Unlocking the Secrets of Data Governance with Carly Van Zandt - 201

WEBVTT

00:00:00.017 --> 00:00:04.417
Hey there. Welcome, and for those returning, welcome back to How I Met Your Data.

00:00:04.717 --> 00:00:07.957
I'm Sandy Estrada, here with my co-host, Anjali Bansel.

00:00:08.197 --> 00:00:11.797
We're your guides through the ever-evolving world of data where the community

00:00:11.797 --> 00:00:13.937
is as diverse as the stories we share.

00:00:14.117 --> 00:00:17.257
From hands-on practitioners to the ecosystem supporting them,

00:00:17.457 --> 00:00:21.117
we bring you insights and conversations that cover everything data-related.

00:00:21.397 --> 00:00:24.477
Whether you're here for the latest strategies, a dose of inspiration,

00:00:24.817 --> 00:00:28.777
or just some data banter, you're in the right place. So, grab a coffee,

00:00:28.997 --> 00:00:30.677
get comfy, and let's dive in.

00:00:30.480 --> 00:00:48.240
Music.

00:00:48.177 --> 00:00:52.417
Hi, Angeli here. Welcome to Season 2 of How I Met Your Data.

00:00:52.417 --> 00:00:56.737
After taking a much-needed break, navigating some transitions,

00:00:57.097 --> 00:01:01.817
stay tuned for more on that in a future episode, and gearing up for this exciting

00:01:01.817 --> 00:01:04.577
season, we're back and ready to dive in.

00:01:04.797 --> 00:01:09.657
To kick things off, we're thrilled to have Carly Van Zandt, Senior Director

00:01:09.657 --> 00:01:13.357
of Data Governance at Fresenius Medical Care, as our guest.

00:01:13.777 --> 00:01:17.777
Carly brings a fresh and practical approach to data Governance,

00:01:17.957 --> 00:01:22.097
including her framework for demonstrating the ROI of governance efforts,

00:01:22.357 --> 00:01:26.337
a topic we know many of you are curious about. Let's jump in.

00:01:28.757 --> 00:01:33.937
Well, welcome, Carly, to our podcast. I'm so happy I was able to have you on.

00:01:34.117 --> 00:01:40.537
We met at CDAO in Boston, and I ran up to you after your presentation,

00:01:40.537 --> 00:01:45.337
and I said, Carly, I am so impressed. I need to have you on my podcast.

00:01:46.197 --> 00:01:49.177
Thank you thank you so carly

00:01:49.177 --> 00:01:52.017
today we wanted to chat with you and the reason why i wanted you

00:01:52.017 --> 00:01:54.837
on the podcast was to discuss the incredible work you've

00:01:54.837 --> 00:02:00.797
been doing for senius related to data governance and more importantly i think

00:02:00.797 --> 00:02:05.277
is you did something very unique in the data governance space i think you did

00:02:05.277 --> 00:02:11.937
two really unique things one you created easy to understand roi on the work

00:02:11.937 --> 00:02:13.417
of governance or bad data,

00:02:13.677 --> 00:02:17.057
like why, giving everybody the why of the initiative.

00:02:17.457 --> 00:02:21.977
And then two, when you talked about the size of your team, I think everybody

00:02:21.977 --> 00:02:25.157
in that room gasped because I think you said two.

00:02:25.437 --> 00:02:27.957
Four, including me. Four, including me. Yeah.

00:02:28.577 --> 00:02:35.057
And I remember hearing an auditory gasp thinking to myself, yes, that is incredible.

00:02:35.577 --> 00:02:39.777
That is a feat not many have been able to conquer.

00:02:40.037 --> 00:02:43.717
So So that's why I ran up to you right after that and said, we need to have

00:02:43.717 --> 00:02:44.817
you on. I need to hear more.

00:02:45.357 --> 00:02:48.917
Thank you for being so gracious and for joining us today. You're welcome.

00:02:49.097 --> 00:02:52.157
I'm happy to be here. I love talking about this stuff, as you can tell.

00:02:52.417 --> 00:02:57.537
And the more we get the word out, the more people will do the same thing over and over. So.

00:02:58.112 --> 00:03:04.092
That's the dream. So maybe we could start from the beginning because my understanding

00:03:04.092 --> 00:03:08.972
was you did not start off focused on governance. You had a different focus.

00:03:09.352 --> 00:03:13.752
So maybe you can walk us through that. Kind of how did you end up with this

00:03:13.752 --> 00:03:15.352
role? Yeah, absolutely.

00:03:15.732 --> 00:03:19.092
I actually, in my career itself, I started as a biologist.

00:03:19.252 --> 00:03:24.032
So I actually started in biology, found out that I was terrible at memorization.

00:03:24.452 --> 00:03:29.372
So I switched to computer science. Well, not switched. I got my bachelor's.

00:03:29.412 --> 00:03:31.092
I was too close. I finished in biology.

00:03:31.432 --> 00:03:34.912
And then I got my master's in computer science because it actually made more

00:03:34.912 --> 00:03:36.252
sense to me with algorithms.

00:03:36.592 --> 00:03:40.592
So biology has always been kind of close to my heart and medicine in general.

00:03:40.952 --> 00:03:45.392
So then when I joined Fresenius, I started in clinical research.

00:03:45.392 --> 00:03:49.052
So I was in the clinical research arm called Fresnova Renal Research.

00:03:49.492 --> 00:03:56.492
And so while I was doing research, I realized how the data quality directly

00:03:56.492 --> 00:03:58.512
impacts our research outcomes.

00:03:58.892 --> 00:04:04.832
And some of the stuff that I was working on in research was enhancing the research

00:04:04.832 --> 00:04:09.292
data sets with data from our clinical systems. because in Fresenius,

00:04:09.492 --> 00:04:13.312
we have, you know, we have 200,000 active patients at any point.

00:04:13.732 --> 00:04:18.432
So you'd think you'd have really accurate, good data, but then we were finding,

00:04:18.552 --> 00:04:21.552
you know, different data entry patterns, different misinterpretations,

00:04:21.552 --> 00:04:23.772
you know, you can't really make assumptions.

00:04:24.032 --> 00:04:28.792
And if someone is going to be using this data to submit to the FDA for a new

00:04:28.792 --> 00:04:32.592
drug, you have to defend exactly what's happening.

00:04:32.712 --> 00:04:36.872
And so that actually brought up, you know, I was always questioning things.

00:04:36.912 --> 00:04:39.992
I was the annoying one saying like, well, how come this doesn't look like this?

00:04:40.072 --> 00:04:43.452
And this doesn't look like this. And so it was actually kind of funny because

00:04:43.452 --> 00:04:47.352
one of our heads of data analytics, her name is Norma Ofston,

00:04:47.352 --> 00:04:49.592
who was what we call the data queen.

00:04:49.932 --> 00:04:54.072
Because I was asking her all these questions, she was like, you would be perfect

00:04:54.072 --> 00:04:59.992
for data governance because you actually want to know how to find this stuff out.

00:05:00.112 --> 00:05:04.772
So she stole me into her team of data governance. and I've been there for seven years.

00:05:05.072 --> 00:05:08.892
And then when she retired about two years ago, I kind of, I took over for,

00:05:08.892 --> 00:05:11.032
for that whole, for the whole group.

00:05:11.412 --> 00:05:14.892
And it's been a journey, obviously, ever since then.

00:05:15.052 --> 00:05:20.852
But I think it was really important to see the problems in the clinical data

00:05:20.852 --> 00:05:25.232
that I was looking at in order to understand what needed to be done.

