Episode 9: How to Build a High Functioning Data Team

Has your company built an in-house data team? Are you planning on building one out? Have you hired a third party company to build out all sorts of analytics products only for no one in the company to use any of them? Or, are you in a situation where you’ve hired a highly skilled data scientist only to gain zero insights from the resulting machine learning model?

This is very common in all industries. This episode will alleviate these (expensive) pains to your bottom line by taking you step by step to building out an All Star data team!

To keep up with the podcast be sure to follow us on twitter @datacouturepod and on instagram @datacouturepodcast. And, if you’d like to help support future episodes, then consider becoming a patron at patreon.com/datacouture!

Music for the show: Foolish Game / God Don’t Work On Commission by spinmeister (c) copyright 2014 Licensed under a Creative Commons Attribution (3.0) license. http://dig.ccmixter.org/files/spinmeister/46822 Ft: Snowflake

Show Notes:

Welcome to data couture, the podcast about data culture at work
at home. And on the go. I’m your host Jordan bohall.
If you like what you hear, be sure to subscribe lows, get the latest episodes wherever you get podcasts. And if you’d like to stay up to date on everything data couture, be sure to follow us on Twitter at data couture pod. Finally, if you’d like to help support this in future episodes, consider becoming a patron of the podcast through our Patreon page@patreon.com.
forward slash data couture. Now on to the show.

Welcome to data couture. I’m your host Jordan and this week we’re on the road in fabulous Las Vegas for the Pega World Conference. Using my very capable Zoom H five hand recorder, I’m going to be interviewing some interesting influencers in the data game, as well as talk to you about how to build a data team. And if you can’t tell, I’m just now getting over the cold. So I apologize for the nasal sound in my voice. In any case, stay tuned.

Okay, so for today, we’re going to talk about three things. The first is what is the data team? And what sort of people does that encompass? Second part is, we’re going to talk about, namely, how I see that going wrong and a lot of companies. And when people hire data scientists and say, Oh, yeah, we’re doing analytics, we’ve got an analytics team. Yeah, that’s not going to cut it. So we’re gonna talk about that. And then the third part, we’re going to talk about how you regardless where you are in the country, you don’t have to be in Seattle, or San Francisco, or New York City, or Austin or DC, to create your own high performing data team at your very own company. So let’s start by talking about what exactly is a data team or an analytics team or they go by a quite a few names data engineering team, whatever you want to call it?

Well, typically the consistent a few different types of people. The first is typically the leader of the group, namely your chief analytics officer, or sometimes called a chief data officer, or a Chief Data Governance officer, or in my case, Vice President of analytics, Vice President of insights, that kind of thing. Again, in future episodes, we’ll get into the details about exactly what that kind of person has and does. But they lead the group, they interface with upper management, or the senior leadership as well as drive the various projects across the organization. second group of people are your data engineers, sometimes called DB as sometimes called Data engineers, these people, typically they grab your data from wherever source systems, they happen to be, they to simplify it quite a bit, put it into a data warehouse.

They also do data integrations, they optimize the warehouse, they clean the data if they’re able to do so. And really, they’re the heartbeat of any true good data program or data analytics team. A second type of person are the visualization engineers, or sometimes called the business intelligence engineers. And these folks are really the ones who makes the data useful by and large for the organization. And so tasks could include anything from doing simple analysis for a particular project for a particular business unit or vertical. Another one could be creating dashboards with visualizations to bring the data to life, so that instead of having to dig through spreadsheets, or whatever else your company uses, they just simply go to an automated dashboard using tools like Power BI, or tableau, or many of them, sure hundreds at this point, other visualization tools, so that one that’s automated, either real time or close there. In the case of banking, we call it banking real time.

