the sake of being able to offer this thing efficiently, right? And thinking about when you begin to have tools like this that give you this higher sense of insight, higher sense of ability to analyze and understand the situation, using that in a way to provide better services, not better, just cheaper, but more relevant, more pertinent to this person as an individual. All the little areas that you could begin to do things in a more personalized, specific, custom way that don't make sense today because it's too hard or it's too expensive or it's too slow, now you start to think about what can I do that I can't do today? And I give you a quick framework on this, right? Like a way to think about it. And to the point of that creepy ick fact, right?

[20:38] One conception, and this has been going on as long as I've been talking about this with people is there's a very quick mental leap that goes to artificial intelligence is like this giant monolithic thing, right? Everybody thinks about the Terminator or Hal or whatever science fiction monster you want to think of that aspect. They kind of think of it that way. And to a certain extent, in the whole, when you step back, there's some truth to that, that it's like things working together. But in reality, the way it works and the way it functions, it's like hundreds of very, very small, discrete capabilities that are put together. And so the way when I start talking to clients about use cases and how to put this together, I say, okay, the right way to think about it isn't mega supercomputer. The right way to think about it is a summer intern.

[21:35] A college intern, somebody who just shows up from school to kind of work for you for the purpose of getting experience for a couple of months, right? They're energetic. They're really excited to be there. They're literate. They're going to show up and do exactly what you tell them to do, but they don't know anything about your business. They don't have any experience or judgment about what you do and how you do what you do and your customers and your market. it. But if you give them simple enough direct instructions, they will execute it to a T. They're great.

[22:07] Here's an industry report on my market. Read through that and give me a summary. Tell me what the highlights were. Here's 10 potential customers we might look at. Take our persona and tell me which one's a better fit and which one isn't. Here's the last three analyst reports that look at my competitors. Tell me what's the gap between what we have and what they have and where that is. And all of those things in isolation, they will do a great job on. They won't do a good job on piecing it all together and kind of helping you from a strategy standpoint, at least not initially, maybe months or the second or third summer they're working for you, they begin to start to piece that together. But they can kind of do all those discrete tasks, right? Well, now imagine you didn't have one summer intern, term, but you had 100 or 1,000. And they were really easy to give instructions to. It wasn't any harder than talking to that one person, right? And it's not just analyze this one customer, it's analyze all of them. Sit down and do it. And you all work together in concert in a very efficient manner. There's not a lot of overhead for me to manage this and do it. And in fact, you can just work 24-7 and I don't even have to buy the pizza for you, right? That's a better way to think about this very easy to scale capability that has.

[23:31] These very specific things that it's able to do, and maybe some level of orchestration, and that's getting better as we move along. But generally, it's a bunch of little things that you can do. And the more creative you are about what kind of tasks can I give them? What kind of questions can I send them to go answer? If I can frame it the right way, and if I can put it in the guide rails about what I want them to do and not do, and maybe even have them work together in a certain sense, you can put together some pretty impressive capabilities around that. So like that, that's a way to kind of think about it. That like good versus bad is the one angle. And then the other one is, it's not one big thing. It's a bunch of little things. And so your ability to kind of break down what you need them to do and what you could do if you had a hundred interns to help you for the summer, right?

[24:27] What could you do today that, what could you do with that that you're not doing today? What new capabilities, what way could you really transform what you do for your clients if you had that army of interns available to you?

[24:42] Okay. Wow. Okay. So I guess let's narrow the conversation down a little bit into some some practical aspects of like, how does integrating AI into a marketing campaign.

[25:03] Actually lead to benefits for the company? Like how, how would you go about doing that if you wanted to start integrating AI and what, what are some like.

[25:16] Results that you would expect to see from that? Like if you have a case study or something from a previous company you've done this for? Sure. I can talk about some things in generalities about how to think about it. I think a really good example of this, maybe done in a very focused way, is Facebook ads, right? How that works when you do it well, when you kind of follow the the methodology that's out there for it.

