Turning Operational Data into AI Ready Assets - with Andrew Deutsch of RR Donnelly and the Fangled Group

Welcome, everyone, to the AI in Business podcast. I'm Matthew DeMello, Editorial Director here at Emerge AI Research. Today's guest is Andrew Deutsch, CEO of the Fangled Group and Director of Operations at RR Donnelly. Fangled Group is a global marketing and sales consultancy that empowers companies with data-driven, go-to-market strategies and deep customer insight to accelerate business growth. Andrew joins us on today's show to discuss innovative strategies for operational efficiency and scaling workflows in complex service industries. Our conversation also highlights practical approaches to integrating technology with workforce solutions to drive measurable business outcomes and improved ROI. Today's episode is part of a special series sponsored by OneTrust. Just a quick note for our listeners that the views expressed by Andrew on today's show do not reflect that of R.R. Donnelly or its leadership. But first, are you driving AI transformation at your organization? Or maybe you're guiding critical decisions on AI investments, strategy, or deployment? If so, the AI in Business podcast wants to hear from you. Each year, Emerge AI Research features hundreds of executive thought leaders, everyone from the CIO of Goldman Sachs to the head of AI at Raytheon and AI pioneers like Yoshua Bengio. With nearly a million annual listeners, AI in Business is the go-to destination for enterprise leaders navigating real-world AI adoption. You don't need to be an engineer or a technical expert to be on the show. If you're involved in AI implementation, decision-making, or strategy within your company, this is your opportunity to share your insights with a global audience of your peers. If you believe you can help other leaders move the needle on AI ROI, visit Emerge.com and fill out our Thought Leader submission form. That's Emerge.com and click on Be an Expert. You can also click the link in the description of today's show on your preferred podcast platform. That's Emerge.com slash expert one. Again, that's Emerge.com slash expert one. Without further ado, here's our conversation with Andrew. Andrew, welcome to the program. It's a pleasure having you. Hey, thanks so much for having me. Absolutely. We're talking today, especially about what we're kind of seeing across many large enterprises, folks who have been on the show. You know, they're talking about or at least where they're really establishing the problems. It's that they still rely on fragmented data systems, legacy tech stacks and tools that capture information but can't drive a meaningful action. And especially the farther we go past the AI hype cycle and actually to AI being really realistically embedded in these organizations, it really doesn't feel like they can drive that meaningful action. The results are, or at least the symptoms are, misalignment between departments, deeper silos, inconsistent KPIs, bad decisions on incomplete data. And at scale, these issues can lead to serious blind spots in how leaders can understand efficiency and value. But just from your experience, especially getting underneath these problems, what are you seeing as the impact of fragmented data sets and those legacy observability tools on enterprise operations at scale? Huge, huge question. And the list is long. So I think we probably need it. Exactly why I wanted to head you off at the pass with the symptoms. Yeah. So, you know, there's so many places where faulty data, fragmented sets, the pre-interview part that we had talked about, probably the most important aspect of it to really look at is organizations that have multiple things that they do. And with fragmented sets, they're all viewed under the same KPIs and the same systems where what makes those divisions unique needs to be somehow looked at within that light so that the organization can see the difference. For example, imagine a manufacturing company that has, let's just pick five locations. Four of them are commodity producers and one of them is a specialty house. So the specialty house is producing short-run special items. Everyone else is doing long-run items. If they have the same KPIs for units per day for how their specialties are done, the data doesn't make sense. You've got this one organization that could be your highest performing, but based on looking at that data, it's the worst performing. There's a company that I've dealt with in the past. They had a division that worked in pharmaceutical. It was a packaging company. They had pharmaceutical packaging, and then they had other packaging. Well, the pharmaceutical side, they had to have special FDA approval through the process CGMP, which is packaging protocols for the pharmaceutical industry. So if you were to look at their headcount versus dollars out the door, it made no sense. They had to have extra inspectors. So the data sets when you're dealing with multiple different things within the organization, but expecting them all to perform to the same KPIs is crazy. and to manually try to adapt those has finally seen solutions with AI systems that they can look at that data in uniquely to create a clear pet image of what each division is doing. Yeah, I'll say this, and this doesn't even reflect Emerge Internal Operations. This goes back to my time in AI vendors. But you've seen this kind of smash where I've been in with editorial teams who then all of a sudden we get some kind of supervisor who's got an accounting background and then we all get accounting metrics. How many articles did you put out today? Well, yeah. Who cares how many we did? Are they any good? Did they meet deadline? Yeah. Our metric is the old newspaper system. And part of that experience was even sitting down with my boss and saying, hey, we're sticking to newspaper systems, right? Deadlines and everything else. Dan was extremely supportive of that. But I've seen that. And I think a lot of folks at home, no matter where they are in whatever industry, you know, a lot of this process started