00:05:25.752 --> 00:05:31.772
Wow. That does sound like a journey. So you've been part of this prior to ending

00:05:31.772 --> 00:05:35.112
up owning the team, you were part of the team for seven years.

00:05:36.201 --> 00:05:40.181
Did I hear that correctly? Yes. I think this is my eighth year. No.

00:05:40.421 --> 00:05:43.681
It's super curious how that journey has looked for you over seven.

00:05:43.761 --> 00:05:48.421
Because if you think about data and analytics and just how data has been managed

00:05:48.421 --> 00:05:51.501
in organizations, especially an organization like Fresenius, right?

00:05:51.921 --> 00:05:57.861
Over that time span, the maturity of how people manage that data has changed. Oh, yeah.

00:05:58.221 --> 00:06:01.601
Have there been themes over those seven years?

00:06:01.621 --> 00:06:06.601
Absolutely. So I think like every single data governance group or a group that's

00:06:06.601 --> 00:06:10.241
working on data stewardship or anyone that's trying to help with data,

00:06:10.521 --> 00:06:14.541
you will always have, or hopefully, you'll have some false starts.

00:06:14.821 --> 00:06:19.901
We had a couple of false starts. And one of the things that we started to try

00:06:19.901 --> 00:06:25.261
and do was take all the categorization and chunking of data domains ourselves.

00:06:25.361 --> 00:06:30.681
We tried to do that. And then we finally had to realize that we can't do that.

00:06:30.681 --> 00:06:37.501
It has to be our clinical services, our boots on the ground to explain how they chunk out their areas.

00:06:37.701 --> 00:06:43.401
So we probably false started maybe two times because we were trying to split

00:06:43.401 --> 00:06:49.381
things in a way that made sense to us, but didn't make sense to the business

00:06:49.381 --> 00:06:51.541
folks that we then started to work with.

00:06:51.641 --> 00:06:57.501
So we started to create this beautiful, you know, medication authorization workflow

00:06:57.501 --> 00:07:02.761
that That we thought was going to solve, you know, the world's problems basically at our company.

00:07:03.301 --> 00:07:07.381
And it got so complex and confusing to folks that it never went anywhere.

00:07:07.541 --> 00:07:10.561
So we spent, you know, probably months and months on that.

00:07:10.981 --> 00:07:14.921
Additionally, we got a tool, which I'm not, I'm not upset that we got a tool,

00:07:15.021 --> 00:07:18.161
but we got a tool before we got the program up and running.

00:07:18.361 --> 00:07:22.481
So a lot of companies make that mistake as well, because you think a tool is

00:07:22.481 --> 00:07:27.281
going to solve all your problems when in fact it's just really the enablement of people to help.

00:07:27.281 --> 00:07:33.541
And we also finally had to accept the fact that people are already doing data

00:07:33.541 --> 00:07:36.741
governance in their own ways, whether it be spreadsheets,

00:07:37.021 --> 00:07:42.101
whether it be emails, whether it be writing in a notepad somewhere and acknowledging

00:07:42.101 --> 00:07:46.201
that and saying, let us help you make it a little bit better.

00:07:46.932 --> 00:07:51.812
We got a lot more traction than saying, you're not doing anything. The data is crap.

00:07:51.972 --> 00:07:55.592
Let's start all over because that was very overwhelming for people.

00:07:56.012 --> 00:08:00.932
So the seven years has had its hiccups. I want to say we really gained more

00:08:00.932 --> 00:08:05.352
momentum the last two years or so. So we started bottom up.

00:08:05.712 --> 00:08:11.552
Technically, we started top down with one upper management person, which was not enough.

00:08:12.052 --> 00:08:17.092
One, one in a company of 100,000 employees was not enough because they had a

00:08:17.092 --> 00:08:19.552
different purpose than all the other management.

00:08:19.772 --> 00:08:24.052
So then we started from bottom up. Let's do technical. Let's plug in the tool.

00:08:24.432 --> 00:08:27.912
Let's start with our technical lineage. Look how cool it is.

00:08:28.032 --> 00:08:30.972
It's cool to us. A clinician doesn't care at all.

00:08:32.112 --> 00:08:35.992
They don't care at all. So then we got stuck there.

00:08:36.192 --> 00:08:40.292
And so then we went middle up, which was somewhat successful.

00:08:40.532 --> 00:08:42.912
When I say middle, I mean the analytics teams.

00:08:43.292 --> 00:08:47.472
So analytics teams are in between business and purely technical.

00:08:48.012 --> 00:08:51.732
And so we got some traction with them. And when I say traction,

00:08:51.912 --> 00:08:54.252
I mean, we made the connections to the business.

00:08:54.452 --> 00:08:59.272
So it was a lot about interpersonal connections and explaining and having our

00:08:59.272 --> 00:09:01.572
analysts vouch for us saying they know what they're doing.

00:09:02.263 --> 00:09:06.263
You need to have a conversation. And then we got the green light to really kick

00:09:06.263 --> 00:09:08.923
off a data governance framework,

00:09:08.923 --> 00:09:18.123
which is where we landed with getting buy-in from the business and all the upper management groups,

00:09:18.323 --> 00:09:21.183
not just one. Not just one. Yeah.

00:09:21.523 --> 00:09:26.543
Across the teams. I guess one question around that, where did you start with

00:09:26.543 --> 00:09:29.443
that framework where you started to get traction on buy-in?

00:09:29.523 --> 00:09:32.283
Like what was the thing that finally turned the needle for the team?

00:09:32.423 --> 00:09:35.383
For us, it was business unit focused.

00:09:36.003 --> 00:09:40.663
So at Fresenius and many other groups, you have groups of people that work on

00:09:40.663 --> 00:09:44.783
different things. Our company is very large in terms of our business units.

00:09:44.903 --> 00:09:49.163
Like we have a pharmacy, we have our clinical dialysis services,

00:09:49.163 --> 00:09:52.923
we have our vascular services, we have a renal pharmaceutical service.

00:09:53.343 --> 00:09:56.263
So business line was where we got the most traction.

00:09:56.483 --> 00:10:01.723
Because if we were talking about, you know, patient admissions to someone who

00:10:01.723 --> 00:10:06.443
works in a pharmacy, that didn't really speak to them.

00:10:06.523 --> 00:10:12.743
So we would have presentations about one area and everyone else was glassy-eyed

00:10:12.743 --> 00:10:13.723
and wasn't paying attention.

00:10:14.403 --> 00:10:19.623
Then when we started realizing that we have to speak to the different business

00:10:19.623 --> 00:10:24.683
units in their own language, we would give examples of pharmacy bad data.

00:10:24.783 --> 00:10:29.583
We would give examples of clinical services bad data. We give examples of vascular bad data.

00:10:29.763 --> 00:10:35.763
So speaking to each business unit then as a business unit really helped put

00:10:35.763 --> 00:10:37.063
it into their perspective.

00:10:37.423 --> 00:10:42.383
When you say you give examples, were you just kind of assuming you gave examples

00:10:42.383 --> 00:10:47.663
and the impact of those examples in terms of, hey, we saw this and here's kind

00:10:47.663 --> 00:10:51.683
the outcomes of why this is not helpful, right?

00:10:52.503 --> 00:10:58.283
But it was different types of impact, too, because we're speaking to different types of people.

00:10:58.543 --> 00:11:02.243
So physicians, we explained the clinical impact.

00:11:02.783 --> 00:11:06.363
This patient's not going to get the care they need. That speaks to a physician.

