Value-First Data - Mar 17, 2026

[00:03] **Introduction** Klemen Hrovat: Good afternoon, LinkedIn friends, Value First Nation. Welcome to another episode of Value First Platform, talking AI data readiness with Tricia Miriam. Happy Tuesday, Tricia.

[00:18] **St. Patrick's Day Greetings** Tricia Miriam: And happy St. Patrick's Day. I'm not wearing green. I could have, but I only thought of it just now. What do you know? Klemen Hrovat: Oh man, so easy for me, and I should know this one as I'm 25% Irish. Um, yep. I got a little green. Casey Chrissme earlier today, a little green in the logo here. But uh, that's about it. But maybe some other shoes. Actually, my socks are super green. So, check that box. Tricia Miriam: Okay. I have some work to do. I got nothing. I got nothing. Klemen Hrovat: Yeah, you got a bunch of green in the background, so you're covered too. How about that?

[01:00] **Shadow Data Discussion** Klemen Hrovat: Um, shadow data is on your mind today. Um, and to the surprise of no one, I'm going to relate this to unified customer review throughout the episode. Um, because it is a big part of that story, like, right? Like shadow data is required for the unified customer view that currently exists inside of your organization at any time. It's just usually like you're pulling that data from conversations and from spreadsheets and email inboxes, and then we're preparing it for one moment of time to not try not to lie about the current state of things with customers or with leadership or with whoever, but it's a fleeting moment. Um, and when we think about AI and readiness, like that moment never goes away, right? So if at any time shadow data like shows back up and data is not aligned, there's gaps. Obviously, just like the humans, we don't want to be misinforming with that data, like AI is not going to do a good job. The difference is that AI will always do something with it and pretend like there is no shaded, shadow data, right? Tricia Miriam: Wait, what do you mean by that? They'll, AI will always do something with it, but only if you share it, right? Klemen Hrovat: Well, it'll do something with the data no matter if it includes the shaded, shadow data or not. Tricia Miriam: Okay. Klemen Hrovat: Right? Like, yes, humans might not be able to act like if they're missing important context or they can just see that something's missing, right? AI doesn't have that mode for the most part, like it's, it's not going to intuit that something's missing. Tricia Miriam: Yes. Klemen Hrovat: Unless there's a serious instruction telling it to look for that thing, right? Tricia Miriam: Yeah, I've had, I've had it, um, lecture me before about, you know, like you, you left out this and we should be incorporating XYZ and I'm like, no, I'm sorry. I didn't, I didn't want to talk. That's all happening. I didn't want to talk about that here. I just want to focus on this over here.

[03:50] **Latest Experiences with Shadow Data** Klemen Hrovat: So, what is your most recent experience when it comes to, uh, shadow data? Tricia Miriam: Um, so I had a win recently in that there was a very important piece of data that finally did get brought into Hubspot, which was revenue. So Stripe payments were sitting off in another world all by themselves. And so that is a win that I had is that we were able to integrate that with Hubspot. So now we've got some revenue data flowing in. But, you know, that's a perfect example is that if you are transacting on a platform somewhere and you don't have that revenue data coming into Hubspot, how are you giving? Um, how are you doing anything? How are you? If you're running ads, how are you communicating to the ad platforms that certain leads are more valuable than others? How is your sales team prioritizing which deals to work on? So that, I mean, that was such a slam dunk for people. But something that you and I have talked about is call transcripts. And of all the companies that I work with, some of them, many of them have their note-taking platforms integrated with Hubspot so that you can go from a meeting in Hubspot, a meeting object, and click on a link and go see the transcript in another platform. Some of them, their call platforms will actually send the meeting summary into Hubspot. But 0% of them are storing the transcripts in Hubspot. And note-takers are not, they miss a lot of stuff. Klemen Hrovat: Yeah. And I personally believe that that AI needs access to the full conversation. And I I still don't know if Hubspot is the correct place to store it. I just know that AI needs to be able to see it at some point. Um, yeah, and it's a very interesting, like, and this is how much of um, you know, SAS is built into AI and straining data. Like most of the time when I interact, it's like, oh, of course we don't want to keep this transcript. Like that's not what you do. We need to summarize it because then it'll be smaller, it'll be easier to maintain, then you don't have to read it. Like I don't have, I'm never reading it, right? You got to do that for me, right? But that's been like best practice. Like we put that somewhere away, you know, and then make decisions in the moment. Um, I was, I was telling Casey about this, like when like we developed some coaching. And this is probably like the most important use case, I would say. Uh, like if you're going to do coaching on calls and call transcripts, it cannot be on some summarized version of those things, right? And like Ryan built it out first, right? And his output was like 100 transcripts reviewed and like 20 session summaries and it was like a brilliant output. And then when I did mine, it was like 95 session summaries and like 15 transcripts and all those transcripts were with the same client, right? And that was like, but I got the same coaching spread that Ryan did, right? And that's where context engineering happening in that moment, right? I know things aren't being taken into account. And when Casey shared hers, like she's only got one client in the repo represented, which happens to be a challenging client sometimes, right? So her coaching was related to all those specific conversations. So that's a a great example of there's so much nuance in the human language and unless you're putting together some really sophisticated like, you know, rag or tokenization of of the bulk data, like leave it in there until you hit some kind of limit or barrier. Right? Because usually if you're not in IT and you're asking that question, you you don't know how to do this correct, so just put it in there. Just, just put it in there like AI, put it in there and then ask AI, right? Like don't make the call. Like, well, this is going to be a lot of work. Right? Like, oh, this is going to be too much context. It will not be um, in most cases for for AI. So, I think the unstructured data situation is like, it's getting so much better. And yeah, another moment I just like, like bring the shadow data into the light.

