Unlocking Data Potential with Claus Torp Jensen

This is Making Better Decisions. I’m your host, Ryan Sullivan. Decisions are where rubber meets the road for organizations. Each week, we’ll be learning from people who are on the front lines of turning raw data into better outcomes for their organizations. This show is sponsored by Canopy Analytic, helping companies make better decisions using data.

Ryan: Hey everybody. Welcome to another episode of the Making Better Decisions podcast. Today’s guest is a distinct distinguished C suite leader with over two decades of experience in transformation and technology. He’s recognized as a healthcare and tech thought leader with numerous publications and patents, a critical thinker and lifelong learner who provides valuable perspectives to companies and boards alike.

Uh, to that end, he is on the board of, uh, and or advisor to numerous companies. Recently unretired to become the new CTO of the [00:01:00] University of Texas at Austin Medical Center. Please welcome Klaus Torp Jansen.

Claus: Thank you very much. Good, how are you?

Ryan: I’m great. Glad to have you here. So if we jump right in, I’m going to give you the same starting question that we give everybody, and then we can get into some, you know, specific conversations, see what threads to pull on.

What is the one thing you wish more people knew? About using data to make better decisions.

Claus: That is always about having good data. the one thing, right? We talk a lot about having a lot of data. Yeah. we talk a lot about using data, but we don’t always talk about, you know, is the data that we have actually good, right? And good means, do we know that it’s the right data? Is it authoritative?

We’re not talking about quality, but just for starters, knowing it’s authoritative data, it’s kind of interesting. Because if you think about it, how many times have you been in a scenario where people say, well, we could take the data from four different places? Maybe you should make up your mind in terms of [00:02:00] which one is the authoritative source.

So that’s my one thing.

Ryan: Yeah, I love that. I think as somebody who’s kind of been a part of a lot of those discussions, right when we’re coming in and either deciding to build some, you know, whether it’s warehousing infrastructure, new reporting initiative, or, you know, advising on strategy for how to decide on KPIs and roll those out to a team.

Like we do always kind of come back to this question. So. I obviously have my own opinions, but I don’t want to bias you. What are some suggestions that you might make to someone who’s going through that process? Say somebody hears that and they identify with that pain point. What are some suggestions you can have for thinking about, okay, well, how do I pick which four?

Do I pick the one that has the number I like the most in it? Or, you know, how do I go about that?

Claus: Uh, it’s a great question and it’s hard to give a short, sweet answer. But I think the short and sweet answer, we’ll talk about what that means is you have to start your life cycle management of data earlier.

Ryan: Hmm.

Claus: When you talk about [00:03:00] data warehouses, when you talk about, uh, lake houses, when you talk about data fabrics, The conversation usually starts assuming that you have the data

and then what you do with it, right?

Okay, you got to push your lifecycle management further left. You actually need to start your lifecycle management with two steps before that. The one step is sourcing, right? So if you go back to good old master data management, We spent a lot of time and effort on sourcing the right data, and we sort of forgot that as we went into this world of big data and AI.

I don’t actually think master data management through and through is the right approach, but I do think some of the tools and techniques that allow us to single source data, and I really mean the single source. I only want to have. One source of one thing. Some of those techniques you should apply in what I call the sourcing step.

When I think about the life cycle of data as it throws through to, you know, [00:04:00] analytics use. And so in the systems of insight sense, life cycles start with sourcing. And then the second step, which we tend to also ignore is ingest. Which is actually making sure that I have that good data I decided, which is authoritative source.

And now I have it up to date just in raw form. I haven’t done anything with it yet. But I placed the raw data in a place I can do something with. There are so many, you know, different solutions out there to talk about data federation. I have yet to find a solution that does a very good job of doing join and memory. So I still remember when you had to write code that did joins, right? So I know how hard that is and it’s been baked into databases and do a great job of it, but trying to do it in memory is slow and inefficient and usually pretty expensive, right? So this whole thing about, There are just some things you can’t do if you don’t move the data.

So that’s what I would suggest. You know, add two steps to your data lifecycle. Add sourcing [00:05:00] explicitly and add ingest explicitly. Move good data to a place that you can manipulate.

Ryan: Yeah. Yeah. I, I love that. And also my, my tech nerd brain love, you know, but that was the tech language version, many times I don’t have the benefit of talking to someone who knows what a joint in memory is. Yeah.

Claus: a good idea. So, so if you have data from two different sources, It is not integrated out of the box. Uh, and let’s say that you have bits and pieces from one and bits and pieces from the other, and what you really want is a view that loops across. So if you think about what you need to do process wise to make that work, you have to map all the identifying keys from the one source to identifying keys from the other source.

The data is probably not going to be in the same [00:06:00] order. So somehow you have to mix and match and map these different data pieces together. The technical language for that is a join. It’s a rather complex operation that requires you to understand the totality of the data on the one side, the totality of the data on the other side, and then merge it.

And that merge, if it isn’t done inside a database, has to be done using some kind of programmatic construct. That’s the thing that ends up getting to be expensive. CMO,

Ryan: I think you did a really, really good job of explaining, you know, what, what I do for a living. You know, it’s like a lot of figuring that stuff, what I think every data professional is doing, right? Like pulling from lots of different places and then somehow making sense of it. I actually was just talking to a very You know, intelligent, but like non data technical professional, their subject matter expertise was in another field.

