The power of a multi-platform stack: CoverMyMeds’ approach to data

Tarush Aggarwal (00:00)

Welcome to another episode of People of Data, where we get to highlight the wonderful people of data. These are our data leaders from various industries, geographies, and disciplines, discover how companies are using data, their achievements, and the challenges they face. Today on the show, we're thrilled to have Scott Sacha the Vice President of Data Science at CoverMyMeds, Scott's leading the charge in AI and machine learning innovation.

‍

With over 20 years of experience in analytics, reporting, and executive management, Scott is known for his strategic vision, leadership, in developing cutting -edge solutions, and a knack for building strong collaborative teams. Welcome to the show, Scott.

‍

Scott Sacha (00:39)

Thank you. Thank you for having me today. I appreciate it. I'm excited. I am. totally real. Let's go.

‍

Tarush Aggarwal (00:42)

Are you ready to get into it?

‍

Let's do it. What does CoverMyMeds do? What do you do as a business? How do you make money? And the question on everyone's mind is how does data help CoverMyMeds play a role in doing what you want to do?

‍

Scott Sacha (00:52)

Yeah, so.

‍

So CoverMyMeds is a McKesson company. So it's a large company, but CoverMyMeds kind of is technology unit of McKesson. CoverMyMeds does is we act as, where our goal is, it is to get medication to people that need it and for affordable prices and make access easier. So we own a bunch of different networks is what we do. So we own the connection between the pharmacy and the payers.

‍

We own it between the doctors and the pharmacies. So we have all these different connections. And basically what it is, is data is flowing back and forth on those. So we run roughly, I mean, over 22 billion transactions just on the pharmacy side of it, not counting our doctor side, another one. it's a great for a data person because there's so much data and so many things you can do with that data. at the end of the day, what we're trying to do is help people.

‍

We work with pharma manufacturers so we can make drugs cheaper for people. We work with pharma manufacturers to help prior authorization. So like a lot of times when you get drugs, you have to get a PA for it and it's challenging. We try to make those steps more automated and easier to get through. And there's a lot of different products we have on top of that. But at end of the day, what we're really doing is we're manipulating data in a transaction and moving it back and forth. So it's a software based company is really what our division does.

‍

Tarush Aggarwal (02:20)

Yeah, how do you guys make money?

‍

Scott Sacha (02:23)

So we make money through many different channels. So for CoverMyMeds, we make money from the pharma manufacturers. So they pay us to help them facilitate to their customers. we also make money from pharmacies our network to switch their transactions back and forth and connect them. And then we have several other like.

‍

different ways to get money, but those would be the main ways is and then we add products on top of our networks. So like, for instance, like if you go to a drug store and your copay were to come up at like $100, we would do my team would do analysis say like nobody's gonna pay $100 for this drug like, or this pocket of people aren't going to pay $100 for this drug. The pharma manufacturer gives us money to reduce that copayment down to a lower amount. So

‍

that money helps to make it so you can cover more of a population because you're not going out and finding the card on the internet or things that, you know, there's cash cards and things like that you can do. But this just happens instantaneously within the transaction.

‍

Tarush Aggarwal (03:24)

Very interesting. I think my next question is very obvious, just given everything you've said, but to ask it explicitly anyway, how is data used on a daily basis? I think what's really interesting is, apart from all of the usual data analytics internally, data is one of your core products. And as a data leader, I'm sure that's...

‍

Scott Sacha (03:35)

Yeah

‍

Yes.

‍

Tarush Aggarwal (03:45)

extremely exciting. So we'd love to hear a little bit of how you see this.

‍

Scott Sacha (03:49)

Yes.

‍

So it's a kind of interesting thing because I started at CoverMyMeds 11 years ago, almost 12, and I was the first data scientist at that company, which was weird to me because it is a data company. I think we've kind of had a little bit of a identity change. Like originally, we were much more of a product -based company.

‍

Now I think we kind of are a data technology software company more. how we use that data is in many, many different ways. Obviously, we're trying to predict how patients will perceive their payments, how we help if a rejection is going to come for some reason at a pharmacy, how we can help them get past those rejections. We're trying to figure out if they submit a prior auth, will it get approved?