[10:10] **Avoiding Organizational Defensiveness** Tricia Miriam: One of the, one of the bullet points when we were, when I asked Claude for a suggestion on what we should cover, one of the bullet points was how to do a shadow data audit without triggering organizational defensiveness. Klemen Hrovat: Unified Customer Review. That's how you do it. Next question. Like, I'm not even joking either, right? Like almost every other avenue is a trap, right? And it's traps that are why we have not solved marketing and sales misalignment. Um, and like leadership not understanding what the front lines are doing. Like nobody has the visibility they need to ask the right questions required to find the uh, the shadow data, right? And it's so hard to, like we have this moment of where you can use the system in this case, but really it should just be the document, like this is our unified customer view. AI sees it. Now AI can ask the questions like of of the humans. So that's one value of AI, like you can make it an objective conversation because the human's not asking you for that thing. But almost certainly, like one department is better at understanding the context of the organization than the other. So it automatically starts from a from an from a team versus team scenario. And it's just hard. Like we're inherently built to like try and figure out what you're, what you're up to. Like, why are you asking me for that? Why do you need it? Just tell, just tell me what you need, Tricia, and I'll get it for you. Like I don't know what you have. I can't see in the shadows. Man, that question like bugs me. Like, do you want us to integrate the systems or not? Tell me what data you have and what you need. So instead, you're not going to tell me, you are going to tell the leader or the concept of a unified customer review. And it's, and it's working. Um, so that's and to our points last week, like of why the the AI leader.org needs to be able to understand this, this and this will trigger human org defensiveness. And if you do any of those things, you're like you, you just delayed everything by like six months or forever depending on how bad it went, right? Um, so yeah, I probably don't need to over, over, uh, explain that one. Um, I mean, I I guess I should like have you seen anything else work, uh, Tricia? Tricia Miriam: Oh, uh, no. Klemen Hrovat: There we go. Tricia Miriam: Um, okay, so next question, you used to be a firm believer that Hubspot should be the source of that unified customer view. Have, have you changed that opinion? Klemen Hrovat: Um, it really, uh, it depends on who's using what system, right? And it really is user-dependent because the reason that we can get out. Like if we get out of the tool conversation, that means we're also getting out of the which one is the source of truth conversation. And we're just pushing data like around where it needs to be. And like we just had a call where it's like, yeah, we got a report on KPIs, which ones are coming out of Hubspot, and which ones are coming out of Salesforce. And the reality is we need many of them in both, both places, which means we need the supporting data in both places. But marketing's over here, they need to see certain things. Sales is on the other side, they need to see certain things, right? It should reduce the effort like of of needing to create a sales-specific view in Hubspot. But that's where often there's a team like you just need the users making the decision. Again, this is another spot where the reasons or source of truth and like um, you know, system of record and all these phrases were to manage complexity and maintenance and all that stuff. And now AI is around to do that for you. So, we can finally put the humans like in a place to be successful.