And I was talking to them about this kind of like very concept of like, okay, why, what are [00:07:00] we really doing here and why are we doing it? And so I use the explanation, you know, of a business. And I think this ties back again to like that authoritative source of where are we getting this information? So for example, like I’ll, I’ll use a modified version of my own company.

So, you know, we do consulting work. We sell. hours of time or, you know, a product of strategy or something along those lines. Well, I have a CRM system, right, where I’m talking to all sorts of people. Hopefully, lots of them want to, you know, come on and be clients and buy stuff for me. So maybe I have, you know, contact X that works at company Y.

And they decide to buy. Well, now they go into a system that we use for actually doing the delivery of work. Am I tracking the hours? Am I keeping track of deliverables, progress? Maybe I want to staff different people to that account. Okay, great. And then eventually, you know, the work gets done. I have to go out and send an invoice to that person.

So [00:08:00] the information for company Y, is in HubSpot or Salesforce or Dynamics or whatever tool you use there. It’s in whatever system you use for doing your job. And then it’s in whatever system you use for invoicing. And that’s, that’s kind of just like a very simple example. Much larger companies have much more complicated stuff.

So being able to figure out, okay, well, company, why is this over here? And it’s this over here and it’s this over here. That is hard, kind of doing that join, that connecting motion, which this is my best definition of join right here, the hand motion.

Claus: And the reality is, you know, each of these three systems has a different identity key for company Y.

Ryan: Yep. Yep.

Claus: So how do you mix those?

Ryan: Yeah, exactly. And how do I get to a place where if I just want to talk about company Y and then I want to see all the information from all the systems, do I have that prepared? Like, is that something manually done? Where like, if I ask five analysts to [00:09:00] do five different types of customer reporting, are they all doing the exact same manual steps to generate that information?

Going back to your question on, how do we get the right copy of the number? So for example, let’s say that I, I want to talk about, okay, well, how much work did I do in January? Well, I know that the numbers of, Of what I did get synced back into my CRM. And obviously I have some invoicing stuff, but which, which one of those is right?

And which is right, depending on the question that I ask, right? Well, maybe I did some work that ended up not being billable. Well, then that won’t be. And to the other side, you know what I mean? So like, for me, it’s, it’s really about, and you touched on this. It’s about thinking back to the actual business process, right?

Your business is doing some function. And then we captured information about that thing happening using these data systems. And the more I think about the actual business [00:10:00] process, going back to your conversation earlier, it’s just kind of like, well, how far back, how close to the truth to the thing that actually happened can

Claus: But it also comes down to how crisp can you be on what you’re looking for. So, so taking your example,

Ryan: Yeah,

Claus: am I looking for the amount of hours worked or am I looking for the amount of hours built? If you, if you didn’t actually go through the trouble of thinking through sourcing and ingest, you may not even realize that those are two different numbers.

And actually, both numbers could be interesting. And sometimes comparing the two and saying what’s the incremental work that I did that I didn’t build could be interesting as well. You may even choose to not surface how much work you did in all cases in your CRM system. So you could have three numbers.

You could have the number of hours you told the client, the number of hours you actually worked internally, And the number of hours that you build, right? So it’s fascinating to think about it that way. And then if you continue the lifecycle, cause we’re not done, we just, now we [00:11:00] have it in raw form, um, take your example with, with the company.

Okay. So how do you know which field in those three different sources represent the company name? Because it might be called one thing over here, it might be called something else here and might come from the third system. So, the next thing you got to do is you have to, at least partially, it’s called conform, streamline, you know, sort of normalize, equalize, standardize the data so that we all understand that this field represents company name.

Even though it came from two different systems, each of them has company name labeled differently, but this is the company name. And once you have got the right data, you’ve got it in a place you can do something with it. You understand what the data means to the degree that you can. Now you can integrate it.

And most people want to jump right to integration. But if you don’t know the other three steps, integration is hard.

Ryan: Yeah. One of, so when I first started my career, [00:12:00] I worked in, in FP& A and I was doing kind of, you know, lots of analysis. And I think, especially under the finance umbrella, you know, I talked about this in a previous episode of the podcast. There’s like this, you know, heavy reliance on making sure that we know exactly how these numbers are being created.

And so I think. In certain places. And again, this was like 10 or 15 years ago. So like times were different, new tools have come out, but you know, like, Oh, well, you know, we can just go talk to the analyst that made the Excel file. So that’s why we have some comfort over the logic. Um, but what that really meant was that I was piecing everything together and doing manual steps.

So like a lot of that stuff would be like a VLOOKUP in Excel or something like that. And what I like to explain to people is like, you’re doing. Data integration and you’re doing ETL. The question is just like, do you want to do it once? And I’ve thought through all the possibilities like you just mentioned, or do you want to have each analyst [00:13:00] times each report redoing that same work?

So one of the things that I think is very interesting, right? Whether you’re at that end of the maturity cycle where, or curve where we’re talking about, okay, well. We’re having multiple analysts redoing the same work, piecing it together, possibly getting different answers all the way up to like, we have like a full amazing infrastructure that’s answered all these questions.

Right. If we have people all along that curve, what are some of the recommendations that you could provide for how to take that next step in terms of scaling? Like how should an organization think about the cost benefit analysis on taking one step? Right. Cause obviously getting to the end of that, it’s like, okay, amazing.

