Nick Pylypiw @ Cape Fear Collective | Building up trust with stakeholders | Data Science Hangout

Transcript#

This transcript was generated automatically and may contain errors.

Welcome to the Data Science Hangout. Hope everybody's having a great week. If we haven't had the chance to meet at a hangout before, I'm Rachel. It's so nice to meet you. Let us know if it's your first time joining if you want to say hello in the chat so we can all welcome you in as well. This is our open space to chat about data science leadership, questions you're facing and getting to hear about what's going on in the world of data across different industries. And so if you haven't been here before, every week we feature a different data science leader as my co host here to help lead our discussion and answer questions from you all.

So we are all dedicated to making this a welcoming environment for everybody and love to hear from everyone no matter your level of experience or background or area of work. It is also absolutely okay to just listen in if you want to. But there's also three ways you can ask questions or provide your perspective too. So one, you can always jump in by raising your hand on zoom and I'll be watching out for that. Two, you can put questions into the zoom chat and feel free to put a little star next to it if you want me to read it. And I can call on you to introduce yourself and add some context otherwise. And then third, we have a Slido link where you can ask questions anonymously.

And so we do share the recordings of each session to the Posit YouTube. So you can always go back and rewatch or share with a friend. But with all of that, welcome, everybody. Thank you so much for joining us. And thank you so much, Nick, for joining us the co host today. Nick Pilipiu is Chief Data Officer at Cape Fear Collective. And Nick, I'd love to have you introduce yourself and maybe start off by sharing a little bit about your role and also something you like to do outside of work.

Sure. So my name is Nick Pilipiu. I'm the Chief Data Officer at Cape Fear Collective. We are a Cape Fear movie. People are always like, is it Cape Fear like the movie? It is based on Cape Fear regions, the kind of southeastern North Carolina region, kind of Wilmington, if you're familiar with that area. We are a nonprofit based out of Wilmington that's really focused on the state of North Carolina now. And we're really looking to use data to move the needle on on these kind of social impact issues, social determinants of health.

My background is more in private sector data science consulting. So I worked for a couple of consultants and was doing work with Southwest Airlines and Procter & Gamble and Lowe's and Baltimore Ravens and Cincinnati Reds and a bunch of different organizations and really looked at this as an opportunity to kind of make an impact in the state that I'm from and that I love. Some things that I do outside data is just kind of building stuff in my yard. I'm building a porch now, a screened in porch in the back of my house. I'm into music and surfing and just kind of hanging out here. And I was still North Carolina.

people like data and analytics if it confirms their preconceived biases and beliefs. But if it counters that, then that's where they start poking holes and wanting to, you know, kind of find fault in it.

So one thing that I've been very successful in my career is kind of starting to build that trust. Find some quick wins. What are some small projects we can work on together that are low stakes projects that kind of gets people to buy into and say like, okay, now I can see why there's value in this way of thinking. So that when that inevitable, yeah, exactly. Don't start with forecasting. That's good advice. When that inevitable cognitive dissonance comes, then you've built up some trust and some capital with that partner.

Corporate vs. nonprofit data science

Patrick, I see you have your hand raised. Do you want to jump in? Thanks, Rachel. And yeah, good morning, everybody. Happy to chat with you again. Nick, you and I spoke a couple of years ago. You may not remember, but you were helpful on a couple of things for me. And so I appreciate that and glad to see you again. I'm wondering about the comparison and contrast, you know, similarities as well of working in those big corporate settings versus where you are now and nonprofit and, you know, social services and that kind of thing. Personally, I've been in academic and nonprofit settings my whole career. And so it seems to me, it seems appealing that, you know, people really have their things together in a corporate setting and everything is well-funded and runs smoothly. And I assume you can dispel that myth for me, but I'm just kind of wondering about your thoughts on that.

Yeah, absolutely. And I think you nailed part of it and Marlene nailed a part of it here too, because there's certainly this, it is a myth that these large companies really have everything all put together because the reality is I saw some things in every, all these Fortune 500 companies I've worked in, I saw things that would make all of you cringe. And I'm probably not going to say any of those because I don't want to get myself in trouble, but things that it's just like, I can't believe you run your stack that way. That is not the way a $80 billion company should be running that. This file should not reside on one person's laptop, right? That should be in a shared platform locked down, et cetera.