00:11:06.703 --> 00:11:10.803
When we're talking to the business financial folks, that's when we say,

00:11:10.943 --> 00:11:15.183
this is how much money you lost because you had this bad piece of data and it

00:11:15.183 --> 00:11:16.203
caused all these problems.

00:11:17.023 --> 00:11:21.043
So, and then when we talked to legal, we say, hey, this something,

00:11:21.183 --> 00:11:24.143
this bad thing happened that if we get audited, we may have to explain,

00:11:24.283 --> 00:11:28.923
you know, so it's speaking to, giving examples in each layer that, again,

00:11:29.163 --> 00:11:33.103
speaks to the person you're talking with about what they care about. That's interesting.

00:11:33.463 --> 00:11:36.983
So did you work backwards then? Did you have a meeting with that group?

00:11:37.203 --> 00:11:40.743
And I mean, obviously you'd been there seven years or five years,

00:11:40.923 --> 00:11:45.443
et cetera. So you kind of already knew the individuals that you were trying

00:11:45.443 --> 00:11:49.563
to sell and what they were caring about at this point, right?

00:11:49.723 --> 00:11:53.843
But that's when our connection to the analysts paid off. Okay.

00:11:54.459 --> 00:11:59.179
So the analysts know the problems. The analysts know the data issues.

00:11:59.279 --> 00:12:03.999
The analysts know, hey, this data was down for two months last year,

00:12:04.019 --> 00:12:05.859
and this happened, this happened, this happened.

00:12:06.039 --> 00:12:11.059
So making those connections with the people using the data day in and day out

00:12:11.059 --> 00:12:17.419
was critical to then getting the evidence and the ROIs and the real world examples

00:12:17.419 --> 00:12:20.579
that we could then present to the management. Oh, it's incredible.

00:12:20.819 --> 00:12:26.819
It's absolutely incredible. So what happened next? You have these real-world examples.

00:12:27.079 --> 00:12:32.039
You've got your business folks agreeing that these are the problems that are

00:12:32.039 --> 00:12:37.099
really kind of impacting them and what that downstream impact looks like. But then what?

00:12:37.439 --> 00:12:41.699
How did that conversation then transition into, let's govern your data?

00:12:41.699 --> 00:12:44.759
Yeah. It's never a fast process.

00:12:45.079 --> 00:12:50.939
So again, with a team of four, including me, you have to put a lot of the most

00:12:50.939 --> 00:12:55.899
of the onus on the business units themselves to govern their own data.

00:12:56.939 --> 00:13:01.179
So for a period of time, you have to work a little bit like a project manager.

00:13:01.379 --> 00:13:03.399
And I know that's not fun and sexy.

00:13:03.899 --> 00:13:08.719
But in the beginning, showing people this is how you do it.

00:13:08.839 --> 00:13:12.439
It's not as bad as you think it is. But let us help you get started.

00:13:12.859 --> 00:13:16.959
Keeping them on task a little bit is really helpful. Because otherwise,

00:13:17.239 --> 00:13:19.759
you know it's going to go into the background of all their other work.

00:13:20.239 --> 00:13:23.859
So the now what is in part of our data governance framework,

00:13:23.899 --> 00:13:27.579
we had our steering committee, which is what we presented to the upper management

00:13:27.579 --> 00:13:29.679
that were like, oh my gosh, wow, we do have a problem.

00:13:29.999 --> 00:13:37.259
And then they nominated what we call the actual regular committee,

00:13:37.519 --> 00:13:42.339
which is the groups of folks that are at like a middle management layer usually,

00:13:42.339 --> 00:13:46.699
but they're not actually programming in the data, but they know what's going

00:13:46.699 --> 00:13:49.079
on in their different business units.

00:13:49.319 --> 00:13:55.499
And then they nominated the work groups. So we worked with this middle group

00:13:55.499 --> 00:13:58.459
of folks to then actually, council, that was the word.

00:13:58.559 --> 00:14:03.779
So this was the council, the Data Governance Council then nominated the folks

00:14:03.779 --> 00:14:06.439
in the areas that they needed work on.

00:14:06.579 --> 00:14:10.339
So they would identify, we're having a problem with this data.

00:14:10.579 --> 00:14:16.619
Here are the folks in the work group that can help implement and document and do the work.

00:14:17.219 --> 00:14:21.939
So then we would have regular meeting cadences. So the steering committee was quarterly.

00:14:22.179 --> 00:14:26.619
They obviously don't want to meet that often. And we only met if there was something to talk about.

00:14:26.999 --> 00:14:29.019
The council was monthly.

00:14:29.599 --> 00:14:33.199
Again, if there was something to talk about, sometimes we would split up the

00:14:33.199 --> 00:14:37.599
councils into business units if we didn't make progress in one business unit, but we did in another.

00:14:37.859 --> 00:14:42.399
And then the work groups, it really was up to them. We tried to even do those

00:14:42.399 --> 00:14:47.099
weekly or biweekly. and that was kind of keeping them on task but,

00:14:47.644 --> 00:14:51.704
what you have to do is work with a problem that they're currently having.

00:14:52.064 --> 00:14:55.524
So they're already dealing with it. They're already having issues with it.

00:14:55.664 --> 00:14:59.444
So adding on this additional work was actually benefiting them a lot quicker.

00:14:59.604 --> 00:15:05.604
So they saw their own ROI, even if it wasn't in dollars, faster because it was

00:15:05.604 --> 00:15:06.984
something they identified.

00:15:07.304 --> 00:15:13.304
That is actually a key win right there, what you just said, because it's always

00:15:13.304 --> 00:15:16.844
like the carrot, the stick, the carrot, the stick, and nobody knows how to create

00:15:16.844 --> 00:15:19.724
a carrot and doesn't want to use the stick, right?

00:15:20.004 --> 00:15:24.064
In terms of getting that working group working. Because my next question was

00:15:24.064 --> 00:15:25.704
going to be, how did you get them to show up to a meeting?

00:15:25.904 --> 00:15:30.344
Because most people just say, I'm too busy. But you answered that by telling

00:15:30.344 --> 00:15:34.004
us that it sounds like you used the steering group,

00:15:34.244 --> 00:15:38.024
the middle management group, to figure out what the challenges are that they

00:15:38.024 --> 00:15:41.784
wanted to address and then brought that back down to the working group that

00:15:41.784 --> 00:15:44.804
obviously was trying to address it for that individual or group of individuals

00:15:44.804 --> 00:15:50.064
within that business line and then use that as the immediate focus for that group. That's genius.

00:15:51.584 --> 00:15:55.744
Exactly. Well, every group has problems. Maybe there was one group.

00:15:56.064 --> 00:15:59.764
We started with like 10 different business lines. Maybe there was one that was

00:15:59.764 --> 00:16:02.324
like, well, we can't identify anything that you could help with immediately.

00:16:02.624 --> 00:16:05.944
So we were like, all right, we'll give you your space.

00:16:06.204 --> 00:16:08.244
When you want to come back to us, come back to us. We can't,

00:16:08.324 --> 00:16:10.224
we couldn't force anyone to do this work.

00:16:10.384 --> 00:16:14.804
But like I said, like that was one group out of about 10 and the rest of them,

00:16:14.904 --> 00:16:17.424
they had a list, laundry list of issues.

00:16:17.864 --> 00:16:24.824
So I don't know any group in any company that cannot say these are the issues we're struggling with.

00:16:24.924 --> 00:16:29.424
Even if it's not a specific data issue, usually you can trace it back to a data issue. Right.

00:16:29.944 --> 00:16:34.464
Yeah. I love that because it's, it's, you immediately know the impact you're

00:16:34.464 --> 00:16:36.604
making. Everybody is on the same page.