out of everybody getting the same report card. And there's a really great Einstein quote about what happens when everybody has kind of the same report card to answer to and how that really defeats education systems. So really, just pulling apart your advice here, it sounds like, you know, you want business goals to be kind of the golden calf here or at least your ultimate KPI. but that place that the different lines of the business, the different silos within the organization, how they're graded to achieve that goal needs to be taken on a department basis. I know we're going to get into, or at least I talked a little bit about in my lead in, that that leads to, or at least what we have now, fragmented data, leads to deeper silos. It still sounds like right there, at least off the bat with individualized KPI, yeah we might need to make the silos deeper before they get more shallow So typically companies are doing sales forecasting based on the previous year Right And from the sales forecast they building their operational their production all of those things based And they all flow in this year in advance. What's not involved are the simplicity of correction factors. So if you've got this data from last year and the market has made changes, the forecasts are not adaptable. So we're planning headcount based on the sales forecast for October, and we're planning that in January. Well, with proper systems that can look at the yeses and nos that happened in sales projects, for example, we base the forecast on XYZ doing a million dollars business with us this year. XYZ in February shows a competitor. Now, how do we update and change that forecast so that we can properly look to what we need to do? In the old style system, it would be, hey, we've got to shut down for a day and go back and recalculate all this stuff. Now with modern systems, we can see those signs and adapt our forecasts and provide additional pieces. In the old days with pipelines and sales teams, you've got so many that you've reached out to, so many that have raised their hand and said, I might be interested in business with you. And each of those have probabilities built in so that you can look at what your pipeline would be. So if you had in the history, 5% of the people who you've reached out to become customers, and during the year that number has adapted, and it's no longer 5%, it could be adapting in real time so that your pipeline forecast of where you really believe you are is a real-time forecast as opposed to, well, we've always looked at 5%. I've dealt with folks that have like Salesforce.com hooked in. They have not adapted those percentages, nor have they gone back in 10 years to see if they're even accurate from when they first put them in the system. So what we're seeing now is that if we can create systems through AI that get the yeses, the nos, and the maybes and recognize how those alter percentages, we get much more accurate in what our output is going to look like. So just even pulling apart the beginning of what you were saying, the goal set in January is everybody knows there's three eternities that take a place between January and October. They're not just quarters, they're eternities for everything that could happen. But the sign of a robust system that's really driving these technologies in the right way is that the understanding of the business goals are so agile that even for what you might set out in January, they have appropriate context and are much more aligned to the facts and circumstances of October by October without you needing to stop and plan. that's that's kind of the sign of a forecast a forecast is based on the assumptions that we're making based on information that's already old and i've i've gotten myself in trouble over the years because what are the assumptions are they just how we lie to ourselves so that then we spend the year trying to prove it true yes yes so so we're working on those assumptions when those assumptions are proving true and false are we adapting moving forward or are we stuck with those assumptions for the entire fiscal year when i've even been in conversation with well hey guys We recognize this. So when we do our budget for next year, we're going to take that into account. And my response is, dude, it's June. It's June. We still want to succeed this year, don't we? It's the same. I mean, if we want to get into the weeds with this, all business in the world has adapted to a non-AI environment, which we started with. So every month, you're going to report out your financials to the team. and then your quarter, your half year and your year, you know, as you go through and you're going to be doing this on a regular basis. So it's now five days before the end of the month and you're the operations team and you go, oh my God, I've got a million dollar bogey to hit. I got to get a million out. I've only got 800,000 out. What can I steal from next month? So I make my number. I got to make my numbers. I got to report out this number and you do it. And then you start the next month. You go, oh my God, I just stole $200,000 from next month. Where am I going to get that. Well, what does that now go back in time? Now you're the buyer for ABC Industries and you're being judged by what's left on the floor at the end of the month when the inventory is carried. So you've adapted now to where you're placing all of your orders in the last couple weeks of the month and telling your vendor, they've got to be there at the beginning of the month with just the right amount to get out by the end of the month so that I've managed my inventory. So we've become adapted to this one month cycle, but then the one quarter cycle and so forth. Imagine if the financials were a floating four week, where your performance during that whole four week period that's constantly floating has to be consistent. We would stop binge buying. We would have to become far more fiscally responsible, not playing the numbers game every month. But what we've encouraged our ops teams, our sales to everyone to sandbag, I know that I