Because typically, the data is about, I don’t know, a day old, best case, we’re trying to get better. In some circumstances, we have data that’s 15 minutes old. So you know, we’re getting close. And in case, these people also serve as consultants with other business units to help sometimes drive requirement gathering, so we know exactly what to be working on in the data analytics team. Which leads me to another type of person, your common business analysts. And so sometimes these sorts of people live within your cmo office. However, it’s good to have one on a data team or if not more than one, so that they can be really the face of the team when it comes to getting those business requirements trying to figure out exactly the problem we’re trying to solve in the data team using our various statistical methods.

Another type of person, one that you’ve probably heard quite a bit about are your data scientists. And so data scientists, they do quite a few things. They only collect the data they need, and go mining for the data they need. But they also clean the data, they transform the data into something useful for models, whether that be a machine learning model, predictive analytic, model, and artificial intelligence model, what have you, and then they operationalize that particular solution so that it can be used, again for making good business decisions. And often overlooked part of a data team is the so called storyteller. This is beginning to get more traction within organizations within the data professions. But a storyteller is absolutely fundamental to a data team.

Because often you have people who are highly technically skilled, they can do all sorts of statistical manipulation, all sorts of ETL processes, extract, transform, and load data into a warehouse from various source systems, the very good good, just putting their head down and crushing whatever project they’re working on. However, at the end of the day, they’ve created something but can’t really tell a story with it, they can’t show the can’t prove business value to the people for whom they’re making these products. And that’s where your storyteller comes in. Sometimes, your storytellers going to be your bi engineer, your GPA, or data scientists or your Ba, or whomever else on the team, or even perhaps the leader of the team. But it can be very helpful to have someone who can show the end user how to use the data product, how to gain insights from that data product, as well as just how to make sense of what was created. Because these are all new things, and especially for laymen who aren’t used to new fancy data products and might seem intimidating. And so the storyteller is also serving as a teacher for everyone in the organization.

Of course, there are many variations, all these particular roles, but we’ll start here. And if it seems overwhelming to you, don’t worry, we’ll make sense of it in the third act, when we talk about how to go about actually building one of these fancy fancy data teams. But for now, on to act two.

Okay, welcome to act two, we’re going to talk about what I see is a very common problem in any organization that wants to build a data analytics team. And there are two problems here. The first is a matter of strategy. And that’s, well, we can’t hire a data team. And so why don’t we just outsource and use a bunch of third party vendors to build us bunch of dashboards to maybe even prop up some sort of predictive analytics algorithm. And then we can say that we’re doing analytics, we’re on the cutting edge where our modern company. Now here’s why that’s a problem. Yeah, there are plenty of companies that build plenty of data solutions using your own data. And they do a very, very wonderful job, I certainly don’t mean to undercut any of them, because I know quite a few. And I know the work they do, and it’s absolutely outstanding. However, the problem is, they spend time with you with your data, they, after a number of weeks, or number of months, build the various data products, they integrate it into your system.

And then they’re done. They’ve done their job, they’ve fulfilled their contract, they’ve done every part of their statement of work. And now you have this monolithic data system or platform or set of automated dashboards. And you tell your employees, hey, we’re a data driven now use data to make decisions and that kind of thing. And that’s, that’s what you can hear. That’s what a lot of companies consider being data driven, or at least having an analytics operation happening in their company. Well, I talked about this a little bit when I mentioned the storyteller role. But even though you have this really cool set of products or product from particular third party vendor, that doesn’t mean that the majority of your organization knows how to use it, it just means that you paid a probably pretty penny for this particular data solution.

And now, it might never get used, because no one’s very sure about what any of it means. Right? Well, so then you think to yourself, and this is the second difficult to that I see Oh, well, I should have an in house data team. And again, there are two pieces to this. The first is trying to solve what to do with this third party solution that we just paid an arm and a leg for. And so you hire somebody in, they have to learn the system, they have to learn the data, they have to learn all those pieces, but then be able to communicate out to everyone in the organization, how to use it. Well, here’s the problem with it. One, you’re not going to get a quick turnaround on this, it takes quite a while months, perhaps more than one year for somebody to actually learn the data that is within the enterprise to It’ll take a while for this person to be able to then understand how all these systems all these data solutions were created by the third party vendor. And so you’re not going to get that turnaround, you’re not going to get the ROI that you’re hoping for out of these data pieces as quickly as you may have hoped for. And the second part is with that this person might not be at the right level within the organization to actually enact positive change.