One of the biggest differences though, for me, is in large companies, people can hide, you know, like there's, there's kind of, there's room, your errors and your kind of weak links on your teams don't kind of expose themselves as readily as on small teams. On some of the small teams I've been on, my previous company was 35 people and currently at K4 Collective, we're a team of 12 and there's no room for somebody to just kind of not, not be pulling their weight.

Another big difference is I think a lot of the Fortune 500 companies, whether or not they're actually implementing it is another thing, but they at least know that they should be. They've heard enough times through all the articles and through the conferences and through their leadership, like you need to be doing things, data, KPIs, measurements, dashboards, et cetera. They know they need to be working on it. There's some pressure to kind of do that. Whereas in the social space, a lot of the stuff's very new and it's a dramatic cultural shift. A lot of the nonprofits that we work with are used to being able to just, and I'm going to kind of say this kind of tough, you know, for a minute, they're used to kind of just patting themselves on the back and saying like we did good work and that's good enough. And the thing is like for the funders and the philanthropic organizations and grants that are funding them, it's not good enough anymore. You have to be able to show us that you're moving a needle, which means you have to set some KPIs, measure them, collect data, and show that you're making impact against that or else you're not going to qualify for that next round of funding. We can't just do feel good funding anymore.

Getting your first data science role

I see a few anonymous questions coming in on Slido and one was, can you talk a little bit about getting your first role in data science? Any tips for jumping from data analytics to data science?

Sure. And this whole analyst versus data science thing is such a weird, ambiguous thing anyways. If you ask, you know, there's people that I know who call themselves analysts or titles analysts and they're programming in Python and R every day. And it's like, that's a data scientist to me. And then there's other people who are like, well, I'm a, I'm an analyst. And it's like, well, are you a business analyst or are you a, you know, like there's, so titles kind of all get kind of, get kind of wonky anyways.

But for me, you know, my getting my first role, I really wanted to be somewhere where I had other folks who could challenge me that I could learn from. I think a mistake that I see in a lot of young devs and data scientists is they go right to a startup or right to a small company. There's nothing wrong with either of those. I've done both, but you're on an Island. You don't have the support that you need. But a lot of these larger companies have support structures in place. There's going to be senior data scientists, senior analysts who can help you and know the data sets and can, you know, so find one of those large companies with a solid structure, latch onto a couple senior devs and be a sponge, right? Just absorb everything that you can.

And then the other thing that I can provide that I always tell people who are trying to break into the spaces, nobody likes somebody who pretends like they know everything because none of us do, right? Like there's power in saying, I don't know. And if you apply for a job on my team and say that you know how to do natural language processing and modeling, and I put you on a project that requires that skill set, and it's obvious that you don't, I'll know like it's that you're not fooling any, you're only fooling somebody until it's time for you to use that skill. So be honest about what you know and how good you are at it. If you say you're a 10 in R, I don't even consider myself a 10 in R. So you better be really good if you say that.

And Marlene, that point about like, every time I think I'm an expert, I think there's a graph that shows that kind of like, the more you learn about something, and then you hit a point where you're just like, like, there've been times in my career where I'm like, I'm an expert Excel user. And then I'll meet somebody that I work with. I'm like, wow, that I, I'm like a two. And then I'll be like, okay, I've learned now I'm like a seven or eight, I'll meet someone else. I'm like, four, you know, so definitely, there's always someone out there who's better at it than you.

Navigating resistance to new tools

Another anonymous question was any advice for someone working for a large corporation that's very resistant to new tools to progress in the data science world?

Yeah. Brush up your resume? No. I think it's a, it's a hard thing moving. Someone else made the point earlier that, you know, big ships turn really slowly or something like that, you know, and it's, it's true. And yeah, I think other than just kind of keep, keep beating the drum and keep trying to prove value. Unfortunately, a lot of these, these companies just sometimes, it's going to take way more than you to get it. So what I do, or what I recommend to people is when you're interviewing for that job, make sure the questions that you ask are kind of hinting at what that data culture is. One of the questions that I think is a really good question to ask is what is a model or a process that your team has built that is currently being used for the company to make decisions on?