‍

How can we help the doctor fill in those forms using AI and ML to help automate those forms and read in things? And then because we have so many different connections to data, also pulling in from the, we get rejections, we can read those rejections and figure specifically why that's happening. And maybe there's a pattern to it that we can recognize and manipulate that to help the person get the drug.

‍

Tarush Aggarwal (05:06)

Yeah.

‍

Scott Sacha (05:06)

and

‍

manipulate it when I mean, I don't mean that in like a nefarious way. I mean, it more or less, you know, just manipulate the situation to make it better. it's, it's super interesting, just from the aspect that there is so much data. But it's also very interesting, because it's very regulated. So we have a lot of government regulations we have to deal with, we have a lot of data rights in compliance contracts,

‍

aspects of it make the job more difficult than probably most would expect.

‍

Tarush Aggarwal (05:35)

Yeah, totally. Every time governance and especially health care data gets involved, there's so many layers which are non -trivial. What's the company culture like?

‍

Scott Sacha (05:46)

So it's interesting because McKesson is obviously a huge company, but CoverMyMeds started more as like a startup it holds that a little bit more. was able to build my group from scratch and just kind of started the way I wanted because we that culture. Like it's very, I can talk to any level, C level.

‍

any level of person at the company, no problem. It's easy to communicate with each other. it's a, and we're based in Columbus, Ohio. We have a really cool building there. It's actually really cool. have like themed rooms. You have like a spaceship room, a old lounge. So it's very startup culture -esque.

‍

of the things I've always loved about there is I've worked in a lot of other places where when a problem arises, a lot of times people panic and freak out and do that. They don't do that here. We systematically and logically solve problems. And I think that's, that's a, it's a joy to have that kind of and team around you.

‍

Tarush Aggarwal (06:41)

Yeah, that's really, I love the concept of the space room. It's so fun.

‍

Scott Sacha (06:51)

Yeah, it's cool. They have a lot of cool rooms. have they also have a room that's like a like a speakeasy and it's kind of behind like this thing and you open a refrigerator and it goes in there and it's like all these leather chairs and stuff. It's really cool.

‍

Tarush Aggarwal (07:03)

I remember at WeWork, at HQ, we would have a music room, which was really popular one. There were instruments all over. What's the culture like on the data team? how does, very often, I think some data teams have this microculture brewing over there, just given that it's so cross -functional. I would love to get some thoughts over there.

‍

Scott Sacha (07:09)

Yeah, that's a good one.

‍

Yeah. So

‍

for me, I've always kind of prided myself on, I started as, you know, 20 years ago doing data before data scientists even existed. So I kind of evolved into a data scientist. I wasn't just given it. So I've sat the one cool thing for me, I feel like as I sat in every role that everybody that works for me is in at some point or another, I had that job. So I, I understand that role. And I think our culture is very

‍

We're very innovative. We a lot towards trying to come up with new ideas. We're good at also working as a community and rejecting each other's ideas and being able to accept that and learn from each other. Because always believed that bad ideas are good ideas. Because what happens is a bad idea, if you have the right people in the room, can turn into a good idea.

‍

would say one thing that I've always thought is interesting, managing data people is different than managing other types of people because they're very intellectual, they're very smart, they're usually very motivated and high functioning people. So I don't really, you know, you don't micromanage these people, you just direct and guide and bring in the work and fly and cover, play coverage for your people, make sure that, you know, they're not dealing with

‍

all the politics and stuff like that, because that doesn't bring value to the team. that's my job to cover on that stuff at this point. So think we have a really good culture. I think that honesty and transparency is the key thing. I think you can't fool a data scientist. You can't fool people like that. So they can read you better than you think they can. So you just have to be honest with them.

‍

I think one thing that's funny for me, I always say like, I'm honest to a fault. Like I don't even think I know how to lie at this point in my life.

‍

Tarush Aggarwal (09:07)

Yeah,

‍

I love it. Honestly, the best policy. what's the company footprint like? How big is the company? How big is the data team in the company?

