Transcript:

Brian: Welcome back to Experiencing Data. This is Brian T. O’Neill. Today I have David Krauza on the line, VP of Enterprise Data Strategy and Products & Governance… did I get all that… at Comcast?

David: It is a mouthful, but yes, you got it correct.

Brian: [laugh]. So, we met in the hallway, I think with Juan Sequeda, who was on the show recently at CDOIQ, the Chief Data Officer, IQ Symposium. Was that your first time, by the way, or have you been there before that?

David: It was my first time attending that event. So.

Brian: Okay, got it. So, you were in the hometown, and we had a nice hallway conversation, and we started talking about, like, buy versus build, and I was I think telling you a little bit about how, you know, my business used to serve both internal data teams and also doing consulting for on-platform products, analytics products in particular, and started having a little, like, buy versus build with all this AI stuff happening. So, I asked if you would be willing to come in and just talk about your experience a little bit less about, like, the product operating model inside Comcast, which is probably where I would have [laugh] dug into a year ago. And now, really, as a customer, as someone that’s, you know, receiving pitches all the time and has been through evals and all this kind of stuff to help our founder-operator audience here to hear what it’s like, some of the stuff that maybe you can’t directly tell them, but you can tell through an intermediary. You can talk to me about it, and hopefully they can grok some useful insights here. So, I appreciate you coming on and being transparent about that.

David: My pleasure. Thanks for having me, Brian. I really do appreciate it.

Brian: Yeah, yeah. So, I think one of the burning questions for most operators right now is, you know, the sense, you know, AI-assisted coding is so big right now, it’s like the largest use case for this technology that’s out there, which poses this thing: well, if you have a lot of resources, you’re a large business, a lot of engineers and development staff, why buy anything when you can build it yourself? Has anything really changed for Comcast about, like, buying versus building in the last year or so? Like, what’s just the, kind of, general temperature about this? Are you getting pushback when you want to make a purchase, and they’re like, “Why can’t we just build this in-house?” And, you know, “Why do we need to go buy this? This looks like something we could do in two weeks.” [laugh].

David: Mm-hm. Yeah, I think with the rise of vibe coding, Claude Code, Codex, things like that, I think there’s a lot of folks… more than just technology folks, to be honest with you… who think that they can just go out and build whatever. I was talking with Taylor Culver this morning… I don’t know if you know him… and you know, we were talking about CRM systems, and we’re like, sure, you can vibe code a CRM system into existence, but is it going to be any good? So, you know, I think there’s a lot of discussions about what people can do from building themselves, whether it’s technical folks or non-technical folks, using a vibe coding tool, but I think really what it comes down to, it’s like when you look at what you could potentially buy versus what you could potentially build, I think what you have to think about is like what’s the long-term roadmap of that, right? If you think of, like, a CRM system like a Salesforce or a HubSpot or somebody like that, they’ve got a multi-year roadmap, a multi-million dollar, multi-billion dollar budget behind those platforms to evolve and develop them and things like that. And I think part of, you know, the trade-off with, like, building is like, do you also have that? Are you willing to invest in that multi-year roadmap? And so, I think that’s what organizations need to ask themselves, is like, do you want to be in that business or not?

Brian: I think that’s wise to be thinking out beyond just the initial, like, oh, it’s done now. We can start [laugh] start using it. Has there been any experiences in, like, the last year or so where Comcast decided to go with an internal solution because it’s good enough, or do you tend to see down the road where it’s like the maintenance on this, or the integrations, or things are moving so fast they’re going to break, and we’re going to have to maintain this, has the general posture been that most of the time, if you don’t want to be in the software business, it’s more of leans buy… it still leans buy regardless of all of this traction that we have with Copilot and stuff?