00:16:37.124 --> 00:16:43.004
And it isn't this build it and they will come fiasco that I see so many organizations.

00:16:43.424 --> 00:16:47.284
That's where some of the false starts were obviously like, build the technical

00:16:47.284 --> 00:16:49.544
lineage and everyone's going to be so excited.

00:16:50.184 --> 00:16:55.424
What is that? What is that? Like they say. Your poor, your poor little anical

00:16:55.424 --> 00:16:57.884
nerd heart. Your, your nerd heart was, was hurting.

00:16:58.364 --> 00:17:01.784
They say they need it. I need to see visibility and traceability.

00:17:02.004 --> 00:17:05.864
Okay. I'll give it to you. never looks at it. That's not what I wanted.

00:17:06.464 --> 00:17:10.764
Accurate data. Yeah. They look at it to check the box to say, yep, I've got that.

00:17:11.484 --> 00:17:15.404
But the usability is relatively low.

00:17:15.624 --> 00:17:20.524
So Carly, I love this approach. And it's one that we try to evangelize,

00:17:20.624 --> 00:17:23.104
right? It's like really, let's stop the bleeding.

00:17:23.364 --> 00:17:25.824
What's your biggest problem today?

00:17:26.344 --> 00:17:32.304
Let's bring all hands together to start focusing But I think one of the things

00:17:32.304 --> 00:17:36.904
that we tend to run into or I've seen happen at organizations is,

00:17:37.499 --> 00:17:44.899
we get some wins, right? But then continuing to drive the program stalls because

00:17:44.899 --> 00:17:46.939
it feels like we've addressed the challenge.

00:17:47.279 --> 00:17:52.579
We've moved past it. So now what, right? So having that ongoing program for,

00:17:52.599 --> 00:17:56.019
you know, now the two years subsequent to those conversations,

00:17:56.019 --> 00:17:57.639
how do you keep that going?

00:17:58.139 --> 00:18:05.119
Well, first of all, we had such a large chunk of ramp up that we still are working

00:18:05.119 --> 00:18:08.679
on a lot of the stuff that have been initial, I mean, people gave us,

00:18:08.759 --> 00:18:11.779
like I said, laundry lists. It's not just one problem.

00:18:12.479 --> 00:18:18.199
So, but the ones that you really, I don't think you ever get to a state of fully governed.

00:18:18.399 --> 00:18:21.759
I really think that's kind of a pie in the sky. That's not real.

00:18:22.399 --> 00:18:26.239
Maybe it's just because we have such a huge company, but I feel like there's

00:18:26.239 --> 00:18:28.879
always room for improvement of our data.

00:18:29.199 --> 00:18:33.639
I mean, our company has been around for, I think at least 25, if not 30 years.

00:18:34.019 --> 00:18:37.979
So that much historical information, that much data that we've gathered over

00:18:37.979 --> 00:18:42.159
time, we're still doing kind of backfills of that.

00:18:42.319 --> 00:18:46.619
But I would say that if you get to a point where it kind of slows down a bit,

00:18:46.859 --> 00:18:48.199
there's different flavors.

00:18:48.439 --> 00:18:54.019
So we start gently, if that makes sense. So when we start out with a group, we start gently.

00:18:54.299 --> 00:18:59.359
We say, all right, start with just a business glossary. Tell us the words that you know.

00:18:59.899 --> 00:19:03.719
There's much more advanced data governance things out there like approval workflows,

00:19:03.959 --> 00:19:10.659
policies, you know, SOPs, procedures, there's so much more to it that if you

00:19:10.659 --> 00:19:14.659
just throw them into the deep end, they'll never make it.

00:19:14.919 --> 00:19:18.119
So there's really never an official stall.

00:19:18.579 --> 00:19:21.959
You just need to make sure that you bring them to the next step.

00:19:22.159 --> 00:19:26.899
Like, great, you've done this very baby step. Let's take the next baby step

00:19:26.899 --> 00:19:28.539
when you're ready. That's fantastic.

00:19:29.667 --> 00:19:33.467
You just sounded like my daughter's kindergarten teacher. Alex,

00:19:33.867 --> 00:19:35.087
tell me what words you know.

00:19:36.607 --> 00:19:40.607
I hate to say it, but really, that's what you have to do. You have to make people

00:19:40.607 --> 00:19:43.427
realize that it's in their best interest.

00:19:43.687 --> 00:19:47.767
And I mean, the carrot, but you also, you know, we also had,

00:19:47.867 --> 00:19:50.107
you know, a data governance winner of the month.

00:19:50.247 --> 00:19:54.887
It was like the star of the month. And I mean, we didn't do anything super fancy

00:19:54.887 --> 00:20:00.447
except like a nice little email. But it really, it helps tell the story,

00:20:00.627 --> 00:20:01.867
like the kindergartner story.

00:20:02.087 --> 00:20:07.107
And not to say that people act like kindergartners, but people want recognition

00:20:07.107 --> 00:20:09.627
and praise for the work that they're doing.

00:20:10.007 --> 00:20:18.307
This is not sexy, show you how beautiful AI, if the word AI doesn't exist in this space right now.

00:20:18.627 --> 00:20:22.787
But, you know, this is not something you're going to present to the management board.

00:20:23.227 --> 00:20:26.927
Because it's not something that they care about. But

00:20:26.927 --> 00:20:31.307
giving them recognition for the hard work they're doing and showing how it is

00:20:31.307 --> 00:20:37.947
benefiting is what they need to have this snowball effect to keep going to the

00:20:37.947 --> 00:20:44.287
point where we actually got it as part of the JDs for some clinical folks to maintain,

00:20:44.287 --> 00:20:46.567
you know, good data stewardship.

00:20:47.681 --> 00:20:53.101
Going back to the point you made earlier, where you were able to document the impact.

00:20:53.721 --> 00:20:57.281
Let's take that a step further, because the thing that I was so impressed with

00:20:57.281 --> 00:21:02.581
was the full ROI, not just the impact, but the cost and the benefit aspect of it.

00:21:03.101 --> 00:21:08.101
Can you walk us through, like, at what point in this entire journey did you do that?

00:21:08.341 --> 00:21:12.561
Was it, you know, the way you got the buy-in at the executive level? Is that what won it?

00:21:12.681 --> 00:21:15.781
And what is the methodology there that you can share with our listeners?

00:21:15.781 --> 00:21:18.341
Because I'm sure many people would get something out of that.

00:21:18.961 --> 00:21:23.421
Yeah, it was what we started with when we went back to the management.

00:21:23.661 --> 00:21:28.761
So you remember we had one person buy-in, but then we needed to get a much larger group buy-in.

00:21:29.101 --> 00:21:33.021
That's what we started with. And the way that we did that was,

00:21:33.161 --> 00:21:36.861
again, with our connection with the analysts, we identified problems.

00:21:37.341 --> 00:21:43.221
And so we kind of thought of, all right, so what problems can we trace back to finances?

00:21:43.741 --> 00:21:47.541
Everyone says this, but it's really true. Any data issue that you have,

00:21:47.661 --> 00:21:54.841
you can get back to dollars, whether that be, you know, technical debt in data storage,

00:21:55.041 --> 00:21:57.801
whether that's time and effort, which is what we really focused on,

00:21:57.921 --> 00:22:01.101
whether that's like when you get dirty data, what are the repercussions?

00:22:01.621 --> 00:22:06.421
So we picked a couple of examples, especially ones that just happened that were

00:22:06.421 --> 00:22:11.541
kind of fresh in our mind, like, hey, this data was not flowing for,

00:22:11.961 --> 00:22:14.821
I don't know, a month, and nobody really knew it.

