Giving Back and Building Your Brand as a Data Science Leader

DAVE COLE

Hello, and welcome to another episode of the Data Science Leaders podcast. I am your host, Dave Cole. Today's guest is Sidney Prescott. Sidney, how are you doing today?

SIDNEY MADISON PRESCOTT

I am excellent! Excited to be here.

DAVE COLE

Awesome. Sidney is the Global Head of Intelligent Automation at Spotify. We'll discuss a little bit about what intelligent automation is today. And Sidney is on more boards than I can count. She's a graduate of Georgia State University with a degree in Philosophy, which I think we'll also get into. I think you have a very interesting background, Sidney.

Our three topics for today start with diversity in data science. We’ll talk about Sidney's journey in becoming a data science leader and a little bit about building your own brand. If you run a search on ‘Sidney Madison Prescott’ on LinkedIn, you'll be amazed at her background and how much she has on her plate today.

Last, but not least, we'll talk about robotic process automation, or RPA, and how to incorporate this into your data science leader toolbox.

Let's get things going here. Again, Sidney, I think you’ve had an interesting path. It's a path that I hope becomes more normal and a lot less “interesting.” Why don't you tell us a little bit about your journey to becoming a data science leader?

SIDNEY MADISON PRESCOTT

Absolutely. The journey started in undergrad. I began with an internship, which was really my foray into technology and, subsequently, the data science and governance world. Within that internship, I was working with a global payments solution company. They focused on becoming a more digitally aware company. We did a lot of work around connecting all of these disparate databases, making sure that the timing of the data is on point, that the data being received is accurate and is something we can rely on.

I learned so much about the importance of data quality and governance in relation to all of these different automation platforms, whether they are SAS systems, running in the cloud, or on-premise, legacy systems like mainframe, which are still very much out there. It was a great opportunity for me. It set the stage for me to prioritize the importance of data in my career, as I moved forward.

After that role, I sat within the configuration and asset management side of the global software team there. Within that team, we realized that we didn't have a handle on exactly what was happening within these systems in relation to data integrity. I sat at the helm of an initiative that we began, which was really my first foray into data quality. I led a global team on data quality and governance. We specifically looked at all of the systems, whether they were third-party vendor systems, homegrown systems, SaaS solutions, on-prem etc.. We looked specifically at the ETL: how to extract and transform all of this data; how to make sure that it's on point and doing what we needed.

We initially started with just pure analysis around our expectations for these systems in relation to the business processes that are running on them. We quickly realized that our expectations for the data did not match the reality of the extracts coming out of these systems. That sent us on this huge journey into uncovering the issues. How do we start walking through a process of remediation of these data issues? How do we also maintain the integrity of the data, once we get it to a state that we believe is viable for our business in the future? How do we make sure that we keep an eye on what's happening kind of underneath the hood, so-to-speak?

I took that and ran with it in terms of where my career progressed. That led to a foray in intelligent automation, which deals more specifically with robotic process automation, machine learning, artificial intelligence, optical character recognition — all of which heavily rely on data — making sure that we are doing right by our business stakeholders. Their needs that we serve are specifically the data that is coming out of the system and the decisions that are being made (the business decisions off of that data).

DAVE COLE

You were getting your major in Philosophy. During your internship, it sounds like there were some philosophical conversations in the hallways. How did you get that first internship? How did you find a company that was willing to take a chance and have you work in the area of data?

SIDNEY MADISON PRESCOTT

It was a little serendipitous. What I was looking for at the time was an internship where I could be involved in legal contracts. I was really interested in software contracts: how companies understand what their obligations are in relation to their partnerships with other vendors and businesses.

The internship that I landed was within the configuration and asset management team because they were going through a huge transformation of all of their contract obligations around software. They wanted to bring me in for the internship to help review and understand the ownership policies around all of the contracts that were signed and the utilization of the software. The question underpinning my role was: are we truly using all of the software that we've purchased or are we underperforming in that area?

Working in all of these software contract true-ups, I began to get projects passed to me in relation to the challenges that were happening in the environment . These were a lot of different siloed projects that were going on to help clarify why outputs from different systems were not where they needed to be. There was also activity in the direction of clarifying why certain systems were underperforming in relation to their ability to handle the amount of data that was being put through them.