[25:47] Meaning you don't go in there and just plaster up one ad and then just show it to the world, right? You're not going in there to buy a Super Bowl ad where you're going to do it once. What you're doing is I'm going to do maybe eight or nine versions of it. And they will automate this for you to a certain extent, right? Here's one version with this creative. Here's the exact same ad, but the creative is a little different. And now I'm going going to use a different angle in the photo. And then I'm going to do those three or four versions and I'm going to test different headlines. And then I'm going to test different sub-headlines and text under there. And then I'm going to test the call to action. What does the button say? Sign up, call me, visit my website, apply, all those different things. What are those elements in combinations? Which one are important? Which ones aren't important? Which ones in combination nation are more relevant? Are people more likely to click through? Which ones are the people that click through more likely to follow through in the whole process?

[26:47] It's this element of experimentation that you kind of go into it with a sense of what's going to work or what is going to be most effective and efficient to reach my customers. But then you go in and you test it and And you try a bunch of different versions.

[27:04] And really, the element is not that you get it right the first time. It's how quickly can you begin to optimize and go, well, I learned a little bit on this version. Let me show it to a wider audience. And we'll try this. And I'm going to do the red font instead of the black font, right? And kind of go down this iterative path where you're able to objectively look at the results. And this is the nice thing about digital advertising, right? It's so well instrumented. You get this like very quick feedback. And so your learning cycle, your cycle time on conducting experiment, learning what worked and different and being able to go do the other experiment to do that very, very quickly.

[27:44] And so having that data available to you about, I did X and this was Y response, the more you can integrate that into the rest of your business and the faster you can go to learn, the quicker you can hone your offering, whether it's as simple as your messaging or subject lines in your email or your entire campaign or the features and functions that you're actually offering customers,

[28:10] the faster you do that, the quicker you get. And I'll tell you, I mean, there's myriad examples about attribution and, you know, doing campaigns, but I'll tell you a more practical one from IBM Watson. We rolled this out. So it was a cloud. The offering that we actually had was a cloud service offering, which meant you could, it was largely for developers where they could show up and they could take some of the capabilities that Watson had, particularly from natural language processing standpoint, and integrate it into their products. Right. So be able to do massive text summarization, sentiment analysis, entity extraction, things like that. Things that would kind of be very tedious to do from a manual standpoint, you could integrate it. We had a great way to do it. And then we rolled it out worldwide in a phased manner because we had to do localization from.

[29:04] How things went to market, money that we would accept, where the platform was available, and largely language. language, like language localization. One of the things that we would do is we would watch the usage of the product as it went into a new market. And we would look at how people were using the market. And we would like, look at things like error codes.

[29:26] Like if somebody was using the product and they would get back these like different types of error codes and we, you know, after you do it a couple of times and you're looking at, you know, tens of thousands of customers in each one of these rollouts, that level of data you could look at. You obviously see patterns and you kind of knew what things would happen. So one of the things we would look for was a certain type of error code that said somebody had tried to use the service, but they used it in a way that they didn't ask the right question, if you want to think of that from a programming standpoint. This is code stuff, so it's a little nerdy, but we could look at that. And we knew generally what the percentage of those would be. So if we went into a new market and we saw suddenly that type of error code spike, we knew we had a problem. And generally, it was going to be in our documentation, in the way that we were teaching customers to use our product. That if we went into a new language or a new region and saw the sudden spike in errors, number one, that's a frustrated customer who's not getting what they want. But number two, we knew there's something about the way we're showing them how to use the product that we're messing up.

[30:42] Doing something wrong because they should be getting it faster. And so that gave us this very quick feedback that said, you need to fix what you're doing. You need to fix the way that you're bringing this product to market, the way that you're teaching your customers to get value out of it for this market, for these specific customers. You've messed up somehow, right? Because you went into it with an assumption about what you were doing and they, through their use of what they were doing, not directly.

[31:12] But just through log files, error files, they were telling us that we were making a mistake, we were screwing up. And it allowed us to very, very quickly make those fixes and provide a better, more relevant service to them. So that whole idea about using the data that you have, using the data that is being generated, even if it's just by usage of your product, not necessarily like feedback or customer reviews. use. I mean, that's a no-brainer obviously there, but just the feedback of the product, being able to use that to make the product better, that's very much one of those like white hat use cases, right? How quickly can you iterate and improve on what you're doing based on the information that's given to you, right? And using a lot of the tools that are available via machine learning, via AI, helps you do that, helps you utilize them at a larger scale. And certainly use data that maybe isn't easy to get to, to analyze.

