Episode 19: Factoring in the Human Element of Data Analytics

00:01

Welcome to In:confidence, the podcast for data ops leaders. In each episode, we asked thought leaders and futurists to break down the topics and trends concerning it and data professionals today, and to give us their take on what the data landscape will look like tomorrow. Let’s join the data conversation. 

 

Tina: Hey, Anne Marie

.

 

00:23

Thanks for joining in confidence. Today, I’m really excited that you agreed to be on our show.

 

00:31

I’ve known you since you were back at

 

00:37

Amgen. And then you went to see you and I did I say that right?

 

00:45

All right. And now you’re you’re leading a team at at Zillow or partner analytics, but I was really excited to bring you onto the show to get your input as a practitioner and leader in this space. But before we dive into it, let’s get warmed up. All right. Sounds good, burning questions that our audience wants to know such as, do you prefer red or white wine? 

 

Anne Marie: White wine for sure. Unless I’m eating a steak and then read any particular varietal? Usually Chardonnay or Rosae? Are my go to is really my true answer. This question is my favorite kind of wine is

 

01:27

poured

 

01:29

in my glass.

 

01:31

Yeah.

 

01:33

So yeah, I’m pretty easy in that regard.

 

01:37

Tina: Right. Okay, so chocolate or cheese. 

 

Anne Marie: Cheese, for sure.

 

01:43

And I live in Wisconsin. So that’s the obvious and traditionally, I think you had to say that. Yeah, yeah.

 

01:51

Yeah, we’re always listening.

 

01:56

Tina: Okay, waterski, or wakeboard? 

 

Anne Marie: Waterski, for sure. I did wakeboard briefly in my teenage years, but I’m not a person that likes falling down both from a ego standpoint and like physical harm standpoint. And in order to be really good at wakeboarding you have to fall down a million times while learning all the tricks and skiing you’re just go back and forth. Okay, but single ski or double ski. Oh, yeah. Deepwater slalom start, although my my father would say, Oh, yes. My, my father would say I’m a huge disappointment because I don’t do a duck start. So

 

02:35

really? The same thing. The day we didn’t talk? Like it’s not even allowed anymore.

 

02:43

It’s not allowed. My dad did good reason. Right. Yeah, there’s good reasons not to plus, I mean, you know, let’s just be face it. The engine sizes are different. Back then compared to now. Oh, for sure. Yeah. It’s the number of horsepower on these votes are exponentially rising.

 

03:05

As all the lawsuits, so.

 

03:09

Tina: Okay, so last one true crime or sci fi true crime? 

 

Anne Marie: For sure. ello. I do, like both genres.

 

03:18

Definitely get pulled into the documentary side of online streaming or before I would get pulled into sci fi.

 

03:29

Hmm, okay. All right, doc. So visual, not podcast, not audio book, or book.

 

03:38

Yeah, I have not gotten into any true crime

 

03:42

books, print or audio books. I do only read nonfiction, because I am a huge weirdo. But I have not gotten into true crime. It’s mostly like memoirs and economics books are right.

 

Tina: Okay. Because you you have a master’s degree in economics. Yes. So true to form. Okay. All right. Yes. Well, there you have it. Okay. So we won’t talk about any of the like recent podcasts on true crime. What I really wanted to turn the conversation to, though, is

 

04:18

what is missing?

 

Anne Marie: From the conversation around data science for corporate use. What I’ve seen missing from corporate use for data science is a lack of strategy around how machine learning and AI models are leveraged. And I think that it comes from a place of difference in how corporate thinks about solving problems and how academics think about solving problems and data science teams when they were started 1015 years ago, we’re

 

05:00

I build with academics. And academics come in and they see a lot of data. And then they build models leveraging that data. And then they write white papers.

 

05:11

And then they go around, and they share those white papers with, with their friends and their journals. And they peer review. And it’s a very important process that moves science that moves academic research forward. When you apply that to corporate use, you then end up with a bunch of expensive trivia. Because you’ve taken data, you make a model, and then you write a white paper. And a sales director says, What am I supposed to do with this? Now,

 

05:39

you may not be predicting a thing that’s actually usable. So business problems, start with a current process where there is some sort of problem at a point of that process. If you then take that approach to data science and say, here’s where our problem is. What what information? Should we know at that point? Or could we know to make better decision decisioning points, what what

 

06:06

decisions are, could be automated, or a lot of times what I’ll ask is,

 

06:14

what’s the most boring part of your job, if I’m talking to, let’s say, an underwriter, to an insurance company,

 

06:20

and those boring, those boring things, the things that there’s totally easy for a human, but the only difficulty is the volume. Those are the things that are really exceptional to leverage data through automation, and Intelligent Automation bringing in and machine learning models.