‍

Scott Sacha (09:14)

the company like CoverMyMeds is about 5 ,000 people and the data team is about 200. would say it's actually fairly small in comparison to what we do in the reality of it. do a lot with very little, which is good and bad, right? Like you don't really have

‍

We have a lot of points of where you just don't have as much redundancy as you kind of wish you did at times because you're not bringing people up as fast as you can, you know, and so that's kind of scary at times. But, you know, we're getting there.

‍

Tarush Aggarwal (09:50)

From a numbers perspective, makes sense. We typically see this is just a high level indicator. We typically see that technology is about 20 % of the size of companies and data is about 20 % of technology. So 4 % of companies add about 5 ,000. That number would be right around 200. So just.

‍

Scott Sacha (10:08)

Yeah,

‍

that's interesting. That's a good, good, good fact to know because so I always look at when I hire people, I'm like, what ROI, and I don't say this to the people, but you kind of think like, I want to get three times the value of whatever I'm paying somebody, right? when you when you're doing that in a consistent basis, it's like, why wouldn't we just keep hiring more of these people like,

‍

Because like our you're sitting on data so you can you can answer questions that other people can't answer. You know, a lot of people think data people are arrogant, but it's not arrogance. It's we have the answers.

‍

Tarush Aggarwal (10:44)

about data teams, what does your data stack currently look like?

‍

Scott Sacha (10:41)

yep.

‍

Yeah, so we're a combination of four different companies. So we have a lot of data stack. And we're trying to kind of centralize that a little bit more now than we used to. So we have we have Databricks, we have Oracle, we have SQL, we have Snowflake. know, we use things like Toad, we use obviously, we have Alteryx we have. So we have all these things. But we just recently purchased a new platform that I'm really excited about. called DataIQ.

‍

And with Data IQ, our goal is to kind of be able to create more reusable objects. So one of the things I think in machine learning that a lot of people miss and in AI is if you're not doing it in a systematic way, you're not really getting as much value as you could. Right. So you want to create something where you can reuse parts over and over and over again. so with Data IQ, we're going to be able to do that more. And it also

‍

allows us to have a no code environment where we can start bringing in more junior data scientists that like you don't have to be an expert. I actually, I actually chose this tool for a specific reason. One thing I believe in data science is that's very, very important is and solving problems. So we all solve problems in different ways and I never want to take that away. So the nice thing about data curious is agnostic. So you can use, you know, you can use

‍

Python, you can use R, you can use so many different things. you know, I think about it from when like I began, use, you know, SAS was where you started in like, I don't think anybody uses SAS anymore. And it's like, so the things are all evolving. So you want to make sure whatever tools you're using, those tools can evolve with what the market's doing. Because I mean, obviously, to me, Python is probably the most powerful thing right now, in my mind.

‍

Tarush Aggarwal (12:15)

What?

‍

Yeah,

‍

that makes so much sense. And it's very interesting to see you have all of these different data stacks with given, as you said, have four different business units. How do you look at the centralization versus decentralization? Because in theory, let's have a single data stack mix is a dream. How do you manage the problem?

‍

Scott Sacha (12:36)

Yeah.

‍

Yeah.

‍

So it's complicated,

‍

right? And I think one of the things that I've learned over time is having one data stack is not the route you want to go because then lose negotiating power, right? Like that company has you at that point where you can no longer have any leverage on them at all. So if you're only using SQL Server, they can just keep running those prices up there's nothing you can do about it. But if you have two kind of different

‍

options if you're a SQL and Oracle, it's good. So I think having multiple is good. Now centralizing is the ideal situation, right? Like if all data is flowing together, working together, that's ideal. you get there is so much more challenging than I ever thought. one thing with it, you kind of have to have good ETL in there so you can connect in different ways and forms. You can move data to connect it.

‍

And you buy tools that allow you to connect to multiple data sources and pull it together. That's really where we're at at this point. We're working towards a more centralized data. We're more towards a centralized data, but I think it'll be a while before we fully get there. Again, also because of contracts and data rights and things like that.

‍

And moving to the cloud too is another, with medical information, was just a few years ago really to where they were certified for PHI. we're probably behind compared to some other types of businesses, but we had to be. we're kind of hybrid cloud now.