[32:12] I have a sidebar here on structured and unstructured data, but I'll let you ask your next question.

[32:18] Yes, that is, yeah, it's fascinating to hear about like how AI has been used

[32:24] like historically, because IBM Watson was like one of the leaders for a long time in the AI space. So yeah. And it's interesting to see how that's coalescing into what business leaders can do in the modern day. Let me just make, I'll just tell you one kind of funny anecdote on this. There's always a discussion about, well, what is AI and how is AI different than machine learning? And the insider joke is machine learning is anything that we can do today. AI is anything we haven't figured out how to do yet. Meaning the distinction between the two really isn't there. Like it's more the thing and then how do you do the thing? And so the difference between those two terms is way fuzzier than everybody thinks it is. Okay. Yeah. I tend to think of it a little bit interchangeably. Yeah. That's the correct way to do it. Okay.

[33:20] Okay. So yeah, when we had our pre-conversation, you were talking a little bit about how companies might integrate... Or how companies and marketers might look at the amount of data that they're collecting and be very selective for which parts of their overarching data set about their customers they might want to use in order to feed to a machine learning algorithm or a third-party company. So can you talk a little bit about what's the ideal type of data to be collecting to mitigate maybe some of these privacy and data management issues and how we can be using that with AI systems to get this optimal balance? Right. Yeah. So the big overarching term for this is they usually talk about data governance. You'll hear that term. And that sometimes is a generic term and sometimes it means something very specific, particularly if you're in a regulated industry, right?

[34:27] Compliance is going to be an issue around data. But I think the big thing to take away is data is very quickly becoming a valuable asset, something that may.

[34:42] In ways we don't even completely understand yet, have a value on your business. The data you have is going to directly impact how valuable your business is, Meaning it's definitely the biggest asset you have as a business that isn't on your balance sheet. And so you kind of have to start treating it almost like money, right? That's probably a good analogy, right? That your data stores that you have are almost like a bank account. And you need to start thinking of it and have controls around being very careful and focused on how you treat that. Treat your data like you treat your money. And that's a good kind of just way to begin to think about it. And like I said, there's a lot of like very industry specific things that you kind of get into here when you think about it. But one thing is really clear. We talked about how broadly a lot of the tools are being made available. You know, things like OpenAI, ChatGPT, you know, Anthropic, a lot of the tools like pick Pick your cloud platform, AWS, Microsoft through Azure and their Office 365 offerings and Google Compute Cloud, and everybody else has an offering that's going to help you do this.

[36:00] And one thing that I've seen, and this is somewhat unique to machine learning and AI, is how strong and prevalent open source has been.

[36:14] You look at backing it up past a couple of big technological shifts that have gone on in society and business. We can always go back to electricity, but if we think there was the advent of the mass adoption of the PC, and you got into, well, is it you're going to go PC, IBM, or are you going to go Mac, Apple? Those would be these two walled gardens you could go into. to. And then there was like the internet and that was like this big shift. And then there was mobile and that also had its two, is it going to be Android or is it going to be iOS? These two elements that went in there. And you kind of had to like, a lot of these were, you were picking an ecosystem you were going to play in. That kind of like almost defined your market. And it wasn't until you kind of got big enough that you could start to play in multiple ones and all that. And so now with this, you're seeing how well open source is playing against the incumbents. So the big dog in generative AI is open AI and ChatGPT and a lot of the GPT derivative products that are in there.

[37:27] But you have other offerings that are out there that are open source, that are not tied to any particular vendor, that really have, even at this stage, comparable capability.