 

06:40

So that you get to a point where the data goes into a model, and then you have action that takes place after that model. And that action is ideally embedded into a business processing, so that you are able to capitalize it on it and measure the value.

 

07:01

And oftentimes, these models work really well alongside humans, because the models are really good at doing the same thing over and over and over again at scale.

 

07:12

And then there are inevitably outliers. And so oftentimes, the machine can automate the decisions that you see over and over and over and over again. And then the humans can take care of the ones that are weird. machine can say, this is weird. Please, human look at this now, and humans are great at critical thinking. So how do we leverage the best of the human brain and the best of the computer brain in order to get the most optimized outcomes? Those are business questions. And they’re that to answer that, you need a lot of strategy, you need a lot of understanding of the process, you needed a lot of understanding of what data is available at that point in time. Because if you build a model with all of the data, you may be building a model that can’t be used at that point in time in the process. So it is a complicated problem to solve, which is why

 

08:11

companies have difficulty solving it right? If it were easy, it would be done already. Right? So the data scientists have to have either deep collaboration or deep knowledge of the business process or process ease. Yep, that the models will be used to support and augment. Yes, and oftentimes, when, when a project is done well, there’s more time spent with humans understanding the process, the business process or processes. They all merge together with much more time with the humans and they spend with data or code, which I think is totally different than the way that data science teams were a decade plus ago where it was, here’s the data, here’s a bunch of computers, we’ve given you a space, go and help solve her all of our business problems, let us know when you’re done. But it’s really a collaboration and data science for corporate use, leverages business strategy, subject matter experts within the business, computer science, statistics, and then Visualization and Communication. And I don’t know a person who exists that has all of that information in their brain. So it really is a team sport. That means a lot of collaboration. And collaboration with humans takes time. Our brains can only take so much in one meeting. Oftentimes, we’re just trying to document what the process is. Because usually the process isn’t documented. It just lives in so and so’s brain and you just need to go and talk to so and so until you figure it out. So we’ll meet with that person. We’ll write down the process and then we’ll come back a week later and say we wrote down the process is this right and

 

10:00

they’ll say, Well, it’s close. Okay, cool. So then we go back and forth, and we iterate on their own process and where we think the problem even is. And then okay, well, what data is available at this time? Do we need to? Do we need to start collecting data? Do we have the data that we need? Do we need to start engineering new variables or fields? Can we create them?

 

10:22

Looking back,

 

10:24

it is, you know, a big problem to ensure that you have the data that you want, that the model actually creates the results that you intend, and that you have actions after the model that create business value or revenue and profit. 

 

Tina: Can you give us an example maybe on your one of you in one of your path experiences? What does this look like in real life?

 

10:53

Anne Marie: You know, maybe an example of a business process. Yeah, I can give a few examples that

 

11:00

I think are really innovative. There’s one from Stitch Fix. And there was a Harvard Business Review article about it, how they created an AI algorithm to predict which customers would which like, which clothing items, but they found that if they just use the AI model, it would be there’d be, you know, if 10, or 20% of the recommendations are pandemonium, their customers are gonna go, why did you give me the pink and green polka dotted sweater? Right? You know, I only wear neutrals.

 

11:38

So they would have a human then review them. And that combination worked really well. But you have to in that case, you have to know when you’re going to run the model. Do you wait until you have so many iterations for our customer? Are they a repeat customer? You, there’s a lot of these business questions about how to leverage the model that need to get answered before you can put in place. And oftentimes, I’ve seen

 

12:07

data science innovation happens solely at the

 

12:13

model level, where it’s, we would like to build a neural network. So let’s build one, and it’ll be cool.