But like, if I’m. Over here, like, you know, so how do I decide what to invest on so that I can grow my maturity in a financially viable way?

Claus: I think there’s two ways of answering the question. Uh, I have my own preference as to which path you should take,

[00:14:00] but the one path which is Non platform path is you invest in the thing that give you the most bang for the buck in the here and now.

Ryan: Mm hmm.

Claus: And you live with the fact that what you’re doing now may not necessarily carry through to what you need to do next. A lot of organizations do that, and it’s not necessarily bad. You just have to be very conscious that that’s what you’re doing, right? So let’s take an example. Uh, if you just wanted to take the CRM data and the finance data and smashed it together to figure out whether your contract was profitable or not. That’s a reasonable question to ask if you’re a CFO. And if you don’t worry about 17 other questions you might want to ask in the future, what you’re going to do is using a spreadsheet model or, you know, a low bar data integration tool, whatever it is you like as your tool. So you’re going to grab the data from these two systems and you’re going to put it together, you’re going to answer the question [00:15:00] and you, and you’re going to have a quick answer to your question. That’s a viable path. Uh, the problem with it is that you end up recreating the. sourcing and ingest and normalization logic every time you answer a question. So it gives you a quick answer, but it tends to pile up in terms of cost over time, because every time you have another question, you have to go to sourcing, ingest and normalization again.

Ryan: Yeah.

Claus: The alternative path, and it’s not for the faint of heart, because it takes perseverance and time and some amount of investment, is the platform path, where you say, no, actually, I’m going to create a staging ground so that every time I need the answer to a question, I’m going to land that data in one place.

And I’m going to make sure that when I do the sourcing work, the ingestion work, and the normalization work, it leaves me with a reusable piece of a data asset. [00:16:00] That sits in my call to data fabric. And that becomes my single source of truth for everything. And all of a sudden you get the booze and it’s going to take longer.

I’m going to need people to agree and I can’t just do my thing. And there’s truth to that, but it won’t necessarily be a lot longer. As long as you have in place the right life cycle management and you don’t get stuck in analysis paralysis, because in many cases, the reason you don’t move fast is because you get stuck in analysis paralysis, there’s too many answers in the, too many different opinions in the room and nobody to settle on, no, it leads to look like this.

So I like the platform path better because you’re creating something that actually helps you next time, as opposed to doing the work all over again. The reality is. Without a platform path, 80 percent of the work you do on analytics is going to be data work every time. If you go to

platform path, Well, 90,

Ryan: Yeah.

Claus: but if you go to platform paths, that percentage will go down [00:17:00] over time. Uh, cause you’ll be able to reuse the sourcing integration and standardization work, just that. I’m not even talking about anything downstream in terms of what do in the analytical side, just standardizing. The sourcing, the ingest, the normalization, and maybe the data integration will take a significant slice out of the 90 to 80 percent that you use on data work.

Ryan: I think that’s one of the cleanest jobs of explaining what, what I kind of agree with you is like the right approach to that, um, that I’ve heard. Like, I think that was like, One minute. And he just, that was, that was a total dunk. That was a 360 windmill right there. Um, now I do actually think that there’s a door number three.

It’s not a good door number three. In fact, I spend a lot of time, um, possibly to my own financial detriment, talking people out of. What I think door number three is, which is the same thing you’re talking about with the platform, but instead of doing it in an incremental kind of business and return driven [00:18:00] fashion, you already know where I’m going.

They’re like, we’re going to build the data warehouse of the next generation, baby, it’s going to have everything. And I’m like, Oh no, don’t do it. And that, you know, like there are people, you know, or the devil on my shoulder is like, that’ll be a lot of money, man. You should, you should tell them that you should promise them the world.

And it’s like, no, no, no, no, no, no. Cause it, it just, it is so wrong for so many reasons. Like first off, like the requirements and the benefits change so incrementally that if you’re not attacking them as they come up, you end up just missing the mark. And it’s just a recipe for like, valueless. Cost creation for an organization.

The way that you described it is so kind of perfectly aligned. with what I recommend to clients in the way that I describe it. I like to think about a data warehouse and a lot of this data infrastructure as just being a storage location, right? Like, [00:19:00] think about it in your head. It’s just like, hey, we did this work because we needed to for good business reasons.

Instead of letting it languish and die on some analyst’s hard drive, we’re gonna just Put a little bit more effort into thinking about it and vetting it up against other sources. And maybe do we need to expand this? So, you know, Oh, well, if we just tack on a little bonus, you know, couple of hours, we, you know, quadruple the amount of value that we’ll get out of this.

Okay, cool. So we do a little bit of extra thinking, but then we store it. In a public location, and we put that public location with proper resources that can handle the data volume and all of those things, that’s really the best way to think about it. This work is happening in the organization no matter what.

If we’re asking questions, we’re doing this work. The point is just, are we doing it over and over again? And so you talked about that perfectly. I love that idea of just like letting the business use cases and the questions that you’re asking drive what gets focused on, [00:20:00] but then making sure that it’s vetted and it’s stored for future use.

Claus: Uh, and look, I always believed that make sure that you get early value, uh, and you know, you can use the word incremental, but, but you need to get early value out of your data work. And what’s interesting is. Some people might even prefer the raw version of the data. They don’t even want the clean, the pretty version that’s already been mapped, because they’re doing data science and explorative work.