[37:41] Maybe not as broad, but in the areas that they play in, they're pretty good. And all of that is to say, it looks like at this point, the thing that's really going to differentiate the offerings, the thing that's going to differentiate what you're we're able to do as a customer is going to have less to do with that algorithmic layer, that layer of, did I pick Anthropic, or did I pick OpenAI, or did I pick Hugging Face, or did I pick another one of these vendors? It's more about the data that I have that I'm going to use with one of these vendors to kind of help me, again, serve my customers better. The data is becoming more and more valuable, particularly the first-party data that you have. When I say first-party data, just a quick definition here, right? You kind of talk about like zero through third-party data. And then zero-party data is usually like the preference data that a customer will give you, that they'll say, look, I only want you to contact me by email. Don't text me. Don't send me snail mail, paper mail, just email. That's an example of zero party data. It's a preference from a customer. This is how I want you to interact with me. First party data is you as the vendor, me as the customer, we're going to have this interaction back and forth. I'm going to tell you my billing address. I'm going to give you my email address.

[39:07] You're going to know the last three orders that I made. All these things that have to do with our interaction back and forth.

[39:14] And that data only exists between you and I, you may have a sharing relationship with somebody else because that's disclosed and it's the whole thing, you know, that kind of plays into it. A lot of times you look at like.

[39:28] A common one for small businesses is your payment provider, right?

[39:32] So you know your business. I know the business I've done with you, but probably Stripe also knows how much money I'm spending with you, right? That's an example of kind of this third party that's involved. But that's first party data. What you know about me, what you know about everything that I've done, and what you know about all the other customers that you're serving, right? And then you get into like second party data. And that's like that Stripe relationship. It's the partner relationship. You're selling through someone else, right? You have this like view into what's going on, but it involves somebody else's data set and you're only seeing part of it. And then third party data is, you know, basically things that are out there that you might use or scrape. Maybe it's like government level data. So you think about like demographic data that comes from the census, like the Bureau of Labor Statistics. You might also purchase data if you're, say, buying lead lists or buying contact information for customers that you're going after. That's an example of third-party data. The thing about it is third-party data is also available to others, specifically your customers, right? I mean, sorry, your competitors.

[40:41] That's going to be hard to build a sustainable advantage, right? Something where you're relying on somebody else. You might lock it up, maybe that second-party relationship, a customer or somebody you're working with where you're getting exclusive use to something. Those are few and far between. Where the real value is, is that first-party data, the data that you have that only you have. Right.

[41:06] That's at least potentially where you can make an awful lot of unique and high value offerings and procedures and decisions from right when you're doing that. And so this gets into kind of where we were talking before. You want to treat it like money. How do you handle that? Who do you let have access to it? Do you necessarily want to start taking that data and just blindly giving it to everybody?

[41:35] As a marketer, everyone will be very familiar with what I will classify as a bait and switch Facebook pulled about 10 years ago. And this was very much the idea of like, hey, everybody, you've got this great community that you've built. You have a blog or you have an email list or something where you're communicating with this group of people. Why don't you just bring all that to Facebook? We'll set up a group for you. You can do the group and just talk to them and we'll take care of all the headache. And your customers are already here on Facebook. Just bring it all later, we'll do it. And then by and large, Many, many marketers and community leaders and people that were kind of shepherding groups of people together did that to kind of go, well, we're just going to move everything to Facebook. And then Facebook basically went, gotcha. Now I'm going to charge you to talk to them, right? Right now, your your any like organic reach you think you're going to get from from the Facebook platform goes to zero. But I have this great ads product that you can buy that you can talk to your audience as much as you want, as long as I get a cut of every single interaction. Right.

[42:52] But there is a generation of people that are very much in a fool-me-once scenario when they start thinking about their relationship with large technology companies, particularly companies that are now busting through the trillion-dollar valuation standpoint to no small effort on the data that we freely gave them in the past 15 or 20 years.

[43:13] So that is all to say that there's a cautionary tale in that. And I think the vendors are largely aware of that. Like they're very upfront and forthright about how we will use your data, how we will not use your data. I think anybody that uses, say, a marketing automation platform, there are like very, very clear disclosures about what they will and will not use to train their own algorithms on, how it gets used, how it gets stored. Most of the time, anybody of a sizable organization as a vendor is also playing in Europe and they're playing with GDPR regulations, which are way more stringent. I mean, they have the whole right to be forgotten element of what they're doing and what they have to produce. You see a lot of that in a lot of the emerging regulations that come out of the state of California who kind of use that as a model for how they're structuring that. So all of that says is there's a broad realization of first-party data is valuable, and it's getting more and more valuable, and you should treat it that way. Mm-hmm. Okay. So...