That's what really piqued my interest around moving away from the contract side and more towards the process re-engineering and excellence side. It was practically an on-the-job migration of what my initial role was versus where I ended up. I really took to that and I embraced the opportunities that came to me to work on those different projects that didn't fall into that initial category of what I was working on.

DAVE COLE

I think there is this emergence of specializations in college coursework that allow you to come out of college with a specialty or major in data science. I also think that a lot of people can learn data science on the job. I certainly fall into that camp. My background is in engineering, not anything related to data science.

You're a female data science leader, too. What recommendations would you give other up and coming female leaders as they're looking to go down a similar path? What sort of challenges have you faced in your journey?

SIDNEY MADISON PRESCOTT

Oh, many challenges. The first piece of advice I would give is that if you truly have a passion for data science, quality and governance, don't let anyone steer you away from that passion. I feel as if some technologists, particularly when it comes to working with women and people of color, tend to give you advice to steer clear of certain technology areas.

In my experience, those are usually the most fascinating areas, the ones for which you never know whether you actually have a passion. That's the first piece: don't necessarily take the advice from others to steer clear of certain areas. Allow yourself the ability to dive into the domain specialties that you may not be comfortable with.

Another huge piece is to challenge yourself. Constantly seek out new opportunities to learn and to grow. As you said, a lot of that growth happens on the job and you never know which doors you're closing by blocking yourself off to those new cool things that you can jump into. I would say that for the majority of my career, the really big moments have come as a result of me saying yes to everything.

DAVE COLE

We're going to talk about that in a second. That's an understatement. What I hear you say is that the world of data science is one of the fastest paced and ever-changing genres out there, that you can specialize in. You have to be open to learning new things.

I think as a data science leader, if you're looking at your team, don't just think about what they know or what they specialize in today. Recognize your team's aptitude for learning. You're a great example of that, which segues into the next section: how you've built your brand.

When I look at all the things that you have achieved and are currently doing today, it is astronomical. You're on a number of advisory boards. You probably can tell me how many. You're one of the founding members of Chief, which is a network focused on helping women rise to the executive ranks like yourself. You're also getting your Masters of Science at Cornell Law School and you are a book author. All of this, you have done in a relatively short period of time. How have you done all this?

SIDNEY MADISON PRESCOTT

So I would say by being very focused and very purpose driven. At the core of everything I do is the question of why am I doing this? What do I hope to achieve out of this? Do I feel that this is going to further my development in some way: my skill set, personal abilities etc.? I always start with the purpose of anything that I want to take on or be involved in, no matter what that is.

The other piece is just being laser focused. I clearly love to achieve a lot of things, but I'm also laser focused once I set a goal for myself. For example, I wanted to grow my ability to advise other companies in relation to technology. There's that saying that every company now is a tech company, which couldn't be truer.

When you think about the expertise that is needed, when we talk about data science, automation and emerging tech, there is a distinct gap between the needs of all of these different companies and the body of talent that exists, to work on these projects and initiatives. The gap even expands to advice and coaching of other executives in relation to technology initiatives.

I saw that gap in the marketplace. It’s even larger when you think about women and people of color. So I love to share my knowledge and give something back to different organizations without access to someone with the bandwidth to do so."

DAVE COLE

I framed this agenda item as you building your brand. As I think about it and I hear you talk, I don't think that was your intention. Your intention with a lot of this was personal growth and to help others. It seems like it's a mix of both.

Whether it be you writing a book or going back to school, you’re constantly sharing what you learn. A lot of the boards that you're a member of, coupled with the advisory work that you're doing, to me, rings of you just wanting to give back. How did you get into that path? Where did that first advisory opportunity come to fruition?

SIDNEY MADISON PRESCOTT

Interestingly enough, it actually started from a conversation on LinkedIn. I think that is where it started, which then goes back to personal brand. It's important.

DAVE COLE

Your LinkedIn page needs to have enough content for somebody to reach out and want to have you advise them.

SIDNEY MADISON PRESCOTT

Exactly. You need to build a body of work that leads someone to think, "Yes, this is someone that I want on my advisory board." That is definitely the first piece. I think that having a brand that had started to emerge like my own personal brand, really helped me in terms of getting companies to start to pay attention. It supported their intention and desire to have me on their advisory boards.

That first opportunity was, interestingly enough, a non-profit mental health organization in Ireland. I'm very passionate about mental health. It’s very close to my heart in terms of a cause that I like to support.