So they may want to actually get, they may have a lot of value out of just the raw data, even without. Me having normalized it. So you just don’t know what people might find helpful. If you’re a researcher, raw data is fine. Uh, if you are a business operations analyst, probably not. You probably want a semi normalized and somewhat integrated data so you can create a report.

So it all depends who you are. Where it gets complicated is if you’re in a [00:21:00] brownfield scenario and you have a lot of stuff already. Cause what you come up against is not just, Hey, can we think about this incrementally. But people want to do the incremental work in the place they already have, because the place they already have has a lot of data already.

So the incremental work to add to that is smaller than the incremental work to start over in this new place here. The problem is, The places they usually have today, you know, most times they’ve got more than one. So if you continue down that path, you’ll never get to the one sort of single source of truth.

And the other problem is, most of these places that exist today, were not actually built to the conceptual model that you and I just discussed. So, So it’s fraught with peril to continue to just do the incremental work on your current state data repositories, even though it is tempting, right? So, so this mythical door number three or four, depending on how you count, right?

There is a door that says, I’ll do [00:22:00] incremental. So the Big Bang, yeah, boo, that’s a bad idea.

Ryan: Boo.

Claus: But, but, but continuing to do incremental work on a non platformized environment is actually bad as well. And that is a pitfall that I see people fall into uncomfortably often.

Ryan: Do I want a small amount of pain every day, or do I just want to like, you know, take my medicine and move on to a better future?

One of the things, you know, we kind of have a saying that I’m sure is all over, but you know, you hear a lot in, in data and report and analytics development is that people don’t know what they want until they see. Um, and obviously I think that, uh, You know, among many other reasons, but I think it’s one of the big reasons [00:23:00] why data and analytics have leaned so much further into this kind of agile methodology.

So you talked about generating early value. And I think from a, from a financial standpoint, that makes sense, right? Like, Hey, if I’m spending money, isn’t it great to see return and that buys us rope for the next block of spending. And, you know, there’s kind of this like eat what you kill funding model.

going forward. One of the other things that’s something I’m not entirely sure people who aren’t data professionals think about is an additional benefit of that early value. It allows us to temperature check, you know, so like, okay, we get the raw data. We know that we’re presumably going to be doing further investment.

And then I show that to you and you’re like, Oh, this is missing X or hmm, what about this? And so it creates this cycle of increasingly valuable requirements and questions and filters. So like the logic gets better [00:24:00] and the project stays on track, even if that track moves. So

Claus: of the out of the possible. It is the storytelling of the data we have and the data we might want.

Ryan: you said the magic word in data storytelling. So, I obviously think that everybody has a different take on this, right? Like we all like different stories and we tell them different ways. You know, there are, you know, 15 versions of the same movie nowadays. People like different ones, you know? So. When I hear somebody talk about that, especially somebody that, you know, I assess to be qualified to give like a really good answer like you, tell me a little bit about what storytelling means to you.

If there’s somebody out there who’s listening and they, you know, build data infrastructure or they build reports for a living, what can they learn from you as to how to use the data of their organization to tell a better story? And then hopefully, you know, Not just tell that story, but [00:25:00] have that story have impact so that it changes decisions and changes actions.

Claus: Let’s see, why don’t we start with the difference between telling stories and storytelling? Because they’re not the same, in my opinion. Telling stories is just that, telling stories. That’s great. I mean, it’s good for entertainment. It may or may not be useful. Storytelling, in my world, is telling stories with a purpose.

Ryan: Hmm.

Claus: So now you know why you’re telling stories and you tailor your story to promote your purpose. That’s important because that means there are multiple purposes, less problem sets where you can apply storytelling in the world of data. We just talked about one. You have to tell stories about why is a platform approach to a data fabric better than whatever it is you might have done that’s an alternative.

That’s one set of stories. This is about your platform. [00:26:00] There’s another set of stories, which is around what can I do with the data that I have, the out of the possible, which is basically talking about how you can change your organization’s last company for the better using the data that you have. So that’s the second problem set, which is about telling stories.

Uh, there’s a third, which is how do you life cycle manage all this data? And what does this dirty word governance look like if it’s not governance, but let’s call it stewardship instead. So that simple choice of wording, did I use the word governance or did I use words like life cycle management and stewardship is how you tell the story or what it means to manage.

The life cycle of your data. So that’s just three, right? If, if I started thinking about it, I’d probably come up with two or three more, but you have to be really clear on what are you storytelling? Am I trying to get the investment and buy in for a platform? [00:27:00] Am I trying to establish a stewardship model and I really need the cultural change and then buy in for that?

Or am I trying to push the envelope on what I can do on operational optimization where Or is it all about my products I’m selling in the market? And this is what I can do with high powered analytics and the data that I need for it to build differentiated products. Those are four very different sets of stories.

CMO,

Ryan: threads and what you talked about, it’s a lot about providing the right context so that a person has clarity. You know, if I put myself, you know, uh, this is a, probably a silly example, but, When I think about it, like I remember I was a star Wars fan as a kid.

So, you know, you grow up watching the original movies and it’s like, all right, well, like it’s pretty obvious who to side with. And then as they’ve like, kind of started coming, you know, like out with all these different movies and all these different stories, it’s like, Oh, like, [00:28:00] you know, like things change, right.