[44:32] That's very interesting in terms of like the distinction between like the different zero, first, second, third party data. Yeah. It seems like from a, like both competitive advantage angle going forward, like the zero and first party data are like the most important for marketers and also like have the potential to, you know, like protect customers from like getting their data, like just shared around to a whole bunch of different companies, which is a growing or not growing. I mean, it's an anxiety that's been growing in the general population since companies have started to do this sort of on the back end. So I'm wondering if the best way to manage that as a marketer is just to be

[45:28] very clear on what data you want to collect and how it aligns with your business objectives. And then minimize the amount of data that third parties or other companies that you might be working with can access from your customers.

[45:49] Would you agree with that as a strategy? Yeah. I will say that I work in this industry that, again, high regulatory burden, lots of compliance piece to it. But at the core of it, banking, and particularly small banking, like community banking, credit union banking, almost their entire value proposition is around trust.

[46:14] It's you believe I'm, you can trust me with your money, which short of the welfare of a immediate family member is like one of the most high trust relationships you're going to have in, you know, in your, in your whole life, in your, your relationship. And, you know, they take that very seriously, which means there's a lot of disclosure about what we do and what we We don't do some of it mandated, some of it not, but also just, just the like very clear, understanding of your, you're in a, you're in a promissory relationship with people and they trust you to do that. You're, you have that relationship with them. They trust you to do it. And if you break that trust, there are dire consequences. There was a recent example, there's a company called Carta. Word up. And they basically, they had a tool that was called cap table management. All that did was they said, look, if you're a private company and you give shares of the company or options in the company to your employees, that kind of becomes complicated to manage dealing with investors.

[47:31] There's like tax implications to that. There's like a product they have, right? So if you're like this tech startup and part of your compensation is you're giving out shares, they would manage that for you, right? What they wanted to do was get into the business of saying, if you're going to allow your investors or your employees to sell those shares before you go IPO, we want to be in that business of being a broker around that. And that kind of changed the relationship a little bit. And they were very clear, no, no, no, no. There's a big firewall wall between you give us this information for your cap table, and then the other part of the business is going to be the brokerage side.

[48:13] Well, they had somebody reach out to one of their customers and say, hey, would you like to sell your shares? I have a buyer all set up there. And it was not only in violation of directly their relationship with this one customer, but it also broke their promise of we are not going to monetize and try to make money off your data from what you're doing here, right?

[48:37] And it was this huge blow up. And there's a lot of Monday morning armchair quarterbacking about how they handled it from a publicity standpoint. But what they ended up doing was essentially severing that whole part of the business. They exited what was a fairly high value part of their business because they realized they had broken this trust and it was probably going to be the end of the business if they didn't do it. So there's this understanding of not only is there the letter of the law that you have to abide by, but there's just a general, almost visceral response to even a perceived breaking of that trust when it comes to it. So again, using that mind of data is almost like money and your customer's data that you hold on to, it's almost like you're holding on to their money for them. And if you don't treat it well, if you're, if you're, if you're, I mean, obviously if you're duplicitous with us, yeah, you're going to get, you're going to get dinged. But even if you're not diligent with it, even if you aren't security-focused in what you're doing and you have a data breach and lose that data, that's almost like you've lost their money. And I know there's been some fairly high-profile ones of those, but boy….

[49:59] Nothing will ruin a brand quicker than, you know, you just showing that you did not care about that relationship or that trust they put in you as much as they did. Done. Yeah. So on a practical level, what should like business leaders or marketers be doing to make sure that you're like securing that data while also like building your marketing program? Because you need like just enough to be able to develop this personal relationship with your customers but you've got to like guard it under lock and key as if you're you're securing their money so how do you manage that like if you had to kind of go like step by step you know like say i'm a marketer i want like to start building like using ai maybe i want to start using chat gpt or like one of these many other like AI companies that are coming out on a regular basis to improve my marketing campaigns and my outreach.