could do a million dollars this month, but you know what? I'm going to tell them 900. I'm going to under promise and overperform. There's all of these games that we play that now with technology have done properly every morning. We can see where we are within our four week span and plan and work properly through it. No one has adapted a system that does that. But imagine not having that end of the month craziness. I'm sorry, Matt, but you can't You can't take a vacation the last week of the month because that's when we need you the most. Yeah, we were talking about editorial and deadlines before. That end of the month craziness, especially for sales teams, might feel like that's the job I chose. Not so much something that could be changed. But the beautiful thing about AI, it fundamentally might move the baseball really on the things you thought were permanent problems of your occupation. In that way, just for, you know, I know we were starting talking about, you know, system sprawl, you know, too many tools kind of in the toolbox. Predictability, I think, can be one of those things where if that's the only sign of, you know, systems working, then you could still have tool sprawl. You could still have a lot of bad things, you know, going on. But you see the predictability and say, oh, well, something must be going right. What else is the sign of a mature, well-governed AI ecosystem and how that looks in a global enterprise environment, just even beyond the predictability? In that predictability piece, sort of the next step is as things are going right, you're being reassured by all of the data, not just by the event, that you're on that path and you can make immediate corrections. So it gives you not so much the predictability, but in real time, where am I and what do I need to do to continue down that path? You're either getting shocked with a cattle product or patted on the head at each moment as you're making those immediate adjustments. Keep going back to the rodeo analogy from our prior conversation. You know, if you've been given a set of tasks and you follow those again and again and again, getting a negative response, maybe the system should be telling you, let's try something else. Where, you know, a lot of that, what's the ridiculous, the supposed definition of insanity that's not actually true? Right, right. keep doing the same thing again and again expecting a different result you're crazy well this can help us from from repeating those tasks again and again and again that aren't getting us where we're going because we seeing in real time those results yeah there a simple model of that imagine the boss tells you you selling a widget and the boss will tell you every one of these widgets that we sell you get a fidget though You can add that on. That's the thing to suggest to get your, get the, you know, you're going to sell a guitar. So you got to suggest an extra set of strings. Let's get the most out of the guy's pocket. We can 20 people come through the guitar store. You sold them a guitar. None of them are buying strings. And well, why am I still suggesting them? Is it that I'm not suggesting them right? It's no, they don't need strings. They're brand new. What would happen if you sold them a pick? What about a stand? What about these other things? And you're hitting my territory because, as the audience well knows, I'm a musician outside of this. But then you have to ask yourself, are you hurting your bottom line trying to shove guitar strings in everybody's hand? Is the real bottom line understanding what they need, extra cables, whatever, you know, they're amp-repaired, those things. And you can upsell that you have probably way better margins of what you can upsell them if you understand what the customer wants out of the situation. And it's that data that's going to defeat all those assumptions that we had before. You had that definition of idiocy that's not even true. I'd actually say the same thing about assumptions. There's that old euphemism about assumptions. And that to me, you can't wake up in the morning. The fact that you'll wake up tomorrow morning is an assumption. You need assumptions to live. It's about how you get grounding for those assumptions in data and those that survive. And those that don't lead to that other phrase that might not be so friendly, family friendly for a podcast. But, you know, you've been mentioning these other attributes of these well-governed systems. Obviously, especially the more we're talking about regulated industries, financial services in particular, the more I think those a lot of folks in that space are just facing huge data debt, huge system sprawl. You know, they could be they could have bad cholesterol. And we're here. We are telling them to exercise like they might not have ever heard of it. they probably know they need to exercise. But even if they're kind of up the creek, so to speak, keeping that euphemism family friendly, what's your first recommendation to folks who are kind of drowning in it right now? What are those first steps they should take to future-proof their AI and analytics infrastructure? Well, the first is you have to know your current state. Yeah. What is currently going on? And more importantly, the future state that you envision. Where do you want to be? What do you want it to do for you? So much of the media talk of AI is replacing humans with these robots that are going to answer questions and the language models, which is fine. But my belief is that if you can look at this as how do I make my people better at what they do? It's the old, you know, the superhero. There's superheroes that were born with special features. And then there's the ones like Batman, who's got a cage full of them. And there's the guys with the exoskeletons. I'm not a comic book guy, but how do I take these people and make them better? What is it that you want it to do? Because what's happening right now, and I have these conversations all the time with people about AI, is how can I incorporate AI in my business? What do you want it to do? You want