And you'll be really surprised to hear the wide variety of answers that question gets, because it's really obvious to hear someone's kind of like, like fumbling, like, oh, we built this thing, but it never got implemented because of whatever, or we've been working on this thing and for a couple of years and it's never quite gotten the traction that it needs. And there's going to be other teams that are like, oh, we've got this, this thing that's helping with our, you know, search engine optimization. And we've got this other model that's out there that helps determine, you know, where we build new locations. And what you hear there is this is an organization that is putting their money where their mouth is. They're not only hiring people to build these models, but they're making business decisions based off that.

The community data platform

Just to give us a little bit more context to you on some of the work that your team does today, I thought it might be helpful to hear a little bit about the community data platform or maybe like an example of a problem your team is solving.

Sure. So one of the problems that we hear a lot, or one of the kind of issues that we hear a lot is there's, there's not data for that. You know, you have, so our organization's really working to build data capacity across these nonprofits. There's 1200 nonprofits in New Hanover County and there's no shortage of passion there. They all think they're doing good work. And they are, but there's no real data capacity to understand, you know, what's the value in generating a hypothesis and measuring data, collecting data.

So what we keep, what we always hear is, let's say, well, hey, you've been doing this thing for 30 years. How's it going? How's it working? And they're like, well, you know, people really like our food boxes or whatever. It's like, okay, that's cool. Is it actually like moving the needle on food insecurity or, or, or hunger or health issues in your community? They're like, well, there's not really data for that. I'm like, let's see about that.

So one of the first things we did is we pulled in Census Bureau data, Bureau of Labor Statistics, CDC, FHFA, and all these publicly available data sources that for us as data scientists are really easy to grab because we know how to use an API to go do this. We know how to scrape a website. We know how to search Google specific terms that are going to lead us more successful results. But for the average nonprofit community leader, they don't know how to use an API. They don't use Python. They don't know any of this stuff, right? So for the first time ever in our community, we really have this holistic 360 degree view of what a community looks like.

And the way we're using that is to do things like you've got a nonprofit who wants to do something. We're going to find a metric that works for you and help you build a mission around that. We are going to decrease children in concentrated poverty in this neighborhood by 5% over the next 10 years. And we need $200,000 to do that or whatever, right? Like that is a statement that is impossible to make without understanding how many people currently live in that situation, what the current projection is for that neighborhood, and kind of like what a reasonable decrease is for that metric.

So a lot of our work is really kind of around that, but there's some other larger projects. One that I'll talk about is this awesome video that we just put out. It kind of walks through this, but the Michael Jordan Foundation just put these two clinics in Wilmington. Michael Jordan's from Wilmington, and he wanted to put them in areas where they're going to make the most impact for low-income, marginalized populations. The hospital system said, that's great. We know a lot about chronic disease and we know a lot about health conditions. We know very little about the kind of social determinants of health that you've been studying. So help us figure out from a data perspective, specifically where these clinics are going to make the most impact. So that's kind of our flagship project and one that we're really proud of.

Measuring outcomes in the nonprofit sector

Morris, I see you had asked a question in the chat a bit earlier. Do you want to jump in next? Yeah, thanks. No, I think you just answered it. I'm working in a non-profit and I'm a social worker. Transformation is always really the big question. How do we measure that? That's what we want. We're pretty good at tracking outputs, but the outcomes and it's a little bit more difficult. But these are helpful. Thank you.

Yeah, and that's actually a question that we struggle with a lot too. So I'm glad you brought that up because a lot of times it's not fair to say, hey, food bank, it's on you, Food Bank of New Hanover County, to turn this area into a non-food desert by yourself, right? And not only that, but the lag in data is such that the USDA only classifies these food deserts every five years. So it might be five, 10 years before we even know if you've moved the needle on this kind of macro metric. So what we try to do is we try to tie these kinds of micro data metrics to those leading indicators. So we might not know the Census Bureau data for another two years because of the lag, but what we do have is this real-time emergency department data or real-time police data that we have for our region. And if we can find the correlation between those things, then what we can do is we can say, we can see a signal here at a right now level that we know is tied to this kind of macro level. And then we can kind of hypothesize that you're moving the needle at this greater level that you haven't been able to measure before.