And so even though you have this data person on your team, and you have this analytic solution, nobody above them is going to really take to heart what they’re talking about. And then an act, any business process change because of it, and hence, no one in those various verticals led by that particular senior person will then be using the data solution. So there’s a problem with that piece. And the second part of it. And the second part of this whole section is not going to third party room, but instead investing in an analytics team. And here’s how people tend to do it. They have a problem that they’ve seen others at conferences, perhaps solve using a data science perspective, or a data science solution. And so they say to themselves, oh, I’ll hire data scientists. And they’ll knock out this project for me using all those fancy data science techniques that I keep hearing about from all of my peers at all of these conferences that I keep going to. And so you hire data scientists, and frankly, they’re not cheap, and nor should they be. But nevertheless, they get in, and they get ahold of the data, they do all the data wrangling the data mining, they do all the data cleaning necessary for this project, they create a sophisticated model or set of models to then be able to predict or to learn what the right solution is for this right or this particular project. And then at the end of it, they operationalize it using various means.

And then what? Well, you have the same problem is in the first scenario, great, you have this cool project, you have a operationalize using some kind of cool dashboard or whatever else. But your people, the rest of your employees don’t know how to use it. They don’t know what any of it means how the data scientists themselves might not even know which insights should come out of this or how to affect positive business change. And you’ve wasted six months to a year of your capital and your time and, frankly, your competitive advantage going down this rude. And so what do you end up with?
what many consider a bad taste in their mouth for the data profession, they end up with a lot of expense and no return. So what do we do given these two scenarios, which are very, very common, and every industry that I can think of? Well, that’ll be Act Three, we’ll talk about how to build a data team, what to expect from them from the start, how long to expect before they can actually enact any positive change, and then where they can go in the future. Stay tuned.

Okay, so so far, in the first and second act, we talked about the sort of roles that comprise a data analytics team, as well as many of the common pitfalls that we see when companies attempt to create a data team on this final act will talk about perhaps one of the better ways to create a data team. Of course, this isn’t the only way. But it’s a way that I have found that works well, and produces actionable insights and results from all the money that you’re about to spend to build a high functioning data team. So the first as I mentioned, was to hire a data leader. Now this can be the chief analytics officer, the chief data officer, or someone at the vice president level, the Vice President of analytics, say, this person is crucial, because and they should be hired first, so that you can one to find the data strategy.

And the data strategy talks about everything to do from data governance to the direction of how data will be employed across the organization to making sure that all the data policies and practices align with the overall organizational practices and policies. And so once you have this person in place, and again, on future episodes will talk quite a bit more in depth about exactly what a chief analytics officer chief data officer does. But this person will then be in a position to strategically hire various roles so that the analytics team can be as successful as possible. And the very first thing that a chief analytics officer should do is hire a DB a. Now, a DB a is historically someone who optimizes and creates architectures and ETFs processes for a data warehouse. And so really, to start, you’re going to need more of a data engineer DB a type role. Someone who can build custom stored procedures and who can really understand whichever framework your company happens to work in my current one, we’re on a net framework, so the Microsoft frameworks that will be really helpful, were able to have very superior SQL writing skills.