[15:47] **Tools Within the Conversation** Tricia Miriam: Um, so tools as much as we don't want them to be a part of the conversation, they always have to be because in my, I've never seen an organization where every team works in the same tool. You're lucky if you have two teams using the same tool. Um, but operationally, talking about shadow data and unified customer view, it is, you know, if you're trying to give everyone a unified customer view, how do you do that when you have an organization running across multiple tools? How do you do that effectively? Klemen Hrovat: Uh, you ask and then, like so tools come in when we get to the how, right? Like so like in theory, we could put all the tools away and then just write on a whiteboard what is on our unified customer review, right? And then if we do that, then it becomes a question of where is that data, who's in charge of that data, and how is that data going to get to the customer review, going to to the unified view of the user, right? Um, and since we've framed it that way, we can start having conversations like, well, yeah, uh, we've been putting that in Salesforce, but it takes Trish a lot of work to maintain the system to get it there. Hubspot's just built out of the box like to have that data. We're already sinking the systems, so why don't we let Hubspot capture that data, right? Um, I think that's going to happen with a lot of Zoom Info integrations with all of the data available that's in Hubspot now, you don't necessarily need all the Zoom Info data. Um, So just as as one example, um, because when you can get that, like you do need to ask the question of where is the data and who's in charge of it, right? And that's like the the trickiest conversations. Because eventually you get to the, oh, it's on Chris's desktop in a file that he maintains locally and nobody can see and we just updated that file last week. He probably hasn't had that file, right? Um, the only way for us to know that is if Chris is taking us through his process. And like we see when it's like, okay, you said this is a part of your unified customer review. It's not in any of these systems, so where is it? Or it might be in the system and it's not up to date, but we know you closed the deal last week. So, and you just mentioned it on the sales call. You mentioned the closed one reason and it's not in here, but you know what it was, so it's either in your spreadsheet or in your head. Right? And that's okay, now we found it. Now, how can we make it so it's really easy for you to uh, you know, add that data to the system so everybody else can have the unified customer review too, right?

[20:24] **Chris Carolan's Six Steps** Tricia Miriam: So I just told myself a bunch of stories and I think, okay, and how I interpreted what you just said was that in order to so we it's a huge task to make every every tool a unified customer view, huge because of how many integrations you have to put in place. And so what comes out of a discussion and alignment around unified customer view is alignment on which tool teams are primarily going to go to for that view and a prioritization around like if multiple tools are going to be supported, here's the information that will be supported across all those tools. Did I get that right? Klemen Hrovat: Yep. And that's where I mean the the challenge is doing it from the tool side first. Like because then we're letting the tools decide what things need to be named and what gets integrated and what doesn't. When you just flip it around, you're much you're in a much better position to handle the conversation like required to deal with the challenge of like, one, maybe it's okay. Like it's okay if the AI, if AI can help transform this data into the name that I want and the name that you want, right? One of Hubspot's updates today was um like the ability to change the name of a workflow when it's on a one-off workflow card on a record, right? So if you want to enable a salesperson to just trigger a workflow, naming it something like close this deal workflow is better for the salesperson than uh close dash deal dash this dash 31726 dash, yeah. Whatever the admin has decided, which they should decide what the naming convention should be. Yeah. And that's a moment that I'm seeing all over the place right now. Like, right? That's one of the reasons like you have to develop this understanding and help people understand like, no, ask for what you want. The transformation of data, the management of data across systems is easier than it's ever been and it's only going to get easier. Of course, it's easier when Hubspot is one of those systems. Um, and so it's like, so what I tell people is we cannot make any decisions based on perceived difficulty or perceived cost or perceived how long is this going to take? Because almost none of those answers are actually true anymore. For how fast things can go, how hard it actually is and how much it actually costs. Because guess what, those three things all like tie in together, right? Projects are were ridiculous costs because they would throw like 600 hours on there. And like none of this stuff takes that long, right? Because you don't need humans to do 500 hours of the work anymore, right? Um, so again, it's easier to say things like that when you start from a place of alignment that we're trying to get this outcome, um, because there's a budget trap in there somewhere. So many traps.