Another question that was asked anonymously a bit earlier was, how do nonprofits think about the substantial costs of a data team? Are these getting more common with the trend of needing to prove moving the needle?

Yeah, I mean, the reality is it's unattainable for the average nonprofit to have a data scientist on staff. But that's okay because that's one of the things that we're hoping to do because part of our business model is we don't need athletic charity missions of Wilmington, North Carolina. It doesn't need a full-time data scientist. All they really need is 40 hours with somebody on my team to kind of help kickstart the way they think about data. So, instead of saying, hey, every nonprofit in North Carolina needs to go hire a data scientist, it's how do we kind of become that knowledge hub that kind of through contracts, through pro bono contracts. Over the last three years, we've done $2.8 million in pro bono data science consulting. So, what that looks like is somebody on my team sitting right next to a nonprofit looking at their Salesforce or looking at their Excel or downloading their data and saying, hey, did you know that 15% of your clients are from this neighborhood, but that actually the majority of the need in our county is over here? And they're like, hmm, never really thought about that.

Working with multiple data sets

Yeah. I mean, if I wish I had a better answer for it, but the reality is I just started grabbing stuff. You know, I like to say I've never met a data set in this space that I didn't like. And that's still true, even though we've met some garbage data sets, we still bring it all in because we think that the kind of the triangulation and the whole combination of all these different data sets really adds to the story.

And I'll give you a kind of an example of this. We're talking to a hospital in New Hanover County, and we had a hospital who was looking at doing some outreach for COVID testing. And they're looking at kind of where do we need to put the Spanish flyers, right? We've got these English flyers, we've got TV ads, we've got signs, where do we need to put flyers that are written in Spanish? And, you know, we told them, well, you need to put them here. And they're like, well, based on the Census Bureau data, only like, you know, I'm making the numbers up by camera off the top of my head, but only like 5% of the people there are Hispanic or English second language or whatever.

And it's like, okay, that might be true. But for reasons that we don't have to discuss in this call, a lot of times non-English speakers are de-incentivized to fill out the Census Bureau. And actually there's a fear factor there and everything else. So the Census Bureau historically undercounts them, number one. And then number two, based on your data, you have a large portion of Spanish speaking patients who live in this area. So there's a mismatch between the Census Bureau data and the hospital data. Who's right, who's wrong? Don't know, but it's like, if you only had one of those data sets, you might make a decision that's not actually the right one. So by having them both, you have a more complete version of the story.

Regional focus and building community trust

Yeah, it was really natural for us. We had some really early success in the first year where people in New Hanover County and that Cape Fear region were like, we love this, we love it. We started seeing like, all of a sudden people in Raleigh are asking us and people in Charlotte are asking us. And we're like, are we going to go statewide? Are we going to go regional or are we going to be like Eastern Sea? When does this stop? And it kind of auto corrected itself because we kind of got to the mountains of North Carolina and people up there all of a sudden were just like, I don't care where you're from in Eastern North Carolina. You don't know anything about us up here, right? So there still has to be this kind of boots on the ground connection to the region or else for the same reasons we talked about building trust and capital in a Fortune 500 company.

If you think about the state in that same way, if you're from the beach town and you try to go up into the mountains and tell them what they don't know about their public health crises, you're going to get run out of there with a stick, right? So I mean, you still have to build these connections organically. So that's kind of limited our business plan in a way that I think makes sense, not in a bad way like, oh, we can't do that. Now the other thing is we've made it very clear that we want to work kind of by, with, and through and not around. So areas that already have, if there's an area in North Carolina that has an organization that does what we do, we definitely don't want to go in there and say like we're going to replace you, right? It's how do we work together? How do we start sharing some tools on some of this work?

Because we understand that you're here and we can't, there's plenty of work for all of us anyways. So we don't really have our sights set on any sort of, there's so much work in Eastern North Carolina. Here's a stat for you. If you carved North Carolina in half down highway I-95 and you just looked at the East part of it and made that a state, it would be the poorest state in the country. So no Raleigh, no Charlotte, no Winston Salem, no Greensboro. If you just looked at East part of the state, it would be the poorest state in the country. So, I mean, that's the level of challenge that we're working with.

If you carved North Carolina in half down highway I-95 and you just looked at the East part of it and made that a state, it would be the poorest state in the country.