 

12:23

And the thing might work, you might be able to gather all the data, but can the business actually use it? And if the business used it? Would they actually have value? Because I’ve seen where models get created? And then they do the back testing, and they show? Well, if we use this, we would save $1,000? Oh, right. So really understanding whether or not that action after the fact has value. Because the model can predict really, really well. And you can still have a failed project. An example of that is there was a customer satisfaction model that was created on a team that I worked with. And they wanted to be able to call those customers and say, Hey, how’s it going? Is there anything we can do for you, and they called them and the customers were very angry, the model worked success, right? We found all the angry customers. And then they said, you know, what I’ve been meaning to call you and cancel all my products with you. As it turns out, you should just leave the angry customers on if you don’t have an action that leads to value. In this case, the action actually made it worse. So you can identify, you can predict something. But if you don’t know what you’re going to do with it,

 

13:46

then you again have this expensive trivia that you can’t actually leverage a little emotional intelligence maybe goes a long way, maybe, but also also just testing. I mean, making sure that you’re piloting in a way that is designed with hypotheses around what we think is going to happen.

 

14:08

What we think is going to stay the same sort of from a Do No Harm perspective. And that you do this in a small subgroup, you may not be applying it to everyone, because if you scale right away, then suddenly, people are leaving as customers. So coming at it from a, we can build a model. There’s going to be testing on the model itself about does the model function the way that we intend it to. But the real trick is actually a business problem. And I think this actually, is actually a real relief for

 

14:46

business leaders. Like pretend that the math doesn’t exist, like pretend that the math just works, pretend it works. 100% of the time, what are you going to do with it? You can answer those questions. If you know this. What are you going to do with it? You

 

15:00

If you call a customer, you know is angry. What do you think’s going to happen? Sales directors know that answer.

 

15:08

Data scientists don’t often know that answer. So really trying to think less about the model itself with the business leaders and think more about the process and say, if you had the perfect prediction, what would you do with it? How would you make money from it? I think they can answer that question really well. But then it’s creating that alignment between the business problems, the processes, the data, and the analytics is definitely a trick and a gap that’s been missing, I think, in the industry in various industries. It’s been tried to be filled by product owners,

 

15:53

strategy analysts,

 

15:56

and various other consultant style roles that I think are really helpful of being that sort of English to English translator between the business and the nerds and the nerds on the business. Is that the answer then, is to have a someone who speaks both? If that’s the answer is to have someone who speaks both and depending on the size of the team, and how embedded it is with within the organization. If it’s a centralized team, you’re probably going to need a team of consultants, right? If it’s an embedded distributed model for data science, then you probably have maybe the manager or the director of that area can can work that team with their business partners. So I think it really depends on the distribution model for the data science team. And what are some of the distribution models, the various distribution models that I’ve seen are fully centralized, fully distributed, and then and then a hybrid where you may have a centralized team that may take projects that are more complicated, or for any projects that were for a business area that doesn’t have a data science team. And then you may have some embedded teams, for areas that have enough work to really support at least a team of I’d say, probably want at least five, so that you have enough people for coverage for peer review coverage. And then you can align with sort of the Center of Excellence for for that hybrid model, on different standards for checking in code for review and peer review for starting a team for hiring, leveraging job description, banks, that sort of thing. Okay, so along functions, it could be organized.

 

17:53

For instance, you might have a team embedded within a function, or you could have a centralized shared or shared service centrally. Right. Or you could have a hybrid. Okay, and is that like in your you know, you’ve been in financial services, insurance, and now you’re in?

 

18:14

Tina: Well, what would you consider Zillow?

 

Anne Marie: I mean, like, some people would say real estate, but it seems more like it’s real estate data, not so much the real well, I guess it used to be Yeah. So I mean, how are there differences between industry as well? Okay, so answer the Zillow question, I would say it’s primarily real estate at this point.

 

18:33

And or at least in the home buying umbrella that includes that entire process, right, everything that you would need to close on a home. Can I just insert though that, I mean, I use Zillow, just like sort of as a voyeur. Like I like I just came home from Palm Springs. And I’m like driving around the neighborhoods going, Oh, I love this house. Oh, I love this. I wonder how much it costs. So I go on.

 

19:01

Tony Boyer anyway, sorry. That’s just my my love of Zillow coming out here. And certainly there’s very distinct personas of the folks that use Zillow, there’s the there’s the person that is just wants to look up the home value for their friend, they just had a dinner party with, right and they want to know how much their house is worth. Right.

 

19:26

That’s bad, right? I think it’s universal. So if, you know if it’s bad that we’re all bad together.