And you, you know, like who you align with. So. The reason I bring up that analogy is to say, you know, when we were talking about the platform solution, all you did was just explain the options and provide clarity around information so that if somebody could put themselves in that position, they would know what they would do.

So they can make the right decision. I think that’s a, that’s a phenomenal way of thinking about it. Like that, that word purpose that you’re using, like, okay, well, if the goal is to get someone to make a decision and give them all of the information so that they can make that decision.

Claus: Flywire,

Ryan: to some of the stuff that we’re talking about.

There are two concepts that are very, I’d say interrelated, but they have different connotations. One is, uh, MDM or Master Data Management that you mentioned earlier. And another is, uh, Lifecycle Management or [00:29:00] LCM or Stewardship or Governance or whatever, right? So we, we kind of have these two concepts.

Both of them relate to Controlling and having understanding and documentation and support around where does the data come from? How can we understand that if we have somebody new, you know, how can they look up and find new information? To me, it’s kind of a fairly It’s, it’s murky water, depending on who you’re talking to.

So some people, you know, you talk to a project manager and you’re like, yeah, no surprise. This went sideways. Nobody knows where the heck anything is. Right. But then you talk to like a hardcore data engineer and they’re like, all right, well, I mean, do you want me spending time? Taking screenshots and telling people how to find things or do you want me like building cool new stuff?

You know So like there’s a lot of different perspectives and you also had mentioned that at least traditional mdm You had some of your thoughts on what are your recommendations on the real value plays? Within those spaces [00:30:00] so that companies can get value and security and assurance But without doing a bunch of pointless work or spending money, like what’s the balance there?

Claus: I think it starts with being clear on at what stages of progression is data valuable to you and why. Uh, you do need to agree on what are your stages of, of existence of data. I got my, my favorite chain goes like this, right? It’s sourcing, ingest, you know, normalization, uh, integration, enrichment, and then delivery downstream that you can pick your own, That’s, have, but have a model in mind that says at what stages can data exist.

And then make a choice that says, which ones of them do you care about? I actually think in many cases that will resolve the seeming dichotomy. Cause if there is value for you [00:31:00] in raw data, after sourcing an ingest, by definition, you cannot adopt the paradigm that goes all the way to data warehouse before you deliver value, because you just said that there was value in the earlier states of the data.

And I think. There are many things to like in MDM tools and concepts, but it does make the assumption that I always want to go all the way to the end. Because it is relatively closely associated to the same thought model that was sitting behind the data warehouses. If that is not the only place you get value, you already know that what you probably need to apply is a more tolerant approach.

Life cycle management model. But I think it starts with understanding what are the stages of interest of your data as it goes through life and which of these stages have value to whom. How’s that for an answer? CMO,

Ryan: solid. I think that that is like a mental framework that [00:32:00] both gives people clarity but also allows them to make it bespoke to a particular project or to their organization. Right. And again, I personally like it because you’re tying it directly back to the value proposition of like, when is value real?

Realized from whatever we’re working on and then making your decisions as a backward progression, like reverse engineering, what are we going to do based on the value that we’re trying to achieve, right? I pretty much any decision, whether it’s, you know, in data or outside of data, if, if people are pretty relentlessly focused on that, then they can at least feel confident that they, they made it like a really fair shake at that decision.

In fact, I think that data is specifically designed to help us answer some of those questions, right? Like, okay, like here’s a goal. How do we directly connect that back to the value that we’re trying to get, um, and make the right decisions. Now, that was extremely helpful. And again, I think one of the, the, The better takes that I’ve heard in my career about how to make good decisions there, because it’s [00:33:00] both like clear and definite, but also flexible based on the situation.

If I can change gears again, one of the things that I want to get into is a little bit of some of the specifics. So thus far, we’ve been talking a lot about kind of data in general and processes and data engineering and all that stuff. And it’s, You’ve honestly crushed it. So it’s been super valuable. But I also think that one of the things that the listeners will find valuable is whether they work in healthcare or healthcare adjacent fields, or whether they’re just interested in, you know, seeing new problems or their problems mirrored in a different industry.

I think one of the things that’s very cool is talking a little bit about, you What is different between all of the guests, as opposed to what’s the same, which is data expertise. So talk to me a little bit about data in the healthcare industry. What are some of the unique challenges that you have there?

And then what are, you know, some of the things that you’ve been able to do to help, you know, achieve solutions.

Claus: [00:34:00] I think there are three unique challenges, at least, uh, one is there are no standards that cover all the data of interest in healthcare. I mean, period. They don’t exist. You, you can see. Fire, which is F H I R. That is sort of the only standard we have for clinical data right now, but it’s mostly to manage payments inside the healthcare ecosystem.

It doesn’t actually capture all the things that a physician might want to know about a patient. It’s not made for that. So, there is no real standard for what, what is the shape and structure of healthcare data. So, so that’s the one thing that’s peculiar and it’s been tried many times that we just don’t dare.

The other thing that’s peculiar is There’s a lot of legislation that says what you can and can’t share, uh, and what kind of consent. And, and it’s pretty brutal when you require consent, uh, people also able to take back consent. And now how do you untangle [00:35:00] data that you put into, you know, a data fabric somewhere because people said you can’t use it anymore.

That gets kind of complicated. So, so, so that’s the thing is, is a complex equation. And the third one is that the problems we want to solve are human problems. Okay.