David: Yeah, I think it really depends upon, like, the use case and really what we’re trying to accomplish of whether or not you know somebody wants to build or buy, right? You know, I’m coming from a unique perspective, too, which is, you know, I’m not a software guy by trade, right? I’m innately a data guy, so essentially, like when I am thinking about a buy versus build decision, the first question I go back to is, like, okay, what’s the data that’s going to get thrown off from these systems, and how is this data going to get thrown off from these systems, and is it viewed as just exhaust, or is it viewed as something that we can use to incorporate into our broader data ecosystem, right? So, I think it really it’s a case-by-case basis, right, but I think, you know, I think increasingly what I’m seeing is not so much build versus buy; it’s integration of data into the ecosystem because that data is going to feed not only reporting and analytics as it always has, but also AI in the future, right? Like, you know, your system, your software, your whatever is going to be basically giving you new material with which to run your AI platform. So, that’s a third lens I would, kind of, add to that.

Brian: Yeah, you told me that you thought the biggest miss that a lot of the vendors coming in is that the data is seen as maybe a byproduct or secondary benefit, but it’s not, like, the core benefit. So, you could see something like a HubSpot, like a CRM, if you’re a sales team, it’s kind of like, measure our deal flow; where are we? Are these deals going to close or not? Marketing has a different angle. One could, a data person, would be thinking about this is a repository of all this knowledge that later, en masse, could become insight that we can use to strategically move in better directions.

I could see though that someone in sales may be like, well, I’ve tried out the UIs of HubSpot versus, you know, Salesforce and this. We don’t want… we hate Salesforce. We want to be over here, but data is thinking, like, the ecosystem around Salesforce is massive. Every connector you could ever want. We don’t need to buy data engineering services. Like, tell me about that. Like, where is the data capture side or access to the vendor data considered in the Comcast ecosystem when you’re selling into an org like that? Is there always a seat at the table for analytics and data to make sure that it checks the box on, like…

David: A hundred percent.

Brian: Oh, okay.

David: Yeah, a hundred percent. I mean, and it wasn’t always the case, but I think anybody who’s been around this space for a while knows that, you know, with the explosion of AI over the, what, past two three years, right? That seat has grown increasingly more important, right? You know, there were points in my career over the past… ashamed to say this… 25 years of [laugh] doing this, that data wasn’t always considered. It was, did this meet the need for the specific use case? You know, to your point, like, deal flow or reporting or whatever, right?

But increasingly now, it’s like, okay, what is the data that is going to get captured off of this system? Are the vendors treating the data that comes from their system as a first-class citizen, right? Do they make it easy to get after it? Do they have the right integrators to bring it into your data ecosystem, if you’re bringing into, like, a Snowflake or a DataBricks or what have you, you know, things like that. So, you know, those become, like, the first questions from a technology perspective that I start asking.

It’s like, at the end of the day, what I’m going to be held accountable for is not so much the feature itself that gets executed, but how is that data going to be brought into an environment where my analysts can then look at that data to measure the efficacy of the tool, but then also, you know, integrate that with other aspects of the business, so that, you know, we can tell a more complete story of the customer or the product or what have you, right, so that we, you know, we can do better AI, we can do better BI, we can do better analytics.

Brian: Got it. Is the data as a first-class citizen, you had mentioned this as, like, the single biggest miss, so what I’m wondering is, is the miss in the fact that it’s not in the marketing and it’s not in the sales conversation, or the product itself is short on supporting those use cases? Are they just not talking about it and it’s there, or it’s often just missing? They haven’t thought about it at all as a value-add that might help close a deal?

David: So, I think data is just such a… I mean it’s been around forever, right, but it’s still a burgeoning space in here, right? And I think a lot of time, and I’m guilty of this too, not thinking of it first, right, when we want to bring a product in there. So, you know, I have worked with some vendors in the past who, like, “Oh, by the way, you know, we have this extensive suite of connectors that you can, you know, bring data in.” And then there’s others that’s, like, “Oh, crap. We’ll have to build an FTP process for you to get at that.” So, I think it depends upon vendor maturity, I think it depends upon the maturity of the organization that you’re talking to.

You know, I think at least what I’ve tried to do is try to raise the visibility of data at Comcast, so that, you know, those become first-level decisions that we’re talking about when we have a vendor come in there. So, I mean it’s kind of all over the place. I do think though it would be interesting to see what would marketing for data integration look like? It doesn’t seem like something that would… be, you know, snazzy [laugh].