Nevertheless, this heart of your team, the CPA slash data engineer, folks or person will be able to proceed forward with creating a data warehouse that will be optimized for data analytics, data dashboarding, machine learning all these sorts of pieces, which leads to your second hire, your second hire should be your bi engineer, your business intelligence engineer slash visualization person. So now that you have data minutes growing and your data warehouse, you have someone who can unleash the power of that data using various statistical methods using various visualization practices, so that then they can start democratizing the data allowing people to see the results get the insights from this previously unused, extremely valuable asset across the organization, this person will allow you to create these ultimate pieces that are then accessible by everyone across the organization. Which means that when folks go to meetings, they pull up the dashboard, they’re not passing ships in the night, so to speak, they’re speaking from the same data to make business decisions that will hopefully drive your organization forward. Know, the next person you need is your Ba, your ba or your storyteller either way, because now that you have a few dashboards pushed out, you’re going to be flooded with all sorts of requests. And most of these requests are going to be of the static report variety. And so your ba or your storyteller, the people who can gather the requirements, but also teach the people how to use what’s produced, will be able to work with the business unit or the vertical to determine

Well, what really needs to be solved? I know you want the static report, or what is most usually a static report. But what do you what are you trying to solve? What problem? Can’t you get your head around? Are you having difficulty getting your head around where a data solution might be the right answer. And using that you can improve processes, and you can get the business line leader or anyone in a various vertical to understand that, look, these are very dynamic pieces, these data processes these data solutions, they’re very dynamic. And they can unleash all sorts of previously unforeseen circumstances and unforeseen possibilities to make a profit, for example, that perhaps wasn’t considered at the inception of the request.

And so that’s the third person you need to hire. Now, you can see where this is going, you’re going to get more and more requests as you get more and more cheerleaders as you start getting more and more wins. And so you’ll have to add another DB or two, you’ll certainly have to hire more bi engineers, you’ll probably have to get another ba or storyteller. And so you can see how the team quickly pushes out to 10 or 12.

Now you might be saying Jordan don’t I need a data scientist?

And the answer is probably not not yet anyways, why? Because your analytics maturity of your organization isn’t there yet. You started with static reporting using Excel, you started building a data team. So now you jumped up a level or two and you have automated dashboarding. Next Level using your storytelling MBAs to push in the Self Service analytics where lots of people across the organization not just the data team has access to. So your Power BI or tableau, only then are you ready to start doing predictive modeling machine learning and AI. And so it’s not until a year or two down the road where you’re really in a position, let alone with the amount of data you’ve collected and stored so far in the data warehouse to employ a data scientist. So it’s not until you’ve reached that maturity level, that you then hire the data scientists who can truly help with creating the predictive models and the forecasting and all the various other types of wonderful outputs from AI and predictive and machine learning.

And so that’s how you go about building a data team. And notice that this is a multi year process. And so you’re not going to see any kind of return in the first six months, probably not the first year, but hopefully by the end of the second year. And certainly the third year, you’ll be in a position to reap great business and opportunity rewards from your data team. And this will avoid the two issues that we saw in the second act. Namely, you’re not going to be hiring a third party vendor to come in and make a beautiful data solution for you and the nobody using it because you have a team that is injected into every process within your organization, working with everyone in your organization to build things that matter to those people in the organization. And then you’re having the storyteller and the data, analysts go out and teach the people how to use it. And so you’re not going to be left with a great monolithic data product, project or product that no one knows how to use.

Going to avoid the second piece, namely, merely hiring a data scientist to come out and knock out a project and then nobody knows how to use it. Because again, you’ll have waited to hire the data scientists until your analytics maturity is at the right level, at which case people will be prepared for an eager for the data scientific solution.

And again, you’ll have your storytellers and your and your leadership to understand and explain how to use this predictive model. So I appreciate everyone listening today and I look forward to hearing any comments or concerns you may have to talk to you soon.

That’s it for the show. Thank you for listening. And if you liked what you’ve heard, think consider leaving a comment or like down below. Stay up to date on everything data couture, be sure to follow us on Twitter at data couture pod consider becoming a patron@patreon.com forward slash data couture music for the podcast. It’s called foolish game. God don’t work on commission by the artist spin Meister us under a Creative Commons Attribution 3.0 license, writing, editing and production of the podcast is by your host Jordan bohall.

Liked it? Take a second to support Data Couture on Patreon!

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.