Ryan: Hmm.

Claus: They’re amorphous in that what you really want is to help people live the healthiest life that they’re willing to live. I didn’t actually say the healthiest life they can, because some people don’t want to make the choices that give them the healthiest possible life.

But the healthiest life they’re willing to live. So think about that. The human motivation is, what am I willing to do to live a healthier life? There’s signal value in how do I prod people at the right point in time so that they will have the highest possible propensity to take healthier actions. There is the issue of data integration, which is there is no common key that allows me to track an individual from one hospital [00:36:00] to another hospital to another hospital and the three insurance plans I’ve had.

Every time you get a new insurance plan, you get a new member ID. Every time you get to a new hospital, you get a new account ID, a new patient ID, and none of it actually ever maps. So this whole amorphous lack of cohesion in the larger healthcare ecosystem means that data integration and this whole notion of smashing data together we talked about is uniquely complicated Hmm. ecosystem.

I think those three peculiarities make healthcare very difficult place to work on data. The price is phenomenal. If we could actually create a digital twin type representation of an individual’s journey throughout healthcare, there’s enough science behind this now that we know we can give a healthier life, deliver better clinical outcomes, etc.

If only we were able to put all the pieces together. And then the last complication is you can’t always trust the [00:37:00] data. And here’s an interesting data point for you. It tells you how little you can sometimes trust the data. So, let’s say that you have a patient, and this is in oncology, so cancer care, because I worked there for some number of years. What would you think is the percentage of patients that are on the inbound to one of the top tier academic medical centers, without naming names, Uh, what is the percentage where you’re going to change the diagnosis, not the treatment, the diagnosis? Everybody knows that the better hospitals have better treat, you know, higher grade treatments, but even more fundamentally, how often do you think we change the diagnosis?

Ryan: As a data informed professional, my official answer is, I don’t know the answer to that. I don’t have the data. Um, but I know that you want me to guess, so I’m happy to guess. I would guess probably, uh, for oncology, [00:38:00] more than half.

Claus: About half,

Ryan: Yeah.

Claus: which is actually a little bit scary because it’s not just that you can do a better treatment. You didn’t actually have the diagnosis right in the first place. So that’s what I mean when I say Everybody has great intent and everybody does the best they can, but some diseases are just so complex that if you’re not a very specialized expert in that exact disease, you may not get the diagnosis exactly right.

And that’s okay, because there’s a reason that you move between the tiers in the system. But if you’re thinking about the data, was the original diagnosis bad data? Yes or no, it wasn’t the correct diagnosis, but it’s important to know that that was the correct diagnosis. It’s also important to realize it wasn’t the right one and that the diagnosis that came later at a different hospital was the better diagnosis.

So all of those things make healthcare super complicated when it comes to [00:39:00] this.

Ryan: and if I can say in intelligence, I think allowed you to describe that in a, in a better manner than I would have, but I have experienced all of those problems, obviously having been a consultant for as long as I have now, I’ve, I’ve worked with a lot of these and, you know, obviously HIPAA is, you know, Not just, uh, problematic when it comes to like solving certain data problems, but it’s also like a little bit scary, right?

Like, you know, like, you know, like, you know, oh, I don’t, you know, can’t, you know, lots of times we’ll be working at, you know, you have to even build very specialized isolated infrastructure to even dig into some of this data to make sure that it’s not, you know, on my machine or that, you know, it gets, you know, leaked or anything like that.

There’s a lot of that stuff, but, And when I think when people talk about health care, they talk a lot about that type of problem. The thing that I found most exciting to talk about [00:40:00] in your explanation there was when you talked about the human element of things. And to me, some of the most interesting questions around health care are You know, I’m a, I’ve mentioned it on the podcast before, but you know, if any of the Freakonomics people are listeners here, first off, I’m super pumped, but also second off, I’m a huge fan of you guys.

So, uh, thanks for making the podcast. Definitely a huge part of me starting this one, but they had a talk, you know, about the healthcare implications of behavioral economics. So like these little nudges that you can give to people. And if you just word things differently, does it impact, you know, some of those You know, the answers to those questions are the most exciting about it.

And like, can we use data to actually make people healthier and make them live longer? Um, it’s just super, super exciting, but you taught, you know, I, you know, the hurdles that are in the way, as far as being able to even say, Oh, well this [00:41:00] person, it’s like, okay, well, you know, the insurance changed or it’s a different hospital or any of that stuff, um, really, really becomes a quagmire.

So I’m, I am, I’m Hopeful that through kind of new technology and new information and, you know, all of the stuff in that landscape, that that becomes a more solvable problem in the future. Cause it’s just, I think would be, you know, massive for humanity. So I, you know, every little bit that you’re doing there helps everybody.

So thank you for that work now. I mentioned in your introduction, you know, on, on the topic of, of healthcare that you recently unretired to come in and be the, uh, the new CTO of UT Austin Medical Center,

Claus: is the

Ryan: cool. About that from kind of like gets, you know, gets a tech nerd builder, very excited is to a certain degree.

I mean, obviously you’re integrating with, you know, other systems and partners that, you know, are a part of that, but you [00:42:00] have a little bit of a blank slate, so I think it would be really interesting to share with the listeners, you know, how are you going to use that freedom? How are you thinking about coming into a brand new place and kind of getting to start from grounds, you know, from ground level?