Brian: Well, I think the difference is, does the product support what the data team might want to do, and it does, and we just haven’t thought that anyone cared, so we don’t talk about that stuff too much, but it’s all there, versus we didn’t know… which your FTP comment suggests… they don’t even know that’s a consideration, let alone what the product needs to support in terms of features, integration, simplicity, like, whatever. You know, so that’s why I was curious about it; is it more the products are lacking, or they’re just not talking about it, and they should dial that up more?

David: I would say they're probably not talking about it. Because I think, you know, in 2026, everybody knows that this is kind of important, and we kind of need it. You know, I’ve been on a number of pitches over the past year or two where people have talked about, you know, there’s embedded Snowflake integration already. So, you know, if you have a Snowflake environment, you almost immediately have access to the data on the back-end, you know, but like, I had to know to ask that question, right? You know, it’s things like that I think would help, like, you know, your listeners understand, okay, when I go in there, okay, I know what I need to do to impress the IT folks, but to impress the data folks, let’s talk about how we can bring data into the ecosystem to support whatever their needs are.

Brian: So, how do you… when I think of something like, you know, like, I think at CDOIQ there are several vendors in this, like, context management space for AI, right, holding enterprise knowledge so you can improve the responses of the large language model applications and things like this. That’s very, kind of, amorphous. Like, how do you measure the value of, like, the responses are overall better? Talk to me about, like, when the intelligence is invisible, how do you decide whether to renew? How do you… if you have to make a business case, how do you do that? Like, what’s your process for some of these more ethereal, [laugh] ephemeral, kind of…

David: So, I mean, like, just taking a step back, even when we were talking about, like, these context tools and things like that we saw at CDOIQ, I mean, for those that really work well, it becomes how well is the software vendor and the company working together to even define those meanings? Because, you know, most companies… probably us included too, right… it’s like, do we have business definitions around a hundred percent of every term we use? No, we don’t, right? So, I mean, like, the very first thing that you’re going to need to drive out is, like, have we driven to a point where we have agreed upon definitions across, kind of, all the key things we do as a business, right? That itself, which is part of the process of bringing in one of these contexts, well, that itself, I think, would be one of the benefits that you could point to that you can say is, like, hey, revenue means X, Y, and Z now, as opposed to okay, this division thought it meant Y, and this division thought it meant Z that, like, you know, we now have these common things so that we can pull these things together. That’s the first thing.

The second thing I look at too is, like, okay if they’re saying they can make better decisions, it’s like well, how have we defined making better decisions before we even went in, right? So, to me, for these things, it goes back to something I think we had talked about before, was it’s all about the pre-work you do before that, right? It’s like okay, the technology, I think getting the technology in, working with the technology, that’s probably the easier part, right? I think it’s that pre-work to define okay, like, our success measures are going to be, you know, these three or four things, right, and we are going to explicitly define these three or four things as, you know, as specific as we can. You know, we’re going to be clear on business definitions, we’re going to be clear on inputs to measures, right, just so that everybody’s on the same page at the start. That, like, this is objectively how we’re going to evaluate this.

Brian: Right, so that’s the operational side, though, right? Like, figuring out when some, you know, a business person asks a question about, like, what’s our revenue, well, it’s always revenue and before refunds and after this thing, and it’s gross revenue, and in the US only, but if it’s not, it’s then in Canada it subtracts this, on and on and on. So, you get all that right so the answers, in theory, are now more accurate. But I guess in a way, this system is, sort of, intelligent. It doesn’t have its own user interface.

Let’s say you’re building an application, it’s a platform, so it’s being built on top of it, and I think what the founders are thinking about it’s like, well, how are you guys measuring or deciding whether to renew or not because it’s so invisible. Like, effectively it’s sort of invisible; you can’t see it each time unless there’s active product work to go in and figure out where should we make the value visible. I’m just, kind of, curious how you think about that side of it, like, to make a renewal argument. Or not; a non-renewal argument.