00:22:14.821 --> 00:22:19.901
And so this, what we did was we figured out, all right, we talked to the analysts

00:22:19.901 --> 00:22:21.481
who was impacted by this.

00:22:21.601 --> 00:22:24.221
And they, you know, people were like, yeah, we were impacted.

00:22:24.241 --> 00:22:26.361
And we said, okay, what was impacted?

00:22:26.661 --> 00:22:29.601
Oh, this algorithm, this predictive model. And then we said,

00:22:29.661 --> 00:22:33.101
all right, where does that predictive model go? Well, it goes to the nurses.

00:22:33.621 --> 00:22:36.721
So then we went, all right, let's talk to our head of nursing.

00:22:36.721 --> 00:22:41.481
So we went to our nursing department and said, all right, if this data is bad.

00:22:42.073 --> 00:22:45.573
What happens? What happens in the clinics? What happens to the nurses?

00:22:46.053 --> 00:22:51.553
And they said, the nurse then has to go use this algorithm that's like 10 pages

00:22:51.553 --> 00:22:53.193
long and figure it out themselves.

00:22:53.673 --> 00:22:56.513
So then we said, how much time does that take a nurse?

00:22:56.913 --> 00:23:00.793
And they were like, it takes about 10 to 15 minutes to do that rather than having

00:23:00.793 --> 00:23:02.493
it right in front of them. Okay.

00:23:02.993 --> 00:23:07.793
So then we went back into our data and said, how many people in this time period,

00:23:07.793 --> 00:23:11.553
how many patients were impacted by this. We got that number.

00:23:12.193 --> 00:23:15.993
Then we went a step further just to cover all of our bases.

00:23:16.233 --> 00:23:20.833
We went to HR and said, talk about our average reimbursement for our nurses

00:23:20.833 --> 00:23:25.073
across the country so we can actually put, I know we went really far because

00:23:25.073 --> 00:23:28.513
we wanted to make sure we were very solid in that.

00:23:28.813 --> 00:23:32.573
So you can see you, we really, you just keep digging. You keep trying to figure

00:23:32.573 --> 00:23:33.933
out and get back to the dollars.

00:23:34.133 --> 00:23:38.013
So time is money, but then you have to figure out how much money is the time

00:23:38.013 --> 00:23:41.513
and then you have to have a hard numbers of these are the patients that were impacted.

00:23:41.773 --> 00:23:46.313
We were very lucky to that we were able to do that because we had such good

00:23:46.313 --> 00:23:47.433
connections with the analysts.

00:23:47.933 --> 00:23:52.473
Right. You can see how kind of like building that middle tier is important because

00:23:52.473 --> 00:23:55.113
those people can tell you the dirty secrets.

00:23:56.433 --> 00:24:00.113
I'm curious how long did that take like that example that you just went through

00:24:00.113 --> 00:24:03.373
like how long did it take for the team to get from beginning to end of that?

00:24:03.673 --> 00:24:07.493
I think that took about a month. The hardest part was setting up the meetings

00:24:07.493 --> 00:24:09.933
and trying to explain what we wanted.

00:24:10.373 --> 00:24:16.393
A lot of times, the biggest hurdle that people run into is explaining what you're

00:24:16.393 --> 00:24:21.533
trying to do in the verbiage that the person you're talking to understands.

00:24:21.833 --> 00:24:25.633
A nurse is going to hear, oh, the data didn't flow from point A to point B,

00:24:25.693 --> 00:24:27.473
and they're like, how does that impact me?

00:24:27.773 --> 00:24:34.453
But if you say, hey, this specific spot in the EMR was not correct for two months, how did that impact you?

00:24:34.693 --> 00:24:39.353
Then they get that. So it's translating the issue into the different areas it's

00:24:39.353 --> 00:24:43.113
impacting, scheduling the meetings, as we all know, those get tough.

00:24:43.473 --> 00:24:48.633
And then trying to, you know, every single person we met with then gave us another

00:24:48.633 --> 00:24:51.813
person to talk to. Yeah. So that was the hardest part.

00:24:52.389 --> 00:24:57.709
It's incredible. I really do love that story because I feel that a lot of people

00:24:57.709 --> 00:25:02.049
just stop with, well, there's a repercussion. We can't have that happen again, right?

00:25:02.229 --> 00:25:08.069
But we put it in real dollars and everyone mobilizes because that impacts the

00:25:08.069 --> 00:25:09.509
executives all the way down.

00:25:10.409 --> 00:25:15.309
Everybody mobilizes on that. So that's pretty cool. You can also Google a lot.

00:25:15.689 --> 00:25:20.629
So for example, Another example we looked at, so our company in healthcare,

00:25:21.049 --> 00:25:24.529
and not just our company, in healthcare in general, there's a lot of reimbursement.

00:25:24.629 --> 00:25:27.229
There's a lot of Center for Medicaid Services, their National Health,

00:25:27.369 --> 00:25:31.289
the NHSN, National Health Safety Network, a lot of things that are critical.

00:25:31.569 --> 00:25:34.069
And if something goes wrong, there are fines.

00:25:34.429 --> 00:25:39.309
So you can look up previous fines and like why they were fine.

00:25:39.689 --> 00:25:45.689
And then that's also, you know, a little bit of a this is what could happen.

00:25:45.989 --> 00:25:49.749
So that is not you can't always just say what could happen.

00:25:50.029 --> 00:25:54.969
But also, you don't need your own. If you don't have access to your specific

00:25:54.969 --> 00:25:59.449
data, like we did in the previous example, you can still find examples that

00:25:59.449 --> 00:26:02.869
could relate to you, that could have a possible impact.

00:26:03.069 --> 00:26:06.729
It's a great call out, particularly if like you find that, okay,

00:26:06.909 --> 00:26:10.509
if this happens, this could be the possible impact and doing a little research

00:26:10.509 --> 00:26:15.409
to see if there's any oversight in that data element and the pathway there so

00:26:15.409 --> 00:26:17.669
that you can kind of make a case as well.

00:26:18.269 --> 00:26:21.609
You know, this is loosey-goosey. We've been lucky. I've had a couple of clients

00:26:21.609 --> 00:26:22.609
where that has been the case.

00:26:23.309 --> 00:26:27.549
Exactly. Yeah. Yeah. So Carly, I was going to say, so for other organizations

00:26:27.549 --> 00:26:33.529
that may be listening to this methodology and are kind of thinking of implementing

00:26:33.529 --> 00:26:36.529
something similar for their own organizations.

00:26:36.829 --> 00:26:40.609
What are some of the key steps that you would really recommend they prioritize?

00:26:41.029 --> 00:26:44.769
I would say make a connection with the analysts. Talk to them.

00:26:44.989 --> 00:26:49.289
Have conversations with the analysts because they are the ones that can explain

00:26:49.289 --> 00:26:50.669
what they struggle with.

00:26:50.889 --> 00:26:54.469
Like what issues do the analysts have? That's where we started.

00:26:55.169 --> 00:26:59.489
Another way you can start is you can start with the business units as well and

00:26:59.489 --> 00:27:01.289
ask them what issues they're having.

00:27:01.629 --> 00:27:04.909
And their issues may not be data-related, but again, like I mentioned,

00:27:05.009 --> 00:27:09.869
you can usually kind of explain that metadata is still data.

00:27:10.329 --> 00:27:15.309
So we start with our business units. So let's put the analysts aside for the

00:27:15.309 --> 00:27:18.109
moment. With our business units, we start with the business glossary.