Claus: is an interesting question. And it’s one of the reasons I only take it because you don’t actually get that semi greenfield opportunity very often. very often. Look, it’s never 100 percent greenfield.

Ryan: Yeah,

Claus: There is some stuff, but it’s more greenfield than most things. I mean, Austin, 3 million people in, in great Austin area, no academic medical center today.

And it’s the largest metropolitan area in the U S that doesn’t have an academic medical center. So in 2030, that’s, that’s five years ahead. We know it in 2030. There will be a new hospital, there will be a research organization, and there will be a medical school. And those three will be put together as an academic medical center.

The only thing that exists today is the medical school. It’s been [00:43:00] existing only for about 10 years and it does its research on premise and its practice somewhere else. So if you think about that, you’re going to grow. The revenue by a factor of 100 over five years, uh, 50 to 100, uh, you’re going to grow the staff by a factor of four or five, uh, you’re going to expand into operating a hospital that didn’t exist before, uh, you’re going to do a whole lot more on the research side.

So how do you think about that? Um, I think you have to be thoughtful and careful because it’s very tempting to say I’m going to do everything different. For some things, you should. For other things, you shouldn’t. Let’s take some examples. Are you really going to build your own electronic medical record platform?

That is probably a really bad idea, right? Because that is not going to be where you differentiate. You might [00:44:00] hate all the ones that are out there, but guess what? It is an enormous amount of investment to build your own. Probably a bad choice. Again, really old, it’s our platform. You probably don’t want to do that, right?

Because the same reason this is something that you should buy, whether that is, you know, Workday or one of the others, whichever one you like, um, probably shouldn’t do that, but. There are some things you should do differently. So, does education need to look the same, or do we need to think through what does generative AI do for education?

Both in terms of the curriculum, what do I need to know about generative AI? But also what can I do that gives people a different learning experience? Must I have inside the hospital a nursing station on every floor? Or is it possible that using technology, maybe nurses can be combined hybrid present physically and virtually without having a nursing station where all the nurses sit on every single floor in the [00:45:00] hospital?

Uh, do I have to discharge people using rounds the way we’ve been doing for hundreds of years? Or is it possible to combine clinical decision, which is what the rounds really do, right? Uh, with some kind of predictive algorithms and manage the flow of patients that get admitted, uh, and discharged from the hospital better.

And the questions just keep piling up. But being clear on where is it appropriate to ask the question, do I have to do it? The way it’s been done the last 100 years versus where the answer is, yeah, you know what, you probably should not try to innovate this one because you can’t do everything. Now, even with five years, you can’t do everything completely differently.

And I gave you examples of both. So it’s a fascinating journey. It’s probably going to involve a lot of storytelling in terms of where different is appropriate and where it’s not. I think it’s a fascinating opportunity in a group of people [00:46:00] They’re pretty much all joined because of this semi Greenfield kind of opportunity to do something that’s different because we haven’t had that many.

Maybe there’s one, I’m not actually sure. Either none or one other academic medical center, which is more than just a hospital that’s been stood up post COVID, and I do think it’s different now.

Ryan: yeah. I mean, honestly, even just the examples that you were giving me were kind of wildly exciting, both from like a technological kind of me being a nerd standpoint, but also from the possibilities of like improvement. You know, patient care, um, you know, like anybody, you know, hopefully people, you know, I hate to say this, but hopefully people have as little experience with hospitals as, as possible.

But, you know, I think most people, um, you know, have some, you know, like whether it’s for themselves or for a loved one or something like that. And like all of those things, you know, really, really, I think, speak to, to points that. You know, people experience it. It’s, it’s extremely, extremely exciting. Now, one [00:47:00] thing that I will mention is I’ve actually been to Austin, I think more times than any other city, um, than a city that I’ve lived in.

Um, and in general, I am probably the least healthy version of myself in Austin, just because of the. Massive abundance of extremely high quality barbecue. I think my cholesterol

Claus: is good.

Ryan: every time I go there, I’m, I’m doing it, you know, two or three stops in a week long trip for barbecue. And I come back, you know, with the meat sweats, but, uh, I don’t know, maybe with a generative AI, you guys will be able to prompt me emails to I don’t know, but barbecue is hard to

That’s right. That’s right.

Claus: Well, I

Ryan: I don’t know if anything would keep me away from it. Now, one of the things that I think is most exciting is, especially after somebody has gotten to share a lot of their thoughts and their experience, giving the listeners an opportunity to get to know you the human a little bit better.

[00:48:00] So tell me a little bit about you. Like what do you like to do for fun or like to do outside of work? Like what are some of your kind of passions or hobbies or what do you like thinking about and doing?

Claus: talk about science fiction, so it’s like, that’s what I read and that’s all I read. When I’m off the clock, you know, I only read science fiction literature, so I must have gone through thousands of novels over the years. So that’s, that’s what I do when I just have 15 minutes quiet time. Um, I love traveling.

Uh, I’ve so far been to 37 states in the U. S. Uh, I moved here in 2008. I’ve actually been to more, but I’ve been on vacation to 37 states. So, so I still have 13 to go. Uh, just driving through doesn’t count. You have to actually stop and do something that feels like it’s vacation like. Otherwise, it does not get added.