David: So, the way I think about it, too, it’s like, you know, most packages I would think, or most vendors, I would think about, they’re coming in there with a pitch to help you improve some aspect of your business, right, be it increased sales, increased retention, et cetera, right, we can quantify that, right? Like, we can say we can do, you know, the measures say, like, this is what it looked like before we had the tool, right? You know, we were seeing X amount of revenue or Y amount of revenue, but with the tool, with everything that we’ve sort of agreed upon as, like, the general direction, right, we have seen this lift. Now, again, this is probably a bit more biased to the spaces I focus on, which are sales, marketing, and retention, but still, like, you can still do that, kind of, pre-post analysis there to say, like, yeah, we have seen a significant revenue lift, or yes, we have seen a significant reduction in customer churn. And you know, that’s the argument for the renewal for it.

Brian: Mm-hm. Do you build that yourself most of the time, or did the products have some kind of… one of my founders called it a meter… like, there’s a metering experience in the product such that you can see this stuff without… like, the look at what it’s doing. Here’s the reporting it’s built in versus, like, no, you go run your own analytic? You know…

David: So, we would tend to do is, we’d also build our own analytics around it too, right? Because there is yeah the meter too, but like, you know, it’s the we trust you, but we still want to verify your numbers as do, because, you know…

Brian: Trust, but verify [laugh].

David: … we just want to make sure that it’s got our lens on it and not… we don’t want to be like your mother grading you, right, because we figured that would probably be like a really great one.

Brian: [laugh]. Right. So, you talked about this, like, single biggest miss and the data as a second-class citizen and that this is a miss here. Is there a general place where they are spending too much time? What is the space being filled with in the conversations about features or whatever else it may be that it’s, kind of like, “That’s nice, that’s nice, that’s nice. I don’t care. I get that, but…” where are they wasting time?

David: So, again with most of the pitches I see, they spend a lot of time… I see a lot of pitches on, like, the technical features of it. It’s like, oh, cool, it can summarize, you know, this pieces of data for you, or it can summarize these things for you, right? And it’s like, cool, but I kind of expect your tool to do things like that, right? You know, I do like the technology, but like, I also want to see, okay, what’s the lens you’re going to apply to the business here, right?

I think I’d mentioned this to you before. You know, pitches I’ve seen where I can tell they’ve read our SEC filings or listened to our earnings call generally have a better track record because they’ve seen… they’re at least articulating in terms of, like, major, you know, macroeconomic things that we’re trying to accomplish as an organization. So yeah, basically in short, probably a little too much on features, probably not enough on business [laugh].

Brian: Got it. Is there a particular story you can talk about where, besides the salesperson or I don’t know if it was the, you know, like a founder and it’s more startup stuff, or whether you’re talking to a seasoned software company coming into you… I’m sure as a buyer, it feels better when they’ve done their homework and it feels like they have dialed into what your current business situation is, and has listened to earning calls, beyond just having listened to the calls, how was the buying process different? Like, where did that make itself very obvious to you that it changed the conversation and it helped you decide to do business with them? Like on a very, like, tactical or productive level, that reading the earnings calls or sitting and the listening to them and all of that, tell me how that helped.

David: Yeah. So, I mean, a particular example I can think of here is, you know, earnings call, we always talk about sales, we always talk about customer acquisition, so you know, it was specifically targeting their feature in their application for the platform and saying, like, “We heard on the earnings call this is one thing you’re going after. Let’s explore how this can help you with that.” And it’s not so much that they come in and say that, like, okay we’re going to solve this for you immediately. No. It’s really more of, like, now these vendors know where to ask the questions, right, the questions that I get interested in.

Because, like, I don’t think anybody, unless they were coming from inside the company, right, really knows all the problems or issues we’re trying to address, right, but if they have an idea of where we’re headed, then they know how to ask those next-level questions, right, to, kind of, get at there. And I think, you know, it just shows an interest in the company, it shows an interest in my problem because at the end of the day, I’m interested in my problems, not their problems. So, you know, it helps build a better rapport as well, too. So, you know, it feels like the vested interest there, as opposed to saying, like, “Hey, cool. We’ve got this feature and we did $22 million in, you know, this round of funding.” “Great, but I have a problem at the end [laugh]. How are you fixing it?”

