Are We Trading Critical Thinking For Fast Answers

If your supply chain analytics feel shaky, clinicians are probably sensing it too and that’s where savings go to die. We sit down with Steve Kinsella, owner of Data Leverage Group, to get blunt about the healthcare supply chain data quality problem hiding in plain sight: messy ERP foundations, inconsistent item master data, and downstream systems that amplify every error. Steve explains why the real cost isn’t just bad reporting, it’s the loss of trust that keeps value analysis teams from even getting to the starting line on standardization and cost reduction.

From there, we tackle the temptation of “fast data” and why artificial intelligence does not equal artificial infallibility. AI tools can generate instant summaries and confident answers, but Steve walks through the real risks when those answers are built on dirty sources or weak prompts. In healthcare, decisions around products and clinical practice can’t be made on autopilot, so we talk about validation, responsible use, and the guardrails that keep AI helpful rather than harmful.

We also get practical about what actually moves decisions forward: a repeatable value analysis workflow that includes clinicians early, runs disciplined trials that don’t turn into black holes, and hands off cleanly to logistics for implementation. The through line is clear: strong process plus clean data protects critical thinking and speeds up the right outcomes.

Subscribe for more healthcare supply chain and value analysis insights, share this with a colleague who fights the item master every week, and leave a review so more teams can find the show. What’s the one data issue you’d fix first?

Transcript

Welcome And Baseball Banter

SPEAKER_00 0:03

Hello everyone and welcome back to Taking the Supply Chain Pulse. I'm Megan with St. Audge Company, and we're glad you're with us. Today we're joined by Steve Kinsella, the owner of Data Leverage Group, to discuss insights from his recent articles on AI in the supply chain and the critical role of data quality in driving better decisions and value analysis. Now let's hand it over to your host, Fred Prantz.

SPEAKER_01 0:28

Steve and I are former colleagues, former co-workers, and both uh big baseball fans. Steve is a seasoned ticket holder, the Boston Red Sox, and um I've been a couple games there. It's an amazing thing. Steve, we're glad to have you here. Thank you for joining us. Thanks for inviting me. I'm happy to join your podcast. It's an honor. And and just to make things clear, you are not related to Ray Kinsella from Field of Dreams, who also attended ball games at Fenway Park. Is that sure? Is that true?

SPEAKER_02 1:05

I am not related, although I get that a lot. A lot of people ask, and maybe I should just start saying that I am, I guess.

SPEAKER_01 1:13

Yeah, you know, I mean, uh James Earl Jones is dead, so you very few people can uh refute your statement anymore. True, true enough. True enough. Yeah, that's yeah.

Steve’s Path Into Value Analysis

SPEAKER_01 1:25

Well, Steve, tell us tell us about yourself and about your background, and also um how you got interested in value analysis.

SPEAKER_02 1:36

Sure. Uh I'll start with um I was born at St. Elizabeth's Hospital in Brighton. Uh no, maybe I won't go back that far, Fred, but uh my start really began in Boston as a materials director/slash central star processing manager at Harvard Medical School's outpatient clinic for uh oral surgery and in uh dentistry, way back when I learned pretty much the fundamentals of inventory management, working with clinicians on getting them their stuff. So that's kind of really where I started my career. And then from there um I uh uh eventually made my way to Becton Dickinson working for BD Healthcare Consulting Services, which was really what Becton Dickinson did at the time. This is probably late 1990s. They uh acquired Tom Hughes' Concepts and Health Care Consulting Group. And BD was at that time getting into supply chain consulting. So I I moved from Harvard Medical School to that BD slash concepts and health care consulting group for a few years. And that's really where I learned a lot about data analysis and working with value analysis and supply chain teams on implementing initiatives and the clip, you know, really learning about the clinical component, how that important that is to match up with supply chain and all that fun stuff. And fast forward a little bit into my career, left BD and started Data Leverage Group, which is where I sit today, data leverage group. And I started the company with a focus on two things, which was data cleansing, because I'm a big believer in uh having good data. Uh, I think that that's uh still to this day a challenge for supply chain, although it's certainly better than it was in the late 90s and early 2000s, but it still has quite a few challenges there. And then I also uh uh started a service line for value analysis, and I uh created an online web-based uh project management tool for supply chain and value analysis called VAMS. And so that's what I've been doing uh ever since and working with value analysis teams and policies and procedures and workflow. Primarily, my focus is on the workflow piece, where we help to bring together the value analysis teams and help them organize or reorganize their workflow to really help them compress time and really take their value analysis process to the next level. And we do all that through our VAMS program. And that's where I'm at today.