 

19:37

There’s some folks that are like thinking about it, right thinking about looking whether it’s their first home or their second home or they’re looking to downsize and then there’s the folks that are, like, serious right there. They’ve got the alerts. They know every single I remember when my brother was looking for a home, and my mom tried to send him links and he’s like, lady, we know them all. Up.

 

20:00

Have

 

20:03

you have those people, you’re those different people that are sort of along that like home buying journey, and then even after you buy the house, you’re like, alright, well, anytime somebody list in the neighborhood, I’m definitely gonna take, I’m gonna take a virtual peek. I mean, the old in the olden days that all you could do was go to the open house.

 

20:24

And my,

 

20:26

my parents live on a lake and my dad’s goal in his life is to see the inside of every single home. And now he doesn’t have to leave his couch to do that. So I, in my defense, I used Villo in all three scenarios. So just, you know, I’m not just voyeur, I have actually used it. for investment purposes, though.

 

20:47

Just to defend myself. I think that your question was around structures, or org structures for data science teams is a different like, what? What is a practical because I hear a lot about that. I worked in centralized teams, I’ve worked in

 

21:06

decent decentralized, where you have embedded data science teams, and maybe like a marketing org, versus an innovation area.

 

21:15

And then there’s also a hybrid model, I’ve seen work well, for automation teams, where you have that centralized team sort of for any business area that doesn’t have enough work to really like justify an entire team,

 

21:32

as well as maybe advanced use cases. And then you have embedded teams in business areas that can justify having a headcount of at least five. And that model we had the managers of all the embedded teams meet monthly, with the centralized team to it to ensure that we documented

 

21:59

the different processes that we would use for our peer review standpoint, from a hiring standpoint, what it meant to start up a team. What evaluating

 

22:10

tools and platforms, what are we using, so that there was some consistency across the teams and the enterprise? Our teams do different? Like, if you have an embedded team?

 

22:24

Tina: In let’s say, marketing versus fraud? Are there differences in

 

22:33

like domain expertise that they need to have? Like, you know, we, my company focuses on data security and privacy. So we find that in the customers, we’re companies we work with, there are differences in what a team in a different functional area would need to know just on a daily basis to do their job. Do you have you come across that in your different roles?

 

22:58

Anne Marie: Yes, certainly. And that’s the benefit of having an embedded team, right. They, they know the marketing really well, they know fraud really well. And when I’ve worked in centralized teams, oftentimes you will have, okay, that guy is the one that works with fraud, right? So you might have some people specialize a little bit on that, that’s also part of why for a centralized team, it’s really important to have those internal consultants, oftentimes, those are the folks that are specialized for a different business area. So they learn the business really well. And then they can explain it and get the what’s needed from the business perspective, translated to the data scientist who might work from area to area to area, because sometimes if it’s, if it’s a centralized team, and there’s a large, let’s say, it’s, let’s say, it’s a centralized team of 50. Plus, you’re gonna have folks that are while they make our neural networks, right, they’re very specialized about what types of models are built, the types of engineering that’s needed, you’re going to start seeing a lot of specialization or like that, if it’s a centralized team and a smaller company have less than 10, you’re gonna have more of the jack of all trades of people that can build the engineering pipeline, and talk with the sales leader about what the process is to build the model, that sort of thing. You might have some specialization between more than engineering function versus the model Build function. But you’re gonna have folks that can wear multiple hats, as opposed to a large, centralized team that will have folks who are very specialized. And so the specialization

 

24:47

I guess comes with

 

24:50

your experience, right? Because like for instance, if you if you land your first job in a centralized team, then maybe you’re more general and then if you land your first role

 

25:00

In a functional team or in a small company, then you’ll get that hands on experience in those those specializations? Yeah, I’d say it’s a function of experience and education. So oftentimes for the very specialized, they’re going to have a PhD. They’ve been kind of like this is their track there, as opposed to folks that are more general

 

25:30

than, you know, there are people that are very far along in their careers that are general because they like it.

 

25:37

Right, some of it’s just a personality standpoint, and they, when I’m coaching people for their careers, you know, or even interviewing folks, I’ll ask, what are the types of projects that you’d like to work on. And somebody might say, I just like to make neural networks every day. And I want to, that’s what I would like, I would like to not talk to anybody. And I would like to code all day, great, we have probably a very great role for you.