Brian: [laugh]. How is your 22 million in funding helping me?

David: Yeah.

Brian: Are there other hidden costs that some of these product companies can’t see? It’s like they’re, kind of, clueless to what’s going on inside Comcast as you try to decide whether or not to go with this vendor or that one? So, configuration burdens, user experience, probably, like, oh my God, this is so different than the last thing and they’re going to retrain 4000 people. Tell me about some… I’m making some up, or you maybe those aren’t real ones, I don’t know, but tell me about some of these hidden things that they don’t seem to be talking about and they maybe they seem a little clueless about, they just haven’t addressed it, and you have to live with that.

David: Well, I think it, kind of, goes back to, like, the integration question, right, because a large Fortune 50 organization like mine, right, you know, we’ve got 40 years of purchases and systems and things like that in place, and so it’s, do I need to constantly be, like, swiveling my chair to do something? Like, okay, now I’m in your tool, and I need to do X. Now, I need to go back over here and do Y, and then swivel back over here and do Z again, right? You know, so it’s things like that… and again, those get back to, like, the second and third level questions, which are, kind of, going through the process to understand, like, what is that experience, right? Because, you know, if I happen to have a user that’s swiveling back and forth between systems, they’re going to have a bad experience, right? I mean, the UI could be amazing, but again, just the constant switching is something that would be a detriment to the experience. Probably things I don’t know that a lot of companies or vendors might be thinking of. Maybe they are. I don’t know.

And then I think the other thing, too is, like, you know, it, kind of, goes back to the data integration conversation we just had, too. It’s like how difficult are you going to make this for me to integrate the back-end of the data in. I mean, we don’t have to integrate everything, but like, to bring the data in that I’m going to need to tell the story to justify the tool, to just to or, you know, tell the story of how this improved the customer experience, et cetera, right? So, it’s things like that, like, I don’t know that I see a lot of attention given, in some of the pitches I sit in on.

Brian: It’s funny because I would think especially if they’re selling into your department they’d be keenly aware of, [laugh] you know, wanting to be able to pull data out of the boxed-in environment, the walls of the product so to speak, and really be thinking of it more as a something that sits across the systems and all that. You talked about the chair swiveling, right, between monitors or screens, different going back and forth. I think we’ve all experienced this before. Do you guys have a process for evaluating that? Like, is this, like, any kind of formality around that to decide, is this unnecessarily a burden on the staff that are going to use it, such that, like, all the gains we’re getting just result in more labor costs because it’s three times more complicated, but we are going to get this data out of the back-end, but we’ve just inherited all this new, like, overhead that we didn’t used to have because it’s so complicated. Talk to me, like, do you pre-screen for that? Is it kind of like a surprise that you don’t want to find out later? Like, tell me about that.

David: A willful ignorance. No, no. Um…

Brian: [laugh]. Not will… but sometimes you may not realize it until, like, the workflow is really going, and then you see it, and you’re like, oh my gosh. You’re, like, four application windows are open, and type this into the command line, and then go over here and wait for a bundle to come, then upload it here, and do a transform, and bring it in here, and then get permission from legal. Like [laugh]…

David: Yeah. I would say as part of, like, the IT evaluation of these things, like, the interfaces and things like that are definitely looked at. I think one thing that probably every everybody could probably do a better job of is, like, how is this fitting into your business processes, right? It’s like, you know, you can add, you know, any ERP system, if those are still a thing, you know, Workday whatever, in there but, like, how is that affecting the process workflow? I get it maybe from a data perspective it might be pristine but, like, if we’re having to swivel, like, we need to identify those things and I think that also needs to get factored into, like, our own metrics of how we measure success.

It’s like, you know, to your point, it’s… okay, great, you know, we got all the other things, but now, you know, Brian’s having to swivel between five different screens, and that has cost us X amount of hours or things like that. So, like, I think those just need to get built into, like, the evaluation criteria. And, you know, sometimes it happens, sometimes it doesn’t happen, right?

Brian: Yeah. Yeah, can you think of a time that a… and again, you don’t have to say the name, but a vendor lost a deal because of that or you didn’t renew after maybe after the fact. You found out, like, the downstream cost was like more than you it’s worth.