[51:00] How would I go about like integrating that into my workflow while also like safeguarding my customers, like privacy and data security? No, no, absolutely. So I'll just say, you know, for the record, I am not a lawyer and there's obviously you can kind of get into things that are legal. So don't construe any of this as legal advice. But thinking of it as a business owner, how do I even just have a framework for thinking of this? If I'm going to kind of go down this path, number one is to be very deliberate about it, right? Just to understand that this is an important thing. And even if there isn't a regulatory burden on you, even if you're not worried about it from a legal compliance standpoint point or a legal risk standpoint, just looking at it from the standpoint of, let me be very deliberate in how I approach this, right? That if I'm going to work with a vendor, no matter who it is, and there's going to be a passage of my data or my customer's data, particularly anything that's personally identifiable.

[52:08] I need to look at that very, very carefully. I need to look through the terms and conditions. I need to seek counsel if that's necessary, or I need to change my relationship with them. Now, there's a tension here because we talked about a lot of these things. It's not going to be AI takes your job. Somebody with AI takes your job. Meaning, you can't use this as an excuse to sit it out because that will put you at this huge disadvantage, and maybe an existential risk to your business, right? So it's how do you proceed, right? And this is an area that very much requires wisdom. And much like the AI isn't one big thing, you can't think of this as a binary yes or no decision, right? It's not, do I engage with this or don't I? You need to be thoughtful to say, well, we can take advantage of a lot of these services or apply it to this use case, but not give up the whole store on first-party data. Maybe I don't need to give them everything that's there, or I need to include that in the prompts that we're doing. Maybe I could just put placeholders in there for the information that's important, that's protected, get the value out of what I'm getting from an email generation or from a recommendation engine or whatever goes in there, and then later add back the information that's needed after I get it back from them, right?

[53:37] There is a process out there called retrieval augmented generation, RAG, R-A-G. And if you want to kind of go down the rabbit hole, just Google that. There's like a thousand versions of it to look at. But this is an example of how can I potentially use a generative AI system that will give Give me something that's both relevant and tailored to my situation without giving them everything that I know about the topic or the subject.

[54:06] Let me just boil it down to this very small element, and maybe I can even strip out some of the things that get into that proprietary data and still get the value out of it, and then put it back together. And this gets into that it's not one big thing, it's a bunch of little things. And maybe some of the little things I do in-house, and then I go to somebody like OpenAI, and I get the generative piece of it, and then I reassemble it together. And it's not one big in and out, it's maybe six or seven steps that go along the way where I can get the value, I can get the personalization, I can do the better service, and I'm able to keep control of all the data that I've promised my customers and my constituents that I would.

[54:46] I can solve both problems. It might be a little bit more work, but it's doable. There's elements of that that's doable. And you only get to something like that if you're thoughtful and if you break it down and if you look at it from a practical standpoint.

[55:03] There are whole career paths that get into data governance and data risk analysis and doing data governance audits and all that. And when that makes sense is largely about the scale and complexity of your business and what you have and all that and what the regulatory requirements are. But there's ways to get into this. And again, don't think of it as yes or no. It's not a binary decision. It's what's the important part? Where is the risk? And how can I mitigate that risk while still doing everything I can to take advantage of the tools that are available to me? Okay. And so do you have any like case studies you've heard of, of like people using this process of like, you know, managing the data, kind of thinking through what they need versus don't need. And then, you know, only giving out little pieces of that information to like an AI company. Yeah. Sure. I mean, I'll say we regularly do this with our clients. Again, I'm talking like larger banking financial institutions along that way. But like, here's one example I'll give.

[56:17] When, particularly for like lending product, like where there's like a credit worthiness that gets part of it, a lot of times they'll use a third party to get a credit report, say like Experian or TRW or somebody like that.

[56:33] There's a lot of thought that goes into when do we request that information? There's some actual legal requirements around that about when you can pull a credit report and there needs to be consent and how it's done. And then it's a state level thing. It kind of gets very complicated, but a lot of times it's one step in the process. Process and a best practice is when it gets into this idea of i'm now going to reach out to a third party and potentially expose even when everybody's on board and we've dotted the i's and cross all the t's i'm i'm exposing you know this information to a third party i'm going to wait as late as possible to do that i'm going to do everything else that i can in the process so that if If something falls out, if there's another disqualifying information that has nothing to do with credit report, I want to figure that out before I involve a third party. I'm going to be thoughtful in how I approach my customer and the experience I give them so that when there's something like this that is a potential risk, and it's obviously a cost involved as well, I'm waiting until I know we have to do it at this point.