‍

Tarush Aggarwal (14:18)

Yeah, obviously one of the big questions in the data world is the race between Databricks and Snowflake. And given you mentioned you are a current customer of both of them, was hoping for controversial opinions publicly on how you see them.

‍

Scott Sacha (14:34)

Yeah, well, so it's

‍

it's it's funny because I see like my AI and ML teams, they data bricks wins out through and through there. But then when you just have more casual analysts, it's like snowflakes seems to be more the route they want to go. So I don't know. To me, again, I don't know if one is necessarily

‍

Tarush Aggarwal (14:50)

Yeah.

‍

Scott Sacha (14:54)

I think they kind of now almost have more specialized things. know like Snowflake has like Snowpark, I think it is. It's kind of like their thing. honestly, I think if they kind of specialized in slightly different things, it would probably be better for the market because I think they have strengths in both. I would say as a company, I'm starting to see more people lean towards Databricks, really.

‍

Tarush Aggarwal (15:15)

Very interesting.

‍

What is one achievement which the team has made? I know there are tons to speak about, but if you have to narrow this down to one of them, what's one achievement which you have been really proud of?

‍

Scott Sacha (15:27)

So I'm going to kind of say it in two, which I know is cheating. So I think one thing, coming into this, we were kind of on the outside as a data team. We weren't really a part of the process. We weren't really included in meetings. weren't really, I don't think we were used in the way we should have been. And over time, we've kind of pushed and.

‍

bagged our way into things and shown value, you know, I'm a person that I believe if you if you have a problem, just give me a shot at it. And even if you won't give me a shot, I'm going to try to take a shot at it I want to prove that we're worth what we are. you know, one of the things I think we did really good and at first was we partnered with the sales teams and we really helped them optimize their sales. So, you know, in software business, a lot of times you have a technical salesperson and a regular salesperson. Well, we didn't really have

‍

that. So we kind of came in as more of technical side. But because our products aren't physical products, we have to show value in those products. So we have to generate models that show value for each of our products. And they're so different in the market that you kind of have to like customize this. So it takes a lot of like, it's heavy dive in analytics is what it is. you know, through doing that, we've over time, you know, we've, we've added more machine learning to we've added more

‍

to how we do and how we use things. So I think that the evolution is cool. So I really love that. And then on top of that, we've invented several products that our company uses or enhanced several products that our company use. So we're generating revenue, which is really cool as a science organization. You know, I think a lot of times people love to do all the cool things and do all the fun things, but like

‍

I can do the fun things and fund them. like that makes it a lot easier for me when it comes to go into leadership and asking for things. And I think that's probably a pattern that I've always followed in all my career steps is follow the value because it doesn't matter. All the cool stuff is cool to do, but if it doesn't add value, it's not that cool at the end of the day when you work for a business, you know, because they're trying to make money.

‍

Tarush Aggarwal (17:32)

Yeah, what is one area which has been a challenge which you're currently working on solving?

‍

Scott Sacha (17:38)

So legalities is very hard for us. I feel like I've become kind of like up to speed on legalities and being a lawyer and how to argue what we want to do and how you just do. So a lot of times when we have to do stuff, we have to do it use case by use case. And it's super time consuming, super frustrating, because as a data person, I don't really care. Just tell me if I can or can't do it.

‍

I want those bounds, and then I'm just going to push it as far as I can to get it done. dealing with that has been hard for me. And I think one thing, we're bringing in more data governance as a group. So I think that helps it because in the past, a lot of times, your decisions could be made that you could do something, but then you would ask a different person and a different answer would be given. So it's become a lot easier to be able to do it now. And then the other area is obviously centralization of data and like

‍

easier access to data to more broad throughout the company because it's really hard to get access to our data just because of all the rules we have to follow. it's to understand. It's not as straightforward as you would think it is. So if you don't understand what's going on, it's hard to interpret that data.

‍

Tarush Aggarwal (18:48)

Yeah, that makes it donna sense. Scott, thank you so much for your wisdom today and thank you so much for being on the show.

‍

Scott Sacha (18:53)

Thank you. Thank you. I appreciate it. Thank you for having me.