 

26:02

And then there’s also people that would say, I would get bored of doing the same thing over and over again, I want to have, I want to have options, I want to, I want to have

 

26:14

a variety. And I want to

 

26:19

I want to be able to meet different people learn about the business. All of those different things are aspects of sort of a generalist. In your, in your career. Are there any experiences or projects or initiatives that you’re most proud of that, you know, you would

 

26:43

you you draw on from that experience over and over again, you know, no matter what situations you or initiatives Do you find yourself in? Or is as there have there been any experiences like that? Yes. And I would say that some of them are that I draw on are sometimes more of the failures than the successes. Do tell.

 

27:06

I think

 

27:09

there have definitely been times and I think this goes back to what we were talking about earlier, where

 

27:18

myself and the rest of the team got sucked in by here’s the cool thing we’re going to make, rather than how will it actually impact the business?

 

27:28

Right. And when you sort of put your heart and soul into a work, baby, that is, you know, sitting in,

 

27:39

I don’t know, maybe they scrubbed it and deleted it by now.

 

27:43

But doesn’t get used, right? It doesn’t feel great. And I think that, especially for

 

27:51

data scientists that enter the corporate world, they want to they do want to make an impact, they’ve decided that they don’t want to write white papers, they’ve decided that just moving the science isn’t along, they want to see the impact of the things that they built. And so

 

28:08

building something and getting it all the way to the end, and then hearing Yeah, we can actually use this. You know, think about the the team that built the customer satisfaction model. They got it worked, right? Yeah, the model worked for the AC was fantastic, right? They found the customers who were angry, but it actually hurt the business.

 

28:34

So thinking through times, and that’s why I’m such a big advocate for understanding where in the process understanding what action is going to have a business effect, what is the financial value of this model. And if there isn’t one, let’s not put our hearts into it, let’s do something else. There are so many problems to solve. There’s no lack of problems to solve. Let’s make sure that our efforts our time or energy, our mental strain is put towards things that is going to have a business value. And then like sort of bonus points often sometimes there are weird, weird Venn diagrams where you can actually find dollars, but from a revenue or profitability standpoint, and you can do a thing that drives money. But if those monies, those dollars aren’t assigned are tied to some sort of broader strategic initiative that the executive can say, here’s the thing that we said we would improve and we improved it. If you’ve just now found something over here on Pluto that you improved, even if you found 10 million bucks, kind of not going to care. So it’s really important to have those folks who are connected to the cheer the strategies, and we’re going to move dollars that help those strategies because if it’s over here, and left field

 

30:00

Just down Pluto, again, you may have found 10 million bucks, and they’re gonna go, that was great. Thanks for that. But we also need to solve all of these things right on, right in the day, your executive, your exec has gone before the CEO on the board and said, here’s the things we’re going to do. So if your project doesn’t impact, hear the things that we did, from the things I said we were going to do, then you’re you’re less likely to get resources, implementation dollars, all the things that you need to actually test and implement and execute on any other project. So it’s not just about what I’m hearing is that there’s some politics involved in that if the campaign is on Mars, be sure to tie your work to Mars, not to Pluto. Because Pluto is not this election, right? That’s not not what we’re talking about.

 

30:57

Yeah, okay. If the executive is announced that we’re going to Mars and you’ve gone to Pluto.

 

31:08

To be mad, they’ll be like, but I said, we were gonna go to Mars, what’s, what am I supposed to do with that? And you’ll be like, but it made $10 million.

 

31:15

Cool.

 

31:17

So it is important to not just think about business value from a from $1 standpoint, but also think about alignment to strategic initiatives that your boss’s boss’s boss has promised the boss’s boss’s boss’s boss. This is feeling very personal now. Sorry.

 

31:35

Tina: Is it personal for you? I mean, this is sounds like maybe you went through this before? 

 

Anne Maries: Oh, definitely. I think, you know, starting in

 

31:46

early career versus, you know, mid career, you, hopefully we’ve learned some things. And

 

31:55

 

31:57

came from a background of my parents were both educators, right? And so they just, they would come to work each fall and teach the kids that they were given and then summer would happen and we water ski for a while. And then they would do the whole thing again, right.

 

32:12

And I didn’t really have any sort of background in office politics.

 

32:18

Like those sorts of how do you move up a career within a corporation. So that was definitely an area that I had to lean into and and learn a lot about, because my thoughts were? Well, I learned a lot in undergrad and grad school about crunching numbers, and I should be able to crunch numbers and have successes with those numbers. And then career

 

32:45

has, right.