[57:55] We cannot wait any later, or I'm doing a disservice to my customer if I wait longer than that, right? And it's almost like medical triaging, right? What am I going to do the test? What am I going to do the specific drug or treatment or operation or whatever it is? I'm going to do that with a sense of what's best for my customer, and how can I ensure sure the best possible outcome for them in the order in which I do things. And that's not really doing anything different, but again, the way I approach it and how I think about it. So that idea of when you're involving a third party and when you're involving first party data, again, being thoughtful and thinking of it very much as how can I mitigate the risk here?

[58:46] How we can't avoid Avoid it completely because then you're not in business. But how do I think about it to say, let me wait until I have to, or let me not use as much information as I have. Maybe trial and try it out. Can I get away with less information here? But what's the least I can do? What's the least exposure I can make to both my customers and myself?

[59:10] Okay. Yeah, that makes a lot of sense. Is there any like challenges you've faced in trying to set up a system like that where like in terms of like, how do you make sure that the people on your team know not to spread the data around too much? How do you make it?

[59:34] Like a clear policy on how you're going to use the data and have you experienced any challenges in trying to set those things up? Yeah. Well, I mean, again, you know, I said that's the whole reason why I got involved in this company. The first one was like, this is the, it's the standards and practices and experience of the entire staff for several decades, right. That adds a lot of of value and allows our customers to trust us. Now, obviously, when we're working with a client and we're involved with them, we tend to get our hands dirty on the technical side. So there's a lot of things that we do or don't do based on how much risk and exposure we want to have to them and to their customer, right? And what we can do kind of being a hands-off of not being there. From an inside point of view, like how we do this internally and how we work with clients that are dealing with this as well, like this very quickly, depending on how much involvement there is in data in your business, it becomes a core competence. You treat it almost like you're treating the way that you train people to do the job that they're doing, the way that you approach them from a professional development standpoint. Who you hire, does this become like a selection criteria in there?

[1:01:03] For us working for banks in a regulatory environment, we have to do things like background checks and in some cases, drug testing, things like that that are about screening from that standpoint. Again, it's the sliding scale about what type of compliance environment are you in that you have to deal with. That might be appropriate. I think certainly from the standpoint of the core value of what you do and beginning to tie things like our practices for how we deal with data is directly tied to the trust our customers have with us. And it's very painful for us to break that trust. Carter example as a prime example of that, eyeing that together as a value, as like, this is how we do business here, that is usually way more effective and way more.

[1:02:04] Personable and tiable, and you can assure it way more than any policy or rule book is. is, right? So there's a lot of things you can do when we start talking about data on the technical side. There's several companies that all they do is they provide a layer between a company and something like ChatGPT, and they just screen for PII. They look for potentially an employee that's putting in phone numbers and full names and addresses and things like that, right? And they they just like will either refuse connection or they'll scrub it out or something like that. Like that, that's a, that's a little bit more of a, of a, like a technical hygiene approach, but that's only going to get you so far. And it only helps you solve the things that you've already thought of. It's not the thing you haven't thought of yet. That's going to be a problem. And that's where I say, you know, really trying to tie a lot of this to your values, to the way you operate, to the, you know.

[1:03:06] People who work here do things like this, like those type of norms for how you operate. Making that part of it is like, you probably have a norm for how you handle money, how you handle expenses, how compensation works, how you deal with pricing, how you deal in negotiation with your vendors, like all of those things. There are norms to all of those. And I think the same thing should be true for how you deal with your data, whether it's that zero or first-party data or how you deal with your vendors and your partners from a second and third-party standpoint. And I think if you get to the point where you go, this is a much bigger deal and we're kind of out of our depths, that's where you go find real advice to do this. And I think looking up data governance, data governance consulting, data governance compliance, there's a lot of talent, there's a lot of offerings out there. And frankly, Frankly, there's a lot of education as well that you can just go out and get and just like look for best practices, right? What should a company like me be doing in this specific scenario?