an AI system that's going to brew the perfect coffee for your team? Or do you want an AI system that's going to make you a better salesperson? salesperson? You want an AI system that's going to smooth out your supply chain? What is it that you want? Do you want a system? And I worked on a project for a long time, how to make salespeople from the day they're hired be faster at growing in the business than they are now? What can we do from a predictability model so that salespeople are out of the box faster and quicker? Companies that take a year before a salesperson can really sell their products. How do you speed it up? And we worked with AI to build a predictability model to get them in place. And it came from first doing a marketing persona study about the different customers they deal with. It's quite complex. But what, what is it that you want? You know, it's, it's, it's kind of like, you know, you're going to go buy gasoline or you're going to figure out what car you want. Right. And this is a, so much of a harder question than it is to say out loud, but the, then the real heart of the question is what do you do? And then what, What are your goals? And then backtracking to, all right, now that we know what those goals really are, you needed to probably even take some data, really look in the mirror at what that is. Insert the spiel I've made many times before about the great movie Founder. McDonald's does not become McDonald's because they sell hamburgers. They become McDonald's because they sell real estate. Once you're really at peace with what you really do and what the business goals are, okay, then let's bring in data scientists to say how applicable are those business goals to the current market in array of capabilities that you have. I think there ends up being like kind of a buy versus build or maybe buy to build ratio kind of question in there. I think there's, especially on the regulated financial services side, there's an impulse of like, maybe we can just buy our way out of it if we're just drowning in it too much. Do you have any insight maybe on where to make that call or maybe what are the best signs to really make that call versus just assuming, oh, we're drowning in it. Let's go find somebody to pay to fix it. Yeah, I think that the biggest challenge comes is truly defining, first of all, what are your pain points? Where do you need help? And so often we're out there getting help for the very thing we don't need help for and ignoring the main problems that we want to ignore. When we do, my business is strategic marketing consultation. First question I always ask a company, number one, what do you guys do? Rarely, rarely do they know. It's usually we make this product. Really, you don't make people's lives better. You don't. What is it that you actually do? Then what's your value proposition? Well, we make the highest quality, whatever. And then we do the research on the customer side. There's never an alignment. Never has there been with a company that we've talked to that the customer's value of what they thought that company was and what the company thought of it. So that misalignment is so much easier to adjust when you actually know what the consumer is looking for. I'll give you a perfect example. Steel drum manufacturing company, hard old fashioned company. They make the steel drums that carry chemicals across the country. One of those drums fails. We have a hazmat situation. So they went to market bragging. Bragging is they went to market. We use all American source steel. We can go back to the mill. We can paint the drum at any color. We meet all the ANSI standards. And then the response was, okay, so you do what every steel drum company has to do to just be in business. Why do people do business with? Well, because we do all of that. Okay, that's interesting. Then we did the market study. We talked to their top customers, people who were loyal for 20 years, customers who second source, they buy from a bigger guy and buy from you as a backup. And we talked to people who quit doing business with. And what did we find? We love this company because they answered the phone. If we have a problem with our order, if we have a quality issue, it's resolved immediately. So the real value is personal touch and customer service. We're in the business of making buyers into heroes that have to buy this product while we're protecting and making people feel safe. That's what we do. Right. So if you know that now, how do you then communicate that to the folks that aren't doing business with? And how do you enforce it with the ones who do that? They'd never want to leave. So knowing that that gives you these these cues that should be built into how you manage your data to constantly work towards those goals, not the goals of of making what you're already making. And there a real look in the mirror moment just in that last example that you gave specific to that last answer of for the company in that space it really more about the problems in the market Nobody else is doing this So you know all your customer experience and then you have to be brutally honest about your place in the market versus staying in your music metaphor, or this is applicable to Apple of it doesn't really matter what's going on in, in, you know, digital product markets. Apple always, because of branding, because of who they are always has kind of that leg up people want from them a certain something that kind of almost no matter what the the conditions of the market are and that that's a different situation i'm not familiar with that what does apple do exactly exactly or is it jillion is a jillion if staying in music of you know oh it doesn't really it that's not how they answer the phone it's that 1000 year mythology that you know they're one of the oldest companies of all time making these symbols and you'll only get that quality of symbol from them. Or staying in Apple, you don't want to, you make media, you know about technology, you have to make advanced media products, but you don't want to spend time souping up your