SPEAKER_01 4:51

Well, very

Writing On LinkedIn And Attention Spans

SPEAKER_01 4:52

good. Um, one thing I noticed recently, and I I forgive me for not having noticed it sooner. Maybe you've been doing it longer than uh than I noticed, but you started publishing articles regularly on LinkedIn. Okay, and uh I I like that because uh good content of the articles, we're gonna discuss a couple of them. Uh number two, you're contributing, you're contributing to the discipline. Uh people can people can read and learn and uh sort of uh get perspectives that they may not get otherwise. How how long have you been doing that and how well has that worked for you? Yeah, that's a great question.

SPEAKER_02 5:36

So really recently, I would say since mid-2000, 2020, probably off and on, writing articles here and there. Uh with the encouragement of folks along the way, folks suggested that I write some articles because felt that you know I could contribute and had some interesting perspectives. So I did start doing that back in 2025. Uh right around, yeah, probably July-ish 2025.

SPEAKER_01 6:13

Uh, you know what's interesting about that is I mean, these are these are nice sized articles. And I have uh I have a chip on my shoulder the size of a small building about uh the way Americans uh learn now, or the way Americans don't pay attention, or the way uh we can't focus. Uh I've been we used to I used to contribute to a newsletter uh that would be four or five pages long, and it kept getting shorter and shorter, and it went away because Americans can't pay attention longer than half of one page. Uh and yet so they they don't get details, they don't learn what uh what they could be learning. And for the stuff that you're publishing on LinkedIn, um there's an opportunity there for folks to actually get some meat and some uh some content and learn something uh about a process. And those people that read it, I think you're providing a great service. And I wanted to I wanted to talk about two of your articles today.

The Real Cost Of Dirty Data

SPEAKER_01 7:15

The one has a title that I really want you to get into some detail. The title is this the industry has a supply chain data quality challenge, and it's costing you millions. Take off on that for a minute.

SPEAKER_02 7:33

Sure. So what's been on my mind lately has been the whole idea uh around data cleansing and having good data to do data analysis at its core. And so I wrote that article with a few things in mind that um, and as I mentioned a few minutes ago in my intro, data quality is certainly getting better over time, but it's still not where it needs to be. I continually hear from customers that their ERP system data is still sketchy. Let's say, let's not get too deep into the weeds here, but uh the item file, item master, still not where it needs to be. Those peripheral systems that it supports uh for power levels and all the different systems around that kind of feed off of that. And those systems are still messy, let's say. So I I wanted to write the article and really kind of highlight how important data quality is, especially today, and try to put it into today's context because you know everybody's pressed to save money and you know, uh do cost reduction and keep their dollars down. So I wanted to kind of highlight the urgency uh and the importance of having clean data as it relates to keeping your expenses down, because that dirty data really ultimately, if you don't have the clean data, you're gonna lose trust. And that's really the crux. If you read that article throughout, that it's really about trust in the data. And if the trust in the data isn't there because of the data is not the data quality is bad, then you really can't get to any of those real cost savings initiatives with the clinicians because they simply don't trust you in supply chain or value analysis because they just simply don't trust the data. So you really you you're you don't even have a starting point uh to start with because the trust is simply not there.

SPEAKER_01 9:55

Yep.