[1:04:12] Do you have any like recommendations? Like if I wanted to go look at data governance consultant or find a training program for myself or like my team members and like, are there any like specific things I should be be looking for or any like pitfalls I might come across as I'm trying to like find somebody who's good?

[1:04:32] Sure. Yeah. There are several certifications for that. And what we'll do is we'll put them in the show notes. I'll give you a couple of links we can put in there to look at. That's like a good place to start to think about. Like there are like some are industry specific, but there are some general frameworks around this is what you should do from a data governance standpoint. And there is a certification for data professionals. And then it gets fragmented it from there about what type of industry so okay sounds perfect and yeah so you mentioned a little bit earlier like risks associated with companies like facebook charging businesses to talk to their own customers and then how like working with giving your data to these like ai companies could lead to the same kind of dependency if people aren't careful. Is there any way that marketers and business leaders can avoid that? Or is that part of the same strategy, just being very careful with your whole supply chain of data? Or do you have any other specific tips?

[1:05:43] No, I think it's very much along the lines of being deliberate and careful, meaning reading the fine print about what you're buying and what they're doing and how it's potentially used and not used. This is one of those things where it's evolving very quickly.

[1:06:01] And it's kind of a moving target from that standpoint. It is telling, though, that most of the big vendors will talk about this front and center. I think some of the more well-done and transparent versions, I'll say anecdotally, have come from Microsoft.

[1:06:22] Most of this is through their Azure platform. A lot of their offerings they do. And they're the ones that have the big partnership with OpenAI. AI, the CEO Satya Nadella has talked about this, like what they do and don't do and what they promise at the various tiers of their offerings.

[1:06:41] They're very good because they're so giant and they do so much industry and government level work. They're very good on their compliance work. So when you start talking about something like a cloud service provider and what they provide, They will often have reams and reams of documentation and third-party audits about them meeting certain compliance frameworks like HIPAA and healthcare, SOC 1, SOC 2 from a financial standpoint, for things that deal with the federal government, there's FedRAMP system. These are these compliance frameworks that say, basically, you're doing the thing that you said you're doing, and you have the security measures in place that you say and you advertise that you have, and you're treating the data the way that you are. And they've been very much forefront of this, talking about that, talking about what they do and what they don't do. So, you know, looking through there and I'll find another link for you to, to put in there to kind of look at an example of like, this is somebody who seems to like be cognizant of this as an issue and wants to do everything they can to allay the fears of their customers and their customers, customers.

[1:07:57] Oh, awesome. Yeah, that'd be really helpful. Thank you for sharing those links. And yeah, as we're wrapping up the conversation, I'd love to hear about any upcoming projects you're working on or how people can get in touch with you or your company. Sure. If you want to talk to me or you got questions, you can find me on LinkedIn. That's where I'm most active. I'm Rich Edwards on LinkedIn with Mindspan Systems. I will throw out a quick ad here. We are launching our own podcast. It's called Behind the Vault Conversations with Community-Driven Bankers. It's very much geared towards that small.

[1:08:34] Community-driven banker or credit union that I think a lot of people don't realize how integral they are to our economy, particularly our local economy. They're that class of business that kind of works locally, small businesses, like the community banking and credit union, they punch way above their weight class in their origination and support of the small business administration loans, which is this huge part of everything that you see. Also local community real estate, just about probably every strip mall or small local office building or medical building, dental, that was probably financed through a local small financial institution made that thing happen. So a lot of the businesses and services that most people take advantage of rely on these smaller institutions to do the financing and the access to the money to get that done. And so we're doing this podcast to talk to some of those leaders about how they're staying competitive and how they're staying relevant and in particular, particular, why they're continuing to win, even in a time when a lot of the faith in the banking system has been eroded as of late. So come check it out. Our podcast, again, Behind the Vault, Conversations with Community-Driven Bankers. All right. Sounds like a very fascinating topic for a podcast and very relevant to banks nowadays.

[1:10:02] So thank you so much, Rich. This was an amazing conversation. Thanks, Kira.