laptop. You don't want to spend time thinking about your RAM. You just want a thing that works. You can see I'm speaking on it biographically here. Anyway, but okay, so we have the kind of the end of the race and I know we're a little bit up on time. So just really quick last question. If you got the time, really appreciate a couple of extra minutes. So we've got the end. I know I was speaking in the beginning of maybe the silos need to get a little bit deeper before they get shallow. And we know that silos coming down, especially if it's in line with your business goals, everything else we've talked about on the show. That is really almost the universal gift of AI that AI could bring to just about any organization. Just wondering, how can organizations really unlock that business value of breaking down silos across teams with this technology? And how do they know they're doing it well for those business goals? I'm going to piss off your audience. Please. Not involving the IT guys first. You've got to involve the decision makers in the organization to define what it is that you want. I've never had anything better done by IT guys than when we came to them with a true model of what we wanted. I'll give you the perfect example. Salesforce.com, they're a huge company. I'm not a fan. There's better CRMs out there. Throw all the bombs, Andrew. Throw all the bombs. They're an okay company. If you like to pay for every additional idea that ever came and watch it get more expensive. But when I've worked with customers, clients of ours that have implemented CRMs and they've implemented it and said, well, I've got an IT guy, he'll take care of it. There's not an IT guy out there in most companies who has ever understood a sales process. They're not in sales. So they will set it up to work perfectly if you want to hire IT guys to sell for you. But if you get all the people, all the stakeholders together and understand what they actually do and how it works and how it's going to assist them and then go to the IT team and say, look, this is how we need it to work. IT guys light up and they go, hey, we can do this. And you get this incredible implementation. If you just say to the guy who's maintaining the computers in the building, making sure that all the networks are working, making sure that the software talks, they do all the integration stuff. Now, make this program work for salespeople. You get a system that's not intuitive to sales. Same with the marketing components. How many IT technical people do you know who understand the intricacies of marketing statistics? They don't. But if you build a model of what you want in terms of your marketing campaigns, digital campaigns and all of that, and share it with the IT team and involve them as the technical piece, you can go anywhere. Yeah, absolutely. And I mean, classic problem of hammer versus the nail. But to your point, you want to bring the hammer experts right in at the end, because in making that they know tools, they know tools better than anyone else. They need that last insight on the final solution to say, hey, this final solution is the best possible it could be with our knowledge of tool solutions. Whereas earlier on you go, you want to stay focused on the nail and you want to have nail experts in the room. And the bonus caveat to that in terms of the advice that we give is be prepared to adjust after you've done it, because what you've done is not complete. It has to adapt. And that's where the AI really comes into it is, are we getting accurate pipeline data? Are we getting what we need? And are the yeses and no's being fed into this logical system creating a better path? yeah if you've got you keep doing the same thing that gets you no no no no no no and you keep doing it and the system isn't advising you to try something different then then what's the point right right right and i know i know a couple of it folks you know starting off that answer were probably a little miffed but the ones who know they stuck through it i can just tell i i know this audience that's all right i could use another death threat i haven't had one in a week yeah yeah editor my editor might need to cut that but in the comments in the comments below i hope there's some really good negative stuff about me that'll be fun yeah see i'm a i'm a big fan too i like you know bad press means at least hey i have an audience someone's listening right trolls trolls tell me somebody was listening exactly exactly that's that's the positivity we're taking into 2025 2026 andrew an absolute pleasure i think especially telling our audience what they what they haven't heard, maybe what they don't want to hear, I think is exactly the kinds of insights that we want to bring to them and make sure that that's on their plate, especially for anything as hard as really transforming legacy tech stacks and really getting their heads above water when it comes to the tools we're drowning in. Thanks so much for being with us this week. Thank you so much. I had a great time. Wrapping up today's episode, I think there were three critical takeaways for enterprise leaders focused on data and AI transformation to take from our conversation with Andrew today. First, building a solid data foundation with quality, centralized, and well-governed data assets is essential to AI success and scalability. Second, establishing clear AI governance and compliance frameworks reduces risks and ensures ethical, secure AI operations aligned with business objectives. Finally, fostering organizational readiness through ongoing change management, cross-functional collaboration, and transparent communication is critical to driving adoption and maximizing ROI from AI initiatives. Interested in putting your AI product in front of household names in the Fortune 500? Connect directly with enterprise leaders at market-leading companies. Emerge can position your brand where enterprise decision makers turn for insight, research, and guidance. Visit Emerge.com slash sponsor for more information. Again, that's Emerge.com slash S-P-O-N-S-O-R. Bye.