Fast AI Answers Need Validation

SPEAKER_01 9:56

Um talk to me about the dangerous illusion of fast data.

SPEAKER_02 10:05

Yeah, so fast data, and uh that is really kind of an AI thing, right? Uh it there's a real direct relationship to AI, right? Uh we we type something into Chat GPT or um uh co-pilot, and we blink our eyes and we get a three-page synopsis of everything to know about that particular um uh that that particular topic. And that is certainly uh a great thing. AI is certainly a great thing. I think we all you know have our opinions about it, but by and large, everybody's using it and takes advantage of it and gets value out of AI. But I think um when we're on the healthcare side, um, you know, we're we're we're not to juxtapose where we're at in health care. We're in healthcare, and that implies that there's that clinical component, we have to get things right for patient care. So it's you really can't make mistakes uh in wrong assumptions uh when you get into the uh clinical aspect of value analysis and supply chain, because that can ultimately harm the patient, right? So to juxtapose what we're doing in healthcare with another industry, you know, we're not in the entertainment industry, and we can just type our topics into AI and get some answers. If AI is wrong, there's no patient on this on this uh operating table that's going to be harmed by that answer. However, in our case, that certainly could be the case now, certainly extreme to suggest that a patient could be harmed by AI. And I don't want to suggest that necessarily to be the case, but it can certainly lead to bad uh decisions being made prior to either taking a new product and entering it into the system for use. In other words, approving a product or product line for use because AI kind of led us possibly down the wrong path. So that's really the watch out there is um AI is not number one, always correct in its assumptions, and we need to make sure that we validate those assumptions, especially in healthcare where there's clinical implications.

SPEAKER_01 12:50

Yeah, and uh the other article, I might I might as well merge these two articles together here because we're they're really uh they really go well together. Artificial intelligence does not equal artificial infallibility. And I'll give you a story, just a simple thing that happened. I have these guys that come over every morning and we talk about all kinds of things, mostly sports. And one of the stuff we're talking about was um what's the average salary in Major League Baseball? Turns out uh my one buddy's got his phone, I got my phone, and he asked his, he asked uh one, he asked uh evidently Siri doesn't have the same answers in South Euclid as it does in Cleveland Heights, where I live. But uh but I had asked, what is the major league minimum baseball salary? And the answer came out uh something like $575,000. He asked what the Major League average salary was, and it came out $400 and some thousand dollars. Now, how can the average be lower than the minimum? So it gets to the point where these are both AI generated answers. It gets to the input into the database, uh, what information is there, the quality of that data, this is what you talked about, uh, and and even sometimes the viewpoint of uh the viewpoint of the cumulative data in the database. Is that fair to is that fair to say?

SPEAKER_02 14:20

Yeah, uh that that's an excellent synopsis of AI and the watch outs with AI. How you prompt your question can give it a different answer each time. Even uh a word or two rephrasing by just uh how you word it, your prompt can give a vastly different answer in some cases, right? And uh it's the data sources and who who's to say uh that AI, if you if you don't check the AI sources, that the source that AI used was dirty data itself, right? Where you're not prompting AI to say, before you give me an answer, please cleanse all the data in every data source that you're asking, and then give me the answer, right? So it's uh certainly a tricky thing.

SPEAKER_01 15:27

So part of this, one of those uh subtopics in here is this where AI fits and where it doesn't. Talk about that.

SPEAKER_02 15:40

Sure. I guess the case could be made it could fit anywhere, but I think the point I was trying to make there is that uh you really want to pick and choose. So back to the whole clinical AI question, it can certainly fit asking clinical questions, but uh up to what point, right? Um in on the supply chain side and value analysis side, do are we the right people to be prompting questions about procedures and surgical outcomes and things like that. Now, in some cases, maybe um the deeper we're into an initiative, the we are that is for us, but probably not. So it's uh where it fits, where it fits for us might stop at a certain point in supply chain and where it goes from there, uh AI for surgeons and procedures and outcomes and things like that, it it may fit a little bit differently in context for the surgeons and the clinicians.