00:27:18.649 --> 00:27:24.489
We start with a report catalog because most business units do their own slight reporting.

00:27:24.709 --> 00:27:26.889
It doesn't always live in IT.

00:27:27.229 --> 00:27:31.289
I think everyone knows what Excel sheets they kind of do manipulations on.

00:27:31.729 --> 00:27:36.689
And then what do they maintain in an Excel sheet that's difficult to maintain,

00:27:36.689 --> 00:27:41.949
that they need to know when something changes and they need a little bit of an audit trail.

00:27:42.329 --> 00:27:46.709
So the business glossary was the biggest hit because, I mean,

00:27:46.829 --> 00:27:51.249
turnover, how many new people do you get that have no clue what an abbreviation

00:27:51.249 --> 00:27:53.069
means? You know, these basic things.

00:27:53.289 --> 00:27:55.889
I think those are the biggest, those are the first steps.

00:27:56.229 --> 00:27:59.809
Particularly in a large organization where we have that kind of turnover,

00:28:00.229 --> 00:28:04.629
ramp up is always pretty slow for those reasons or issues come up or wasting time.

00:28:05.089 --> 00:28:08.409
Creating a report that nobody understands because you're using the wrong terminology.

00:28:09.009 --> 00:28:12.869
You know, those kind of things that end up happening when a new hire comes on.

00:28:13.149 --> 00:28:17.329
Yep. Yeah, that makes sense. You mentioned, you know, at the start of this,

00:28:17.489 --> 00:28:19.269
we discussed the two things that wowed me.

00:28:19.389 --> 00:28:23.789
One was the ROI, but the other one was the size of your team.

00:28:25.869 --> 00:28:29.149
I think you spoke to this, though. You definitely spoke to this, right?

00:28:29.309 --> 00:28:34.089
Because it's not necessarily the fact that you went off and hired governance

00:28:34.089 --> 00:28:35.629
people to do governance work.

00:28:35.789 --> 00:28:39.309
That's really not what that was about. It was about enabling the business with

00:28:39.309 --> 00:28:42.249
a methodology and a process and a reason to do governance work.

00:28:42.829 --> 00:28:48.729
So you said you had a team of four. I'm curious, how much of the business do you support?

00:28:49.229 --> 00:28:53.749
Like how many different teams are you supporting? How many working groups is your team supporting?

00:28:54.049 --> 00:28:57.489
And what do the four individuals that you have on your team do?

00:28:58.269 --> 00:29:04.049
We have, I want to say we have about 12 business units. We currently covering

00:29:04.049 --> 00:29:09.189
care delivery, which is our patient type services. Like I mentioned,

00:29:09.369 --> 00:29:12.009
pharmacy and vascular and dialysis.

00:29:12.249 --> 00:29:16.309
U.S. We have an international piece as well.

00:29:16.469 --> 00:29:19.769
And actually, that's kind of what I'm, I have a slightly new role that's called

00:29:19.769 --> 00:29:21.589
clinical data integrity and stewardship.

00:29:21.809 --> 00:29:25.109
And they didn't, sometimes the word data governance scares people,

00:29:25.289 --> 00:29:27.789
as many, many people know.

00:29:28.129 --> 00:29:32.549
And I'm going to be starting to look at more global data, which is kind of a

00:29:32.549 --> 00:29:34.049
new thing that I'm very excited about.

00:29:34.509 --> 00:29:40.729
But back to my staff, we focus on one area, which has about 12 different business units.

00:29:40.729 --> 00:29:45.629
The working groups and keeping everybody in line is kind of not in line,

00:29:45.749 --> 00:29:51.729
but actually helping people navigate their own governance is kind of a full-time job.

00:29:51.909 --> 00:29:56.889
So really one person is dedicated to running the framework, making sure they

00:29:56.889 --> 00:30:01.809
assist these groups with, we use a metadata management software, Calibra,

00:30:02.289 --> 00:30:06.629
assisting these users to get their information in Calibra, teaching them how

00:30:06.629 --> 00:30:12.369
to use it, creating training materials, really enabling the business folks to

00:30:12.369 --> 00:30:13.829
do their own governance.

00:30:14.049 --> 00:30:16.969
So that's really a one full FTE.

00:30:17.812 --> 00:30:23.652
And like I said, it's not always the sexy, fun work, but it's what gets the snowball.

00:30:24.112 --> 00:30:29.932
And that person also managed a JIRA board and is the person that kind of like

00:30:29.932 --> 00:30:33.932
communicates the most with the work groups and the council.

00:30:33.932 --> 00:30:41.092
We have another person that helps with triaging and translating between the

00:30:41.092 --> 00:30:42.792
business and the data folks,

00:30:42.792 --> 00:30:49.652
because that's always a challenge is understanding that communication between the two.

00:30:49.812 --> 00:30:53.412
Because you'll have the business saying there's a problem and the technical

00:30:53.412 --> 00:30:57.952
people saying, no, there isn't a problem, but there is a data discrepancy.

00:30:58.812 --> 00:31:02.712
So we have a person that helps bring bridges, those gaps like,

00:31:02.912 --> 00:31:05.992
oh, well, maybe you don't see it on the back end, but in the front of it looks

00:31:05.992 --> 00:31:09.432
like this and we think it's blah, blah, blah. So we kind of have a translator.

00:31:10.112 --> 00:31:15.172
That sounds strange, but all of the people that we have, we actually stole from

00:31:15.172 --> 00:31:19.652
other areas of the company that had been with the company for a period of time

00:31:19.652 --> 00:31:22.052
and understood our systems.

00:31:22.292 --> 00:31:27.072
And then the last person that we have is a more architecture type person.

00:31:27.432 --> 00:31:31.432
And he focuses a lot on what's the best,

00:31:32.099 --> 00:31:36.699
architecture for not only our metadata management software, but when you're

00:31:36.699 --> 00:31:39.499
creating different governance processes.

00:31:40.079 --> 00:31:44.719
And he is more, once you get the metadata and the data that you need for governance,

00:31:44.939 --> 00:31:48.879
what is the best architecture way to make sure that that is a seamless flow?

00:31:49.219 --> 00:31:51.759
He's the one that worked on our reference data management.

00:31:52.279 --> 00:31:57.619
Also, he's incredibly good at, you know, training documentation of the tool

00:31:57.619 --> 00:32:04.539
itself and the like approval processes of the data. So it's not just kind of wild west.

00:32:04.679 --> 00:32:10.999
So it's then he is putting the controls around what these groups are starting to enter.

00:32:11.379 --> 00:32:16.199
That's great. I heard enablement, somebody who's focused on enabling others

00:32:16.199 --> 00:32:20.659
and, you know, within Calibra, within the platform that you've selected.

00:32:21.039 --> 00:32:25.579
There's also the translator who handles triage and communications and then the

00:32:25.579 --> 00:32:28.899
architect. Did I miss one? Me. Wow. Okay. All right.

00:32:31.419 --> 00:32:35.599
I do a bit of all of it, to be honest. I have the technical knowledge to do

00:32:35.599 --> 00:32:39.159
the programming and the tool. I have the relationships to do the translation.

00:32:39.499 --> 00:32:42.999
And I'm also, I run the steering committee, which is the highest level.

00:32:43.539 --> 00:32:48.839
So my hands in all of it, honestly, I wouldn't say I'm one, I'm the puppet master. That's kidding.