[23:31] **Chris Carolan's Six Steps to Dealing with Shadow Data** Tricia Miriam: Okay, so if I'm going to read your mind and articulate Chris Carolan's six steps to dealing with shadow data. Number one, which I think is going to still surprise everyone, but the number one thing is getting alignment across the team on the importance and priority of developing a unified customer review. So that's step number one, you got to get the whole team championing the idea that the business benefits from being able to see that in a single place, or just having a unified customer view, every team benefits from that. And then the second step is to begin having conversations with each team about, well, where is your data stored and what is this data? And it's only then then you can start that you actually know what your shadow data is. That's the path to even identifying the shit the shadow data. And the first two steps should also help you then have some context to prioritize what should be brought into the unified customer review. Klemen Hrovat: Yeah. I think I mean, I don't know if you can start anywhere else, but like business goals. Right? Like what are we trying to accomplish? And then it's very unlikely that the existence of a unified customer view, if the goal is related to growth or customers, like journey or anything, like unified customer review will serve that goal, right? So you got to make sure I'm trying to figure out the balance between prescribing without knowing the problem even though we know the problem is when we come into these portals, like your data is not good. Right? Um, and so I think always starting there is important. And then like once you establish that, and not even from a documentation standpoint, you just get people nodding and saying out loud, yep, unified customer view. It's like nobody's going to say no to that, right? They have they have no idea what it takes to get one of those. Right. We don't at that moment in time either, so that's why like another thing it's doing is when we show up to the conversation and are constantly like, oh, maybe, oh, it depends, oh, let's figure it out. Let's have six more conversations and then maybe we'll figure it out, right? That's not helping anyone. And most of the time when people come asking for this stuff, they just want to know they're on the right track. They just want to know like what first step to take and be confident that that's the right first step because they know how how how challenging these conversations are, right? So, business goals tie it to unified customer view, then what systems and capabilities do we have in the organization like creating that data and and where is it? Um, and it could be in places. This is where it's like, it's almost it's, it's guaranteed to be inside of the org, it just might be in somebody's head. Like these the concept of like, oh, it would be cool to see, you know, I saw this on LinkedIn the other day and this person that we this ICP was posting about this. And then they commented and then I commented and then we got a conversation and then we won the deal, right? So it would be cool to see that in Hubspot. Right? The data exists over on LinkedIn still, but the the shadow data is in the person's head that they know that context led to, you know, this this closed one reason, right? And that's the kind of thing where if we've never tracked data like that and we put close closed one reason as a dropdown, we probably don't have like LinkedIn engagement like as an option, right? So again, that's why you try to pull out of what these tools, um, are doing and how they're set up. Um, So how many steps was that? That was That was three. Tricia Miriam: I think it was maybe four. You added one, align it like, define the business goals first, then align on unified customer view, then identify all of the systems and where data is being and then you can prioritize what you're going to bring in. Klemen Hrovat: Yeah. And that prioritization. Hang on. Like I'm sorry, folks. Like unified customer review is the easiest part of the process, right? Like deciding that that's going to be the goal is the easiest part, I should say. From there, there's so much that can get in the way. And and where I went there was the ease of deciding which data or what's prioritized can be based on systems proximity, the cost of doing that, the humans involved and whether they're like cooperating or not. uh, and like and the cost are both a financial as well as a time perspective. Right. And often, like the initial goal has to be like, okay, how can we take it from 10% done to 20% done so that we can show that there's progress to be made, everybody understands it continues to be a good idea and then you start to like grease the wheels on anything that that's not maybe moving as fast. You also get more help from AI as the context starts to fill in. Um, so there's almost always opportunities to find things that are already being done and just move, move them. Right? So if you're typing the data into your email, maybe you can type that into Hubspot and email from there instead. Now all of a sudden, like a bunch of fields are getting getting filled out. So and all of that is like conversation, right? So step five is like having the conversations you need to have like while building the proof of concept and making sure everybody's clear on progress. And six, I think, like, profit. Done. Done. Um, in reality, then you get to move to Unified Revenue view. Which we're going to be talking about in the next session in a little bit. Um, But yeah, I mean if you're out there and you've tried this uh, or don't believe it's going to work or you know, whatever is coming to mind and you have other ideas and other things that work, like I'm open to them, but like it's just this human element that I think SAS as a whole has disregarded in terms of uh, like what's required for adoption of these systems. And the same things that have gotten in the way there, like get in the way of AI. And and any readiness like related to that. Namely, personalities. Prioritization in terms of like, I this is important to me. Um, and that's just human nature when you're competing against what somebody else wants, competing for their time, competing for their attention. Um, you have to have a way of showing them, you know, why they want this. You have to have them like believing that there's something really good in this for them before they're going to get on board with you. Um, and being open and transparent about what they're doing and share what they're doing with you. You have to get that first before they'll actually do do the steps, the sharing steps. Um, and in the many traps that you have shared, uh, there there aren't very many great ways to go about it, uh, but unified customer view is one that seems to be at least at this point indefensible. Klemen Hrovat: Yep. And it doesn't mean you get to sit on your laurels and then everybody's just going to fall in line. That's like the proving part. Like you have to constantly show and get agreement. That's why it's like like do we agree that this unified customer view as we're building it like is still good? Like everybody understands what's going on instead of, hey, we're getting this data into Hubspot. Thanks guys. Like, do you like what it looks like in Hubspot? Like, nope. Like they're just going to find ways to to say no or I don't understand Hubspot or there's too much going on. Like everything you've ever heard in your life as an excuse to not change, right? Like we're kind of built for that. Um, and this ongoing effort to like just confirm and this is where AI like helps take care of the admin required to just be human with each other. Like create the space for for that extra confirmation, right? That people are still understanding what's happening. It's still valuable for them and or you're putting the leaders in a position to do that, right? Because any of this starts to touch like commissions and KPIs. If you're not in charge of that team and you're suggesting that there's going to be an impact to the way that they understand what's on the screen and it relates to those things like don't. Like find the leader, align with the leader, let them do that. Right? Because in their head, that's where it's important to understand with shadow data, they have a unified customer view, like in their head. They might be able to forecast to next year super, super accurately even though none of the deal data is in the system, right? And they might not even have a spreadsheet. Like that's what like sales people just get good at human relationships and remembering what people say, right? And again, that's what also makes humans a unique element of this puzzle and still highly valuable in these contexts. Um, so if that's where you got to go to start getting that information, like it's uh it's a fun challenge for me and I think it can be fun for most. You just have to not retreat from the the idea from unified customer review. Like it's so easy to want to explain rationally, right? It's almost never an irrational challenge coming. I mean, it's never a rational challenge. It's so it's kind of like, it's almost political in nature. You're like, I I whatever you just said, unified customer review. Yeah. Sounds good. But we can do that because we know for a fact that will be a good outcome for both AI and the organization.