Sponsor Message From CNOC Company

SPEAKER_00 17:00

Before we dive back in, you may be asking yourself, is my supply chain future ready? At CNOC Company, we don't just optimize logistics. We engineer smarter, leaner, and more resilient systems from end to end, from manufacturing to retail to healthcare. Our experts bring decades of experience in cutting-edge strategies to transform your operations. Visit and discover how we're shaping the future of supply chain one solution at a time.

SPEAKER_01 17:26

Um, yeah. The uh this is uh another thing that I think you're this is the next uh part of the your article.

The Quiet Erosion Of Critical Thinking

SPEAKER_01 17:35

The real risk is the quiet erosion of critical thinking. Talk about that.

SPEAKER_02 17:44

Sure. So we we sort of you sort of uh talked about it earlier uh in the lead up here. Um the we live in a society where uh we just want the answer, you know. There's that saying, don't don't tell me about the labor, just show me the baby, right? I forget who that famous uh coach, basketball coach was, but it's that that's the first thing that comes to mind, right? Because we're so used to just give me the answers, not not how we arrived there. We we get a some summary of everything these days. We never really get the full depth of the discussion, we just get a AI, you know, AI summary, or in a lot of cases, we're kind of spoon used to getting spoon-fed uh something closer to the answer with a summary versus the in-depth knowledge behind that. So we really there's not uh book book sales, uh I can't imagine, are going through the roof these days compared to a few years ago. People just want the the summary of what happened, or they're just gonna watch it as a movie on Netflix, right? So there's that erosion of of thought and and discernment in uh in uh our thought process because we're just getting summaries of information these days versus the in-depth knowledge. So kind of repeating myself there, but that's really was it is my concern that we're really not getting depth of information.

Clinician Inclusion Builds Trust

SPEAKER_01 19:41

So so let me sort of revisit where you started from and where I started. I started before you. I started when uh there was I start I also started out, I don't know if you know this, as a director of sterile processing and distribution at Baptist Hospital. So we have in Miami, so we have uh we have a common starting point. And that was in the days of what I would call Jeopardy purchasing, where the doctors gave you the answer and uh you had to get what they said, okay. Uh you knew you knew you were going to buy a Kleenex, not a facial tissue. Those are two different things. Kleenex is a brand name, it's only one facial tissue you can write requirements for, and then you can go through a process. Um and then along came the uh the product standardization committee, which is different than the new products, and then we're moving toward value analysis where you're implementing a process and you're implementing a critically uh followed procedure to come to conclusions about what you need. From your experience, how have physicians themselves acclimated to a process that takes longer but makes better uh more appropriate decisions?

SPEAKER_02 21:01

Well, I I would say I would answer the question this way from my early career to to till today, I I don't think it's necessarily changed a lot. And what I mean by that is early in my career I I had a a moment of learning, if you will, where uh a surgeon, uh oral surgeon was really pretty upset with me because I changed out a seemingly to me not not a critical supply product. And the the interaction was was this why did you change it out? Why did you not consult with other surgeons or other people? You just kind of switched it out. And it's uh purpose of having a discussion with with me was just include us in the process and we're gonna help you with the process. But if you don't include us in the process, then that's a problem. That was the gist of the early learning and interaction I had with this particular surgeon. And I carry that through to today, and I consistently see that today. As long as they're included in the process, number one, they're fine. They will be okay with you. I don't think that has changed in my career. What we're really seeing is we're trying to with the process, we're we're trying to, we have more information at our fingertips and we're sharing, and it goes back to the data quality. The big thing that I've seen change is the surgeons and clinicians, they really challenge our data in supply chain because they know more often than not it's incorrect. Not all the time. There are some organizations I've worked with that have stellar data. That so I don't want this to be a widespread commentary on healthcare supply chain data, because that's certainly not the case. But the majority of data is not where it needs to be. And surgeons and clinicians, they know that. So when they're presented with supply chain data, the trust level is not where it needs to be.