I’ve been to, I don’t know, 40 countries or more than that. Uh, so travel is probably the one vice that my wife and I sort of share. Um, I’ve got A handful of kids, like, fine. That’s a lot. That keeps you busy as well. [00:49:00] Top

Ryan: Yeah. Okay. Um, so seeing as how I’m also a science fiction, fantasy, all that kind of, um, nerd, give me your top three of all time. If you’ve read that many, give me, give me your top three

Claus: 3! That’s hard. I think the best series I’ve read is actually relatively modern. Uh, so It’s, um, written by a gentleman called David Weaver. I’m not sure I’m pronouncing his last name correctly, but it’s W E B E R. And it’s sort of the sci fi version of the Hornblower series of The Sailing Age. But in this case, it’s a female lead character, a naval officer, and she goes through all kinds of stuff.

So there’s a whole series. It’s the Honor Harrington series. I love that. That’s a great series. Don’t read the last book because he started a story arc that I don’t think he’s going to finish. So read the first 13, however

Ryan: I’m an ending guy. So thank you for [00:50:00] that.

Claus: uh, so that’s one. Um, you want something classical? I don’t know. Um, how about the Foundation series?

Ryan: Really good. Yeah. By I think that’s a good choice. Yes, by Asimov. If you want something that’s a little, you know, more contained, you could read Ivor Hubot, but I think the Foundation series is really good. And then moving into a more useful genre, just to give you different perspectives on

I love it. I love it.

Claus: I like military sci fi, obviously.

That’s clear from my choices. Uh, I think Ender’s Game is a phenomenal book. Uh, it’s really good. Uh, and I actually think they did a good job with the movie. It’s a very difficult book to turn into a movie, but I think they did a pretty good job. I like Ender’s Game. I know it’s a used sci fi sort of genre, but I actually think it’s a really good book.

They’re

Ryan: I, I didn’t want to ask you unless I had thought about what my answers would be. Um, and Ender’s Game [00:51:00] was on my list as well. Um, you know, I read a huge, um, chunk of, of, you know, that. Kind of serious. Cause he did card, did a series of books around that. I also agree that they did a phenomenal job with the movie.

In fact, that’s what got me back. You know, I, I did most of my sci fi and fantasy reading when I was younger. And then I kind of moved into like, well, I’m a professional now. I need to read nonfiction and all this stuff. And I watched Ender’s game with my parents and then I went upstairs. To my room, I was staying with them for the weekend.

So I went back up into my room and I still had some of my books there. And I think I had like Speaker for the Dead or something laying around. And, you know, I sat down and I was like, Oh, I’ll crack this open. Like three hours later, it’s like the middle of the night. And I was like, Oh wow. I still love this.

So it got me back into it. And I reread that one. Um, Being an ending guy, um, it was tough because I just, I knew going into it, I wasn’t going to get like the crazy ending satisfaction that I [00:52:00] wanted, but the Dune series, um, I, I was a fan of, I know that that’s obviously a bit of a trite, um, selection to put in, in the top tier.

Um, but then another one

Claus: just dark, they’re good books, but they’re dark.

Ryan: Yes. Yeah. Yeah, they are. Yes, very much so. Um, and then the last one that I’ll give, uh, I’ll say the Martian, which was also made into a movie that was awesome, but I also like Andy Weir’s other books. Uh, like Hail Mary was awesome. Um, I don’t want to I’m blanking on the, uh, uh, Artemis.

Artemis was also awesome. Uh, so I was a huge fan of those. Anyways, I’m glad that we had a little bit of overlap there. That’s, that’s super cool. Now. In case anybody, you know, likes what you had to say or wants to like reach out and connect with you or talk about anything, what are the best ways to get in connection with you?

Claus: Probably LinkedIn. I’m pretty active on LinkedIn. I wouldn’t say I go through my messages every day, but certainly a couple of times a week. I actually do reply to [00:53:00] my messages. Assuming that you talk nicely to me, I will reply. So easy to find. My full name Klaus Torp Jensen. If you ever look me up on LinkedIn, you’ll be able to find me.

Ryan: I love that. Um, Klaus, first off, like your experience and knowledge and like the level of distillation that you were able to talk about really complicated concepts at, you know, I’d ask like a really complicated question. You’re like, Oh, that’s three things. But you know, it was so apparent.

Um, how knowledgeable and experienced you are. I really, really appreciate you for coming on and sharing that, uh, with me and with the audience and, and for making an awesome conversation. And also I want to thank you for allowing me to talk about science fiction.

Claus: It’s a good topic. Always a good topic.

Ryan: Yeah. So thank, so much for coming on. I also want to make sure that I thank the audience. If you learn something today, if you liked it, if you laughed, please make sure to go on, um, you know, and tell a friend or write a review, give us a like a subscribe, any of those things that [00:54:00] really help keep the show going.

The podcast engine, uh, moving it, it really does help. Um, Klaus, thank you again so much for coming on. And this has been another episode of the making better decisions podcast. Thank you for listening.

Outro: That’s a wrap for today’s episode of making better decisions for show notes and more visit, making better decisions dot live a special thank you to our sponsor canopy analytic canopy. Analytic is a boutique consultancy focused on business intelligence and data engineering. They help companies make better decisions using data for more information, visit canopy analytic.

com. There’s a better way. Let’s find it together and make better decisions. Thank you so much for listening. We’ll catch you next week.