00:32:49.139 --> 00:32:53.219
And we call, I always call my team member unicorns because they have to speak

00:32:53.219 --> 00:32:56.479
multiple languages. They have to speak the technical, they speak the business,

00:32:56.639 --> 00:32:59.719
they have to understand the flow within our company.

00:32:59.959 --> 00:33:02.499
So those are the kinds of people that you want on your team.

00:33:02.979 --> 00:33:08.119
A curiosity I have is, you mentioned that all these individuals were in other

00:33:08.119 --> 00:33:09.259
parts of the organization.

00:33:09.839 --> 00:33:14.179
How did you convince them to do this? Because data governance is not sexy.

00:33:14.379 --> 00:33:17.839
Nobody, you say data governance, people will literally run away from you.

00:33:19.599 --> 00:33:26.939
My wedding personality. I don't know if I had your emails or Slack messages anymore today.

00:33:27.599 --> 00:33:33.039
Like, how did you convince that? Well, one of them really enjoyed...

00:33:33.441 --> 00:33:37.841
Processes and procedures. Like that was her, I know, right?

00:33:38.241 --> 00:33:41.441
She's, again, she was like a unicorn in that regard.

00:33:41.601 --> 00:33:45.321
So she was the one that everyone would go to in our technical part to be like,

00:33:45.701 --> 00:33:47.581
well, what's the standard of blah, blah, blah.

00:33:48.221 --> 00:33:52.081
So, you know, there's a combination of why people choose to move departments.

00:33:52.141 --> 00:33:55.581
And part of it is like, I don't see growth in the place that I am.

00:33:56.161 --> 00:34:00.121
And the way that our department was set up is we were set up outside of IT,

00:34:00.681 --> 00:34:03.421
which gave us a lot more flexibility.

00:34:03.721 --> 00:34:09.001
And I think that appealed to some of these folks that were in IT that were kind

00:34:09.001 --> 00:34:12.441
of restrained by their processes and procedures.

00:34:12.681 --> 00:34:16.861
And it was more limited to think out. They couldn't think outside the box.

00:34:16.961 --> 00:34:20.461
They did think outside the box, but they couldn't implement some of the stuff

00:34:20.461 --> 00:34:21.601
that they wanted to implement.

00:34:22.121 --> 00:34:25.781
So she was all excited. She's the one took over the data governance framework.

00:34:25.981 --> 00:34:29.321
She came over. The second one was an analyst.

00:34:29.821 --> 00:34:35.041
So she was part of the analytical group and actually led a group of analysts.

00:34:35.221 --> 00:34:41.541
She was so frustrated with the problems with our data that she was like,

00:34:41.661 --> 00:34:47.901
it would save me time to join this group and fix the data instead of doing all this disaster.

00:34:48.721 --> 00:34:54.561
So she's the translator. So she was the one who also loved to dig and try to

00:34:54.561 --> 00:34:55.761
understand data issues.

00:34:55.981 --> 00:34:59.901
When you understand data issues, you also understand a lot of what's happening

00:34:59.901 --> 00:35:01.741
in the patient realm, in the clinic realm.

00:35:01.861 --> 00:35:04.041
And it's kind of, it's very interesting to her.

00:35:04.421 --> 00:35:08.141
And then the last person just, he really liked me and was like,

00:35:08.221 --> 00:35:09.501
oh, let's go come join your group.

00:35:09.721 --> 00:35:15.741
He was the person that helped us analyze and look at the data governance software

00:35:15.741 --> 00:35:17.981
and the metadata management tool.

00:35:18.361 --> 00:35:24.821
And he also was in IT and felt restrained that he couldn't do as much documentation

00:35:24.821 --> 00:35:27.581
as he wanted, which I thought was really fascinating.

00:35:27.821 --> 00:35:30.581
I know. See, you have to find these unicorns. I know.

00:35:31.181 --> 00:35:36.461
And he wanted to do things that took a little bit longer, but were going to help later on.

00:35:36.581 --> 00:35:40.681
And the pressure in IT is do, do, do, get it done, get it done, get it done.

00:35:41.141 --> 00:35:45.661
So we found those people by talking to them.

00:35:45.801 --> 00:35:50.301
And just like my mentor stole me because I would ask the questions.

00:35:50.561 --> 00:35:53.941
I asked the questions and then you kind of pick the people out that ask the

00:35:53.941 --> 00:35:57.061
questions because you know that they care and they want to understand it.

00:35:57.657 --> 00:36:00.277
Well, I think we have two more questions for you.

00:36:02.377 --> 00:36:05.457
I'm just fascinated right now, by the way. I know. I'm like,

00:36:05.597 --> 00:36:09.037
where are these unicorns? I've spent 20 years trying to find my people.

00:36:11.257 --> 00:36:17.377
So I think people, I think Mr. Rogers said when there's an accident or something

00:36:17.377 --> 00:36:18.997
bad happens, look for the helpers.

00:36:19.257 --> 00:36:22.757
Yep. Look for those that, and that's similar in this scenario.

00:36:22.777 --> 00:36:26.517
Look for those that ask the questions and that care about the data quality.

00:36:26.517 --> 00:36:28.757
Those are the ones that are going to be your unicorns.

00:36:29.397 --> 00:36:34.197
Yeah, absolutely. I love that. I absolutely love that. So you mentioned that

00:36:34.197 --> 00:36:36.277
you don't report and you're outside of IT.

00:36:37.717 --> 00:36:42.337
What area are you in then? We are in our global medical office.

00:36:42.697 --> 00:36:45.697
So we're more clinically focused.

00:36:46.257 --> 00:36:51.157
And I think it doesn't have to be in some type of medical office that could

00:36:51.157 --> 00:36:55.097
also live in the business of the area that you're governing.

00:36:55.097 --> 00:36:58.137
This is just the way that our company was structured

00:36:58.137 --> 00:37:01.337
and set up in terms of like who gets the headcount and he doesn't and

00:37:01.337 --> 00:37:07.777
but living outside of it i think is really important or if you're in it having

00:37:07.777 --> 00:37:14.757
the kind of autonomy to say this is how we're going to do it is really important

00:37:14.757 --> 00:37:17.077
and another thing is we could help

00:37:17.077 --> 00:37:21.037
without telling someone we're going to charge for our time and effort.

00:37:21.397 --> 00:37:23.417
That is also really important.

00:37:24.337 --> 00:37:29.757
Yeah, I agree with that. I think it's also, I mean, IT always gets a bad rap

00:37:29.757 --> 00:37:31.237
and I always feel horrible.

00:37:31.457 --> 00:37:36.857
I mean, I obviously worked in IT for a number of years in the middle of me being

00:37:36.857 --> 00:37:42.037
a consultant, but I will say that, yeah, they always get a bad rap and it's

00:37:42.037 --> 00:37:45.337
not necessarily the individual's faults that work in there. Absolutely not.

00:37:45.497 --> 00:37:49.657
A lot of people want to do the right thing and help that there's just so much to do, right?

00:37:49.797 --> 00:37:54.477
But I think having it outside of IT also helps in terms of having those honest

00:37:54.477 --> 00:37:59.277
conversations about what's happening and no one's being defensive because it

00:37:59.277 --> 00:38:02.657
really is what is the right thing for the business, right? So you're put in

00:38:02.657 --> 00:38:04.037
that position, which is fantastic.

00:38:04.497 --> 00:38:08.277
I know Anjali is dying to ask about AI. I am.

00:38:08.437 --> 00:38:13.297
I am. So I know you touched on AI a little bit, saying it really isn't something

00:38:13.297 --> 00:38:15.197
that's in your realm right now.

00:38:15.237 --> 00:38:22.357
But are you starting to think about either AI for governance or governance for AI?