SPEAKER_01 23:48

Yeah, and that that that's interesting. And I think that uh, you know, you're saying the inclusion of the physicians. I mean, once again, I get back to where it started out where it was a dictatorship, and we were we were angry about the fact that we got dictated to, finally, it transitioned to a point where we would have physicians involved in the process, and then we'd have physicians who became advocates for the process. And so uh if if I if I could, and check me if I'm wrong, if you look at a value analysis continuum, uh you start out with asking quality questions to determine requirements. Uh, number two, having good information and data. Number three, from that, developing options, reviewing those options and selecting the outcome, and then continuing to monitor afterwards. Is that a fair process of uh of the way to include uh AI and other elements into a data anal, I mean a value analysis continuum?

A Repeatable Workflow That Compresses Time

SPEAKER_02 24:51

Yeah, that's a great way to frame it out. I uh I would say uh I might add to that to have a uh standard uh workflow that encompasses all of that from on the identification side of the new newly requested surgical item or whatever that item may be, to have uh a rigorous due diligence process that includes clinical data, that includes supply chain data and information at a high quality level, and always being inclusive, obviously, of the clinicians and uh nurses, etc., on the front end, and making sure that you have that process in place that's repeatable so that people can actually once again gain trust in it. And then I I see that continuum really being three pieces. That first piece being the due diligence pre-implementation process. In the middle, you always kind of have that, okay, let's trial this because we need the the product evaluation information. It it's great that the literature tells us it's the greatest thing since canned beer, but we need to trial it. So you kind of have that middle process of trialing, and that needs to be rigorous with its steps, and it can, and if it turns into a black hole, then you're in in real trouble in value analysis. So you have to have that nailed down as well. And then from there, after the trial, there's a decision that's made to implement, and you need to have that process in place where value analysis can smoothly hand off all the information to logistics for a smooth implementation. And so if you have those, you can compress time, and everything that you talked about kind of fits right nicely in the middle of those three almost Lego pieces or pieces of a puzzle of a workflow puzzle that fit together.

Process And Data Trust Soapbox

SPEAKER_01 27:09

Yeah, well, uh thanks, Steve. Uh, you know, I've asked all the questions I have. I don't I don't know that much about anything anyway. So uh uh what did what would if I gave you a soapbox, uh what would you like to say there talk about that I haven't asked you about?

SPEAKER_02 27:28

This has been a great discussion. Uh not a heck of a lot. I might repeat myself here, but I would say that you know, to really uh lay down the train tracks for successful implementations of products, whether it's new product requests, item ads, product conversions, cost reduction initiatives, you really need to bolster the trust in your both your process and your data. And so that's really kind of where I no surprise there, that's kind of where I focus with data leverage group. We focus on process and workflow and making sure that uh there is trust in the process and there's trust in the data. So that's pretty much my soapbox uh speech for the day, Fred.

SPEAKER_01 28:29

Great. Well, folks, uh thanks so much, Steve. And if you want to learn more about value analysis and about uh Steve, just uh uh follow him on LinkedIn. I I uh read his articles when they come out, and and I'm I think that I think that I I really appreciate what you're doing with those articles because I think that giving back to the discipline is one of the responsibilities we all have, and I admire the folks like you that do that, Steve. So thanks so much for being on our podcast and I wish you the best. Thanks, Fred. I appreciate it. Okay, take care.

Subscribe And How To Reach Us

SPEAKER_00 29:09

Well, that's all for today. Thanks so much for joining us, and don't forget to hit that subscribe button and connect with us online so you'll never miss an episode and can catch up on all the ones you might have missed. Got a topic you're fired up about, or maybe you want to be a guest on the show? Fred would love to hear from you. Just reach out at F C R A N S at S T O N G E.com. We'll see you next time.