 

32:48

But unfortunately, that’s not that’s not how it works, right? The things that you do in

 

32:56

a corporate job, need to have impact for specific areas, in order to have career success, to ensure that when you’re

 

33:12

when you hear about a new strategic initiative, right, these things like digital transformation, right, like, this is the thing that we’re going to try to do

 

33:23

that you pay attention. And if you’re assigned to a specific business area, that it’s not just, you know, there’s there’s one part of paying attention during the big meetings, or the executives talk about what we’re going to try to do. And then next year, which are probably those meetings are happening right now. Right? This is q4, everyone’s talking about wrapping up the year strong, and then doing great things in the next year, we’ll find out what those great things are. And then bonus points if you have coffee, or digital coffee, with the folks who are close to those initiatives, right? Maybe, yeah, you can always be that person that you know, cold emails, they exec and see if it can actually happen. But you know, getting closer and closer to that person and having coffee, people love talking about themselves and talking about their jobs. Right. That’s, that’s why podcasts exist.

 

34:14

And having coffee with someone and saying, Alright, so I heard that we’re going to have this new strategic initiative. What’s the real story? He will just tell you.

 

34:23

Like, they’ll tell you, you know, what was said in the meeting in front of everybody? And then maybe what the real story is behind the scenes, and then who’s trying to influence what behind the scenes like, you know, get the tea, a little bit of what’s going on with those strategic initiatives, which are the ones that are actually important, which are the ones

 

34:47

that are the most likely to be wins, which are the ones where the executive is struggling and who’s sort of championing what that sort of politics layer to it, right. So it

 

35:00

If I know that Tina really hates initiative three, I’m not gonna go, Hey, Tina, let’s try to fix initiative three, you’d be like, I don’t want to meet with you. So to know, sort of who’s the champion for which program, if you then email that person with that thing and say, I’d like to meet with you about this, I think I have an idea. They’re more likely to listen.

 

35:21

And so building building projects, and at least thinking through projects, around that lens of alignment to strategic initiatives, and then building business value through building models that have actions that drive revenue or profit margin, I think it has been well is a very good combination, but that that combination takes a combination of brain power, that is business strategy, that is understanding data science, enough to know what is possible at this, you know, point in, in the universe?

 

36:04

And understand the data enough to know do we actually have data to do build the model to impact the thing that they all want to impact? What we’re describing? Is this all one person? Or is it really maybe a team of people, some people who are better at the discovery exploration within the business owners?

 

36:29

Versus the data, maybe engineering side? And then the actual model building? Or is does that person exist? Is that like a superhero data scientist, or they’re the superhero data scientist doesn’t exist. It’s a, it’s a fantasy team. It’s really data science is a team sport. And oftentimes, what I’ll say to Business subject matter experts that in draw the short straw and have to talk to the nerds for some project,

 

37:01

you know, we’ll talk about how understanding the process is really important for this understanding their day to day frustrations, what’s easy, what’s difficult, you know, what do they need, they’re a part of the team now. And I’ll say grass, your data scientists added to your LinkedIn profile, like you’re in will teach you some of the basics about what we need from, like training and testing data, and what the output would look like and what you can do with it. But what we really need from you is to understand the process to understand the business problems, to understand where your efforts are not getting you far enough. And then working with the business leaders to know how all that granularity, all the processes line up to those strategic initiatives and challenges.

 

37:51

There are definitely a lot of roles in this space. And I think that it goes back to that conversation around career coaching and like, what do you like to do every day? Do you like to get into a room and code all day? Great, there’s jobs for you. If you’d like to get into virtual meetings and talk to people about these sorts of problems? Great. There’s a role for you. Are you more interested in the data side of things and building data pipelines? And understanding the data so that it can go into a model? Fantastic. Do you like actually building the models themselves? And maybe the like, what actually is, you know, the data? Is the data perfect? Is the data Correct? Maybe that’s sort of less exciting to you. So there’s all these different sort of roles that are

 

38:39

being renamed nearly every, I don’t know, three, five years. But they’re all really, really important. And I think that early career, it’s important to try out a lot of things to fail to figure out what your sort of work superpower is. And then you can start honing in on one specialization, and that specialization may be a generalist, right? Like that may be I just, I like to build data science models at a small company that has embedded teams, and to understand that sort of like, what do you like to do day to day, who do you like to work with, can help answer your own personal questions about what are the types of companies types of teams that you want to pursue your career with? That’s great, really great advice.