Again, it felt kind of serendipitous. It was a perfect alignment of a cause that's near and dear to me and the fact that they were looking for some really innovative ways to include technology in their efforts to advance the mental health of children and teenagers.

DAVE COLE

That's awesome. Out of curiosity, what sort of time commitment do some of these advisory activities take? I'm sure it varies but what does that look like?

SIDNEY MADISON PRESCOTT

Yes, it varies from company to company and then it also varies depending on if it's a non-profit versus for-profit company. Relatively speaking, you're probably not looking at a huge time commitment. For most of my boards I average 15 or so hours a month, at most.

Sometimes it requires a little bit more, especially if there's something really big going on within the organization, like a big leadership shift or something like that. It might be that we're throwing some sort of event and they want me to advise on that too. I have seen some advisory positions ask for more time and I've seen some ask for less. It also depends on the maturity of the organization itself.

DAVE COLE

I know I've had conversations with others in an advisory capacity. I don't know about you but I always take away something from those conversations. If you want to look at it from a selfish standpoint, as well as a giving back standpoint, I do think there's something to be learned in having those conversations. There are new challenges that you see when you're talking to other companies, seeing how they think about the problem. You may walk away having learned something, right?

SIDNEY MADISON PRESCOTT

Absolutely. It's one of those unique opportunities where you have the ability to learn about the way that you work with others, how to tackle challenges when it comes to leadership, team building on a globally distributed scale.

And then, as you said, you also get to see how other companies adapt and meet challenges. I get inspired by that work because a lot of times it will spark something in my mind about a challenge that my own team or company is facing. It’s great because it gets the wheels turning outside of ‘business-as-usual’ role.

DAVE COLE

So the way to kick it all off is to actually build out that LinkedIn page. That's how you got started. Am I correct in saying that it snowballed from there? So, when you published that you were advising one company, others wondered if you’d be open to doing the same for them? Is that how it went or was each opportunity different?

SIDNEY MADISON PRESCOTT

It did. I made a conscious decision to curate my LinkedIn. That was a very intentional thing that I did. I made it a point to highlight anything of significance that I was involved in. I also made a conscious effort to post.

DAVE COLE

That brings us to our final topic here, which is your current day job, among all these other activities that you have going on. So I want to talk about RPA or robotic process automation. Quite frankly, I don't know too much about it. Could just tell us a little bit about what RPA is and why I should care?

SIDNEY MADISON PRESCOTT

I’d be happy to. Robotic process automation is, in essence, software robots that mimic the activities that human users take. This is pretty much anything that a human user does within any given system: logging into a certain system; downloading certain reports; pivoting data in those reports; dumping those reports into another file path. We can code a software robot to, in essence, mimic those steps.

The benefits of the robot include it being faster than a human, typically less prone to error, especially with data handling and manipulation, when compared against humans. We can process more transactions, faster. It enables you to automate any manual task that is very tedious and repetitive in nature.

By automating those steps, you increase your data quality, the efficiencies within a given workflow and your ability to handle more transactions at a faster rate. You can really amplify your team's ability to handle much more as a result of leveraging robotic process automation.

DAVE COLE

Do you have a specific example of it? How many of these processes do you have running at any given day?

SIDNEY MADISON PRESCOTT

Right now we have over 150 robots in production, across a wide variety of business functions: everything from treasury and tax to internal audit. I can give you an example of a process that we've automated actually within our ads team.

The ads team was taking screenshots and recordings of different advertisements that were set to run on the Spotify platform. This sampling of that audio and visual was a part of our quality control. We made sure that any advertisement that someone had purchased with Spotify could be subjected to quality and integrity checks from a digital perspective.

The sheer number of ads and the turnaround time from someone purchasing an ad, to our need to create that visualization, audio, get that ad pushed out to the platform — it didn't leave a lot of time for this very tedious manual testing that was taking place. We wanted to make that testing piece more efficient, and so that's where RPA came into play.

We took our robots and we actually were able to mimic those exact steps: playing the audio back; recording a snippet to showcase the quality; taking a screenshot of the advertisement and then putting all of that in a folder for a human. All the human has to do is just pull those assets and showcase, "Yes, we've tested this particular ad."