[30:27] **Breeze Story** Tricia Miriam: Such a good outcome. The best. I've got, I've got a good little, uh, story about that if you have time. Klemen Hrovat: We do. Tricia Miriam: I, Casey and I are working with a team it's a it's a situation I've seen a number of times, which is a marketing team that is too small and asked to do too much and they never have any time and they're overwhelmed and I would even say traumatized. Like they're just exhausted. Um, and so Casey and I are working with them and showing them some things that they can do to take the load off. Um, part of that, as you and I have talked many times is training your AI with context. And and and backing up into that is helping them teams, the teams understand how they can use AI to help them build context in the first place. And Breeze, Breeze wasn't happy with Breeze in the beginning. I wasn't happy with Breeze in the middle. But I think as of like a month ago, maybe six weeks ago, I've seen a shift in how well, how good a job Breeze does of like answering questions about your own portal. Still not happy with using Breeze to answer how to do things in Hubspot, but if it's looking at the data in your portal, something has shifted. So, we asked the goal was to start using AI to write emails about these, about this company's products, but first we had to teach it what the products are about. So we asked Breeze to generate a product summary, like a product overview for one of their products in particular. And it did a very good job and it was far more details detailed than I thought it was going to be and in the, I don't know, somehow, maybe the dialogue that was happening as it was building it, I could tell it was actually going into sales conversations that had been recorded and stored in Hubspot. It was actually picking through sales conversations to get some of the data that it used to generate the summary. So, if that is interesting to you and if that's something that you want like, what a beautiful way to like get the information out of your sales team's head, like this is what resonates with a buyer and get it directly into the hands of marketing that can use it and be more effective. Ooh, it was so beautiful. Klemen Hrovat: Those are the best moments. Aren't they? Like, because it's like yeah, it's the way that AI is supposed to work, right? Um, and when it does work like that, those are the moments where it's when it's inside of these other change change initiatives, like you just got to leverage, you got to be ready to like leverage that, right? Um, and build trust. Like those are trust-building moments. And whether you know for sure it's going to happen the next time, you start to lay the groundwork for um, for the trust in this. And that's why you're always showing, you're always transparent with these teams because the thing that gets in the way is the uncertainty of it all. I know my process even no matter how much shaded up shadow data it has, I know it works. It's worked for me for a long time. So even though what you're explaining right now makes sense, if then I go click on the thing or you show me how it's supposed to work and then it doesn't work that way, right? I'm, I'm good. I'll spend an hour writing an email and scheduling meetings and like using handwritten notes to understand what's going on because I know it's a known quantity for me, right? Um, so yeah, very cool to to hear about that story. I had several moments like that yesterday when I was unboxing expertise, uh, AI, uh, which is now on the Value First Team website. Um, but it's doing things like that. It's bringing like just uh, conversations, like get your conversations into Hubspot. Uh, get them in front of AI and it can it can help you out. Well, uh, I I enjoyed this this topic of of shadow data. I hope I hope you did as well. Tricia Miriam: I definitely did. Klemen Hrovat: So, until uh, we see everybody back here next week, great Tuesday. Thanks so much, Tricia. Tricia Miriam: All right. Take care.