David: I’m trying to think, you know, in my career, has that ever happened? You know, I don’t think I’ve ever witnessed it, to be honest with you. So, I think I just talked a lot about something theoretical here, but…

Brian: [laugh]. That’s okay. No, it’s good feedback. Is there any other stuff you’d like to add about the hidden burdens that are preventing you from moving forward with things. I mean, we all know procurement and, you know, legal, and it’s just there’s a backlog of other requests and all that kind of stuff, which is really hard for a vendor to control for, but other things that might be blocking a deal from going through that they don’t realize because it’s not coming out directly, but maybe you can’t say it out loud to them? Anything like that. Just hidden costs that they’re not aware of that maybe they could be addressing somehow.

David: I don’t know if it’s a hidden cost or if it’s a missed opportunity. And what do I mean by that? This kind of goes back to, you know, my comment about roadmap, and then my comment about, like, you know, earnings call is, like, are the vendors coming to us uniformly and saying, like, “Okay, this is the direction your business is going. You’re a Fortune 50 company. You’re in the telecommunications business.”

You know, are they coming to us with, like, a vision of how their tool can, kind of, help evolve our strategy? Not necessarily a hidden cost or anything that would prevent a deal, but it could be a sweetener to say that, like, okay, this is somebody who really wants to do business with us. This is somebody who really wants to partner with us. You know, is there an opportunity there? Because I know if somebody came to me and said, “Okay, you know, we’re looking to enhance, you know, your sales and retention AI experience by doing X, Y, and Z,” they have my interest [laugh].

Brian: Yeah, go figure [laugh].

David: Yeah [laugh].

Brian: Focus on the customer. I mean, we’re laughing because it’s so obvious, but apparently it’s not happening, and I’m wondering, do you see any patterns here between when you’re talking… I don’t know how often you talk to founder, like, startup-level companies because they’re the new ones in the space or you guys tend to skew more established companies because of your size, but is there a difference between a founder-led sale and a sales-led sale, in either the questions being asked, like, talk to me about some of that. What’s missing as you move to sales-led sales?

David: What’s really interesting is, like, when… and I do get some founder pitches, right, and when I do hear those, what I hear is the passion for the vision, right, like, instead of, like, a sales team coming to you. What I hear is, like, “Yes, we’re doing this one thing, but it’s in service of changing the world.” And I’m like… so like, I hear a lot more of those, and, you know, it’s definitely intriguing to me. Whereas, you know, if you get from a more established, larger vendor, it’s a bit more of, like, a typical sales pitch or more of a typical sales process, right? The founder-led ones definitely try to sell me on a vision, which, you know, at times can actually feel, I don’t want to say inspiring, but it’s like, you know, here’s someone who’s really trying to make a difference, rather than just make a sale, right? You know, like, they believe in the product.

Brian: It’s not transactional, yeah.

David: Yeah.

Brian: Would you say that’s a very noticeable difference?

David: At least with the pitches I’ve seen, I’d say yes.

Brian: Got it, got it. David, this has been really great. I appreciate you opening the doors a little bit here on what’s going on, inside at least your experience to date. Are there any other… like, I kind of wanted to give you the last word here, but any other closing thoughts for particularly founder operators in the analytics and data space selling these kinds of products that you want them to hear that you didn’t have a chance to say so far?

David: Well, I mean, I’d say, if you’re a founder-led startup type situation, you know, shoot your shot. You know, reach out. Try to get the business with us, right? I would think being a founder, you’ve got a big personality to begin with. You know, use that to your advantage, right? I think there’s still a lot of opportunity and don’t get intimidated by the size of some organizations.

Brian: [laugh]. Got it. Thank you for sharing this. Where can people follow your work and get in touch if they want to? Is there a good place to do that?

David: Yeah, absolutely. So, I’m pretty active on LinkedIn. I publish a newsletter there, and also pretty active on Substack these days, so.

Brian: Great, awesome. I will put those links in the show notes for people. David, thanks for coming on Experiencing Data.

David: Thanks, Brian. I really appreciate it. Great time.