That particular bot runs every single day, multiple times a day. We can have bots running near 24/7. I won't say total 24/7 because we have maintenance of the systems and VMs that they're running on, but you can get pretty close to that. It’s all just based on the amount of downtime you need for the servers and things they're running underneath the hood.

DAVE COLE

Got it. That makes a lot of sense. What sort of data gets generated from these robots? How do you improve the performance of them, optimize them etc.?

SIDNEY MADISON PRESCOTT

There's a lot of data that gets produced. The first piece there is data just in relation to, as you said, how the bots are functioning. Are they performing as intended? Are they falling down at any point, at any process? If so, why? There’s operational data from the robots that we receive. There's also data that we gather from the virtual machines that the bots are running on. From a performance standpoint, we check on how our machines are doing. Can they handle the load in terms of the amount of bots that are running back and forth?

We also have process data that is generated as well. This is more specific to the business process itself, whereby we, in essence, are creating any of the data assets that the process typically generates. Then we're pushing those out to the stakeholder. That could be anything as simple as an Excel spreadsheet with a wide variety of data, that the bot has populated.

It could be the bot actually populating data in a database within a system, that's also a data output, or it could be something along the lines of pushing out a report of some sort that then gets dropped for a human to review later on. All the data really varies based on the business process, the frequency of the runs and how many systems that particular bot process is touching.

DAVE COLE

Do you have any idea of the efficiency and cost savings that these bots are providing for Spotify? I know you also did that at your previous job. Is there any rubric on each bot that saves ‘X’ amount of time or something like that?

SIDNEY MADISON PRESCOTT

We have taken a step to look specifically at the number of transactions that each bot is processing. We compare the automated time savings to the original manual process. We do a basic compare and contrast between, "Okay. It took a human ‘X’ amount of time to complete one transaction. It takes a robot now ‘Y’ amount of time to complete the same transaction. Here is your savings once you multiply that times the frequency of the bot runs."

To date we have saved over 75,000 hours in relation to the processes that we've built. That really is just scratching the surface. We have a lot of additional projects that we can pull in. I've seen companies get as high as millions of hours saved through automation. It really just depends on the speed of growth for your automation program and also the quality of your use cases. Are you taking on the most manual repeatable processes and are those transactions usually being made on a daily basis? That gives you the bulk that you need in terms of generating ROI.

DAVE COLE

On the Data Science Leaders podcast, we talk a lot about uncovering use cases and partnering with key stakeholders in drumming these up. A lot of times it's focused on generating models to produce predictions and finding efficiency gains, but there's a lot of repetitive processes that your various areas of the business experience every single day. Those can be automated away and that's what I'm hearing RPA helps with.

Quantifying the benefits is relatively easy. You load and do a pre-/post-analysis. You look at the way the world was beforehand and the way it is today. I imagine it frees up the time for these individuals to work on other things that might be less monotonous and have a higher impact. Along the way, you're also generating this exhaustive data that can be mined for additional insights. I think that's all very fascinating and maybe an area that a data science leader might not always be thinking about.

There's no question in there. It's just a long ramble. This is fascinating to me.

SIDNEY MADISON PRESCOTT

There is a wealth of knowledge that you can glean out of this data. There’s actually a really big project that we are working on now. We have all of this data and we have pushed some of it into data visualization tools. We’re exploring how to make those visualizations more robust and then even more compelling to our executive leaders, in relation to these transformative projects that we're taking on. How can we use this data to actually tell that story of digital transformation?

DAVE COLE

That's awesome. Well Sidney, this has been very fascinating, I've learned a lot today. At the very least I'm going to update my LinkedIn page over the weekend to get going. I can start lining up those advisory opportunities. I also learned a lot about RPA. I learned about your journey and I think all of us should open our minds and hire from all sorts of different backgrounds.

If anyone has a fraction of the curiosity you have, I think they could be a fantastic member of the team and you might have a future data science leader in your hands as well.

SIDNEY MADISON PRESCOTT

Yep. Absolutely.

DAVE COLE

So great to have you. Thanks so much for being on the Data Science Leaders podcast. Speaking of LinkedIn, if people are interested in chatting with you, can they link up with you on LinkedIn?

SIDNEY MADISON PRESCOTT

Yes. Absolutely. I believe I'm one of the only Sidney Madison Prescotts on LinkedIn. I'm very active on LinkedIn so feel free to reach out.

DAVE COLE

Awesome. Thank you, Sidney.