00:38:22.937 --> 00:38:25.757
Yes. AI governance is...

00:38:26.546 --> 00:38:31.226
It inherits data governance. So you need data governance to do AI governance.

00:38:31.446 --> 00:38:35.526
And then AI governance obviously is newer for a lot of folks.

00:38:35.726 --> 00:38:39.446
And one of the reasons why, like I mentioned earlier, I'm going more into the

00:38:39.446 --> 00:38:46.686
global space is because EMEA and other areas of the world are coming up with

00:38:46.686 --> 00:38:49.166
AI rules and regulations.

00:38:49.806 --> 00:38:54.226
So in order to meet those rules and regulations, you have to meet the AI governance

00:38:54.226 --> 00:38:59.166
standards they put in place, which then you need the data governance to enable the AI governance.

00:38:59.626 --> 00:39:05.386
So we're working on understanding what is needed in the EMEA space,

00:39:05.546 --> 00:39:09.586
which then we would kind of apply to the US space because we know US usually

00:39:09.586 --> 00:39:12.386
follows suit after a period of time.

00:39:12.666 --> 00:39:15.746
And that's actually really helpful because you have to think about fines.

00:39:15.746 --> 00:39:20.246
And if you don't have this in place, what's going to happen?

00:39:20.766 --> 00:39:26.666
And the challenge The challenge with AI governance right now is AI can't be

00:39:26.666 --> 00:39:29.606
left alone yet, especially in the healthcare space.

00:39:29.766 --> 00:39:32.506
You need a second set of eyes on things. You need review.

00:39:32.966 --> 00:39:39.446
So if you, a lot of people, like you have technical folks developing some type

00:39:39.446 --> 00:39:43.066
of AI model, and then you have a physician that looks at it and tears it apart.

00:39:43.726 --> 00:39:49.446
So we have the challenge of not just doing the data governance part of the AI

00:39:49.446 --> 00:39:53.686
governance, but the AI governance part with how do you incorporate it in healthcare

00:39:53.686 --> 00:39:57.346
and make an impact to the patients without getting hallucinations.

00:39:57.546 --> 00:40:01.926
So I think there's two parts of AI governance. I'm helping more with the data

00:40:01.926 --> 00:40:07.046
governance piece, and I'm part of the bigger conversation of AI governance,

00:40:07.246 --> 00:40:14.186
which then has to bring in the experts of the field to review the models themselves,

00:40:14.186 --> 00:40:18.666
not the technical piece, but the clinical and the outcomes and say,

00:40:18.826 --> 00:40:21.466
yeah, okay, this makes sense, or this is ridiculous.

00:40:22.046 --> 00:40:24.926
It's not worth it. Stop altogether. Mm-hmm.

00:40:25.330 --> 00:40:29.930
Yeah. Love that. Yeah. Yeah. And that's what we're seeing so many of our clients,

00:40:30.150 --> 00:40:34.090
especially at EMEA, start to struggle with where they're saying,

00:40:34.210 --> 00:40:40.270
you know, we jumped in the AI pool, but now we got to make sure we don't get fined for doing it.

00:40:40.270 --> 00:40:44.670
So they're really looking at how do we take some of the work that's been done

00:40:44.670 --> 00:40:50.030
foundationally on improving or at least governing their data and applying that

00:40:50.030 --> 00:40:56.310
frictionlessly to AI governance as well.

00:40:56.550 --> 00:41:02.550
So big topic, I think, that's going to continue to explode over the next two to five years.

00:41:02.550 --> 00:41:07.410
And unfortunately, because it's not frictionless, you're going to need members

00:41:07.410 --> 00:41:11.990
of multiple areas of your company. And it's going to have to be a little bit

00:41:11.990 --> 00:41:16.090
of a brute force method until there's a better understanding of it.

00:41:16.250 --> 00:41:21.530
So it's critical to include members of the different areas that are impacted

00:41:21.530 --> 00:41:25.010
by whatever AI you're creating. That's critical.

00:41:25.290 --> 00:41:29.130
The end user, the person in the middle that has to run the model,

00:41:29.330 --> 00:41:32.850
the person in the beginning that's developing the model. And that's what I've

00:41:32.850 --> 00:41:37.770
been finding companies are not doing. They're not including everyone because

00:41:37.770 --> 00:41:39.630
they're like, oh, we don't want to keep having meetings.

00:41:39.850 --> 00:41:44.350
But because this space is so new, you can't leave anyone out yet.

00:41:44.710 --> 00:41:49.110
Not that you would in the future, but it's not streamlined enough yet to do that.

00:41:49.655 --> 00:41:53.595
Yeah. Yeah. That's such a great point. You can't set it and forget it.

00:41:53.715 --> 00:41:55.375
You absolutely cannot. Not yet.

00:41:55.815 --> 00:41:59.255
Not as to, you have to know what data you've poured into the model,

00:41:59.455 --> 00:42:01.015
make sure that that makes sense.

00:42:01.175 --> 00:42:05.595
You have to ensure that the model, you know, stays compliant throughout its

00:42:05.595 --> 00:42:09.595
usage and you need to understand the ultimate impact to, in your case,

00:42:09.675 --> 00:42:12.355
your patient, right. And others, it's consumers.

00:42:12.615 --> 00:42:16.615
But I, I, I find it interesting that organizations are just like,

00:42:16.735 --> 00:42:19.695
Oh, oh, I don't want to have another meeting, but wait a minute.

00:42:19.915 --> 00:42:24.755
This AI capability is actually accelerating a lot of what you're going to be doing as a business,

00:42:24.755 --> 00:42:31.415
and it probably is worth your time to make sure that it is accurate and complete

00:42:31.415 --> 00:42:36.835
and working in a way that impacts your business positively, not negatively in the long run.

00:42:37.835 --> 00:42:42.475
I'm waiting to see hopefully people just do the right thing moving forward because

00:42:42.475 --> 00:42:47.195
they have real reasons to do so beyond just, oh, somebody's telling me to.

00:42:47.415 --> 00:42:49.215
It's more actually there's financial impact here.

00:42:49.515 --> 00:42:55.155
That should be interesting. I like, I am just amazed that I got to meet you, Carly.

00:42:55.255 --> 00:43:00.255
And I'm so happy that our worlds collided a little bit there at CDAO.

00:43:00.675 --> 00:43:03.835
We're probably going to have you back like in a year, like next season,

00:43:03.995 --> 00:43:06.515
just so that we can hear what's going on with the AI stuff that you're doing

00:43:06.515 --> 00:43:08.015
as well, because I'm sure it's going to be phenomenal.

00:43:08.315 --> 00:43:11.975
Yeah, I'll talk about all the world data now. I even went to Germany two weeks

00:43:11.975 --> 00:43:17.775
ago for my first lesson on our global data, which is a whole problem in and

00:43:17.775 --> 00:43:20.095
of itself. I mean, you think about all the different countries.

00:43:20.435 --> 00:43:23.795
You know, I just had to deal with one country's regulations. Oh, boy.

00:43:24.535 --> 00:43:27.015
So in a year, I'll probably have more gray.

00:43:30.015 --> 00:43:34.275
That's awesome. All right, Carly. Well, thank you so much for being on.

00:43:34.435 --> 00:43:37.795
It's been an absolute pleasure chatting with you again and hearing your story

00:43:37.795 --> 00:43:41.255
again. And I absolutely learned so much and I'm sure our listeners did as well.

00:43:41.335 --> 00:43:43.955
So I appreciate that. You're welcome. Thanks for having me.