 

39:27

And also, I mean, it for the data scientist practitioner, that’s great advice, but it’s also a great advice. I feel like for companies who are trying to, you know, figure out how to get the most out of their data science teams, because there’s been a lot you know, in, in, in the industry going around, or the market about how they’re struggling to find value. So this is, you know, I really feel like this is valuable for our listeners. I wondered if you had any takeaways or parting

 

40:00

Words of Wisdom. Well, I definitely agree with that sentiment about corporations struggling.

 

40:08

I think that

 

40:10

the hiring and retention is a real struggle right now for most companies. But I do think that a really important piece for both the hiring and the retention is to have a career path. And that’s more than just well, we have these five job descriptions, good luck,

 

40:34

is also about having a having career conversations that are planned and sincere, and really are trying to encourage people to grow their careers within your company. And so if somebody doesn’t, isn’t really excited about being on the team that you that they’re on, they’re not going to be doing their best job. So I will frequently talk about myself as I am just a really selfish, greedy manager, and that I want the people on my team to be really jazzed about the roles that they’re in, because they’ll just unlock more of their brain, they’ll be happier to open up their laptop at the beginning of the day, they’ll have here come to the virtual office, they might think about work in a positive light, when they’re out on a run or running their errands and go, Oh, my goodness, that’s how I solved the problem, right? Because having good work life experience, leads to better results. And so, for my team, we have quarterly IDP reviews, so individual development plans with just a very simple template. It’s not meant to be an essay contest. It’s not meant to be a huge amount of homework. It is meant to be a guide for their career that involves continuous learning and check ins, because if you don’t set up the check in you don’t do it.

 

41:57

And so we start with what is your next step in your career? If there were a job posting that said this, you would definitely want to apply for it, but maybe there are some gaps. And then we say, Okay, well, what are the gaps from the job description versus your resume? And then how do we

 

42:17

fill those gaps with formal learning, mentorship, and on the job project learning?

 

42:25

And then we check in on that quarterly? And we say, How have we done? Is this still the thing that you want to do? Because maybe that’s changed. Maybe after doing all this other work? You say, no, no, no, project management is not for me, thank you very much.

 

42:36

But maybe it’s still the thing that you want to do. And then we say, Okay, what things what would you do have enough time and space in your day, and week, and the last three months to actually do some of the things that we itemize, so maybe you did great, maybe we need to make more space for that bull. Or maybe we need to add, maybe you’ve knocked it all out, you just went nuts, and you read all the books you met with all the people who did all the project work. Now we need to add more things to that bucket. So I think that having those, that career planning process within your teams, is something that helps to retain people, it helps to have them build their career at your company so that they that cumulated knowledge is retained. And when you’re interviewing folks, it is such a huge add to have them know that our Career Planning isn’t just we have a series of job descriptions, it is something that I’m committed as a manager to do with you side by side. And if that means that you interviewed a guy on another team, that I am happy to help support you to help find out who that manager is to help find out and prepare you for the interview and be along the way for that journey. Because keeping people within the company, keeping good experience knowledgeable people at the company and building their careers is better for the company. And who knows, maybe that person might boomerang and come back, then we have an advocate for data science in another area. It’s only upside. So it’s the gift that keeps giving. Exactly. And and by having that process in place, you can attract more people that are interested in building their careers, because those are the people that you want on your team. So this is very helpful. And I really appreciate that you spent the time with us today to share these best practices in building, developing and leading data science teams and how they can provide value for business processes and initiatives. 

 

Tina: Thank you Anne Marie, and we’ll have drinks soon. 

 

Anne Marie: Sounds great, Tina, looking forward to it.

 

44:43

No matter where you are in your data journey. privitar is here to help. privitar empowers organizations to leverage their data to innovate faster, while protecting the privacy of individuals at massive scale. privitar is unique in combining technology and

 

45:00

Thought Leadership and expert services to help your data operations thrive.

 

45:05

Want to learn more? Our team of experts is ready to answer your questions and discuss how data privacy can fuel your business. Visit privitar.com Thanks for listening to incompetence brought to you by privitar. To hear more insights and advice on how to effectively use, manage and protect your data, subscribe to the show and your favorite podcast player. If you liked the show, leave us a rating. Join us for the next data conversation.