Unlocking Value From Your Data
Whether you’re a startup, a midsized company or a large organization operating at a global scale, data is flowing through your business every second, connecting you to the patterns and trends of your clients and competitors. With so many possible data points to track and analyze, the opportunity for your business to learn and adapt is vast—but making the most of this data and keeping it safe can also pose real challenges.
Hi, everyone. I'm Anish Bhimani. And with me today is Rob Casper. On this episode of TechTrends, we're going to talk about how to unlock the value of data in your company Rob, why is data so important to a company?
Yeah, whether or not you're an industrial company, a financial company, whether or not you make cars, at the end of the day everything comes down to data. Data tends to be the lifeblood of just about every organization today, whether or not you're a digital company, frankly.
And how should companies think about starting to get their arms around their data?
Well, the first thing you need to do is find a business problem, and mobilize around that. And that business problem could be some external regulatory driver. It could be some internal risk-management imperative. It could be something that's just really important to the business, and there's an opportunity to be seized. But that's what you need to really mobilize around-- a data-governance agenda.
And you talk about data-governance agenda. So when they do decide to get their arms around this, what are those priorities for data governance?
I tend to think about the priorities in three buckets. There's data landscape. There's reference data, and there's data quality. In terms of data landscape, what I mean by that is, you need to understand what data you have and where it comes from, and where it goes to-- and then use that to inform the selection of your authoritative sources of data. What are the right sources of data to be consuming from?
Second, reference data. So reference data is data about your clients. It's about your customers. It's about companies you interact with in some relationship, like a vendor. You need to manage your reference data for leverage, because it delivers a lot of leverage for the organization, as you don't want to recreate those data sets all over your company.
The third is data quality. And data quality, we tend to think about in a continuum from least mature to most mature. Least mature is reactive data quality-- where a data-quality issue happens, you have a problem, and then you fix it. That's not ideal.
Then, you move to proactive, where a data-quality issue happens, but you catch it before it has a negative impact. That's better. And then third is the most mature, which is preventative data quality-- where you actually design data quality into your systems, so that data quality issues don't happen at all.
So this sounds like no small thing for people to get after. So when companies do start to get after their data and build a data-governance program, what are they in for? What should they expect?
You know, it tends to be a multi-year journey, if you were not a company that was born digital. And so you should be prepared for that type of commitment. But definitely try to chunk the deliverables in smaller bite size than years, and years, and years.
So we try to structure our programs so that there are meaningful business benefits that are delivered in three and six-month increments. Because if you ask somebody to go away and get better data in three or four years, they're going to disengage pretty quickly.
What's the price of not doing this-- or getting this wrong, I should say.
There's tons of benefits to a good data-governance strategy, or otherwise put, just having good quality data. There's regulatory compliance. There's risk management. If you don't get data right, those tend to be disadvantaged. How do you serve your customers effectively if you don't understand them? That's critical.
And then there's all types of efficiencies that are lost, whether they be technological or operational efficiencies. If you have a lot of different systems out there creating the same data, not only do you have redundant infrastructure, but you have redundant teams that are creating and maintaining this data, putting controls on that data. And it's just wasted time, wasted budget.
In previous episodes, we talked with Joe Kane about machine learning. We talked with Christine Moy about blockchain. How do you think about data's importance in machine learning? A lot of interest in machine learning and blockchain things.
So I guess the first thing I would say is that don't get overly enamored with artificial intelligence, and machine learning, and big data, and robotics. That's kind of the really interesting, shiny object in the corner. It is important. But if you don't put good data into it, you're not going to get great insights out of it.
And so the way I think about AI, and ML, and robotics, and big data, I think of them as a customer. Just like I think of finance and risk in the lines of businesses as customers, the data scientists need good quality data. And so we try to serve them in that same way.
So one of the challenges a lot of companies have is trying to do machine learning without having done the data work beforehand. What happens then?
Again, you're going to get out what you put into it. You put bad data into machine learning, you're not going to get out the insights that you hoped for. So don't pretend as though throwing tons and tons of data at some algorithm is going to magically make the data not only good, but the insights valuable.
Right. If less is more, just think about how much more more would be.
It just doesn't work that way.
Yeah, right, exactly. So this is also kind of a large undertaking for the whole firm. So what kind of stakeholders need to get involved in this. Yeah, a data governance agenda really benefits from having support at the very senior levels, and so you try to get them engaged. The ability to communicate a vision around data governance in a simple, easily-digestible way is really important.
What you tend to find very often is that senior leaders kind of don't understand why you're doing this, because they think it's already being done. So it's a little bit of a chicken-and-egg situation. And it's only through education and letting senior management know that good data governance doesn't just happen, and it might not just be happening-- that they need to invest in very obvious, but things that are not happening. Like, what are your authoritative sources of data?
So engaging senior management with a very clear and simple message is paramount, and then having that cascade throughout the organization.
I remember talking to one of our executives, who really didn't think that there was a data problem. Because what was happening was there was a lot of data going into these systems that came out as nice, clean reports for the executive. Well, it turns out there were 40, 50 people massaging all the data and making sure it was all in the right format. He had no idea.
Yeah, you could have kind of two situations. One is where the data on the report is wrong. And that's just bad. The other possibility is that you have a lot of people in the background making that report good. The good news there is that you can rely on the report. The bad news there is that it's got a ton of inefficiencies all around it, and you just don't appreciate just what it takes to put that in front of you.
Right. And in a highly digital world, where we are looking at things like machine learning, you can't have that much manual effort in the middle of it.
That is not straight through.
Right. So we talk about technology. So what's the relationship between data and technology? A lot of people do look at this as a technology problem, but it's really not, is it?
No. I think people need to realize that data and technology are two discrete disciplines. Data doesn't just come free with technology. But at the same time, technology and data need to be inextricably linked. And so I think of it as a very healthy relationship, but one that needs to be acknowledged as two discrete disciplines-- one complementing the other.
And how do you think about privacy-- a huge issue in the industry these days, and especially around the world, in light of a lot of major issues in the industry. People really care a lot about their privacy of their data-- that it's not being used for reasons that it was not intended for, et cetera. How does privacy security play into the data strategy?
So privacy, security and entitlements, records management-- what you need to retain versus what you can delete-- all very important. And all a byproduct of a good data governance strategy. If you understand what data you have, how sensitive it is, and where it resides, then you can ensure that you're protecting it the way you need to protect it, and that you understand exactly where it's going and whether or not it should be going there. So privacy is something we take extraordinarily seriously, but you need to enable it with a good data-governance strategy.
So, finally, for a lot of our listeners who might be in smaller companies-- maybe don't have the resources of a firm like ours or something like that. what advice would you have for them about how they get after this?
So one of the questions would be, do you have an organization that's responsible for technology? And if you do, then you need to have one that's responsible for data. Now, does it have to be an enormous organization? We have a very sizable organization dedicated to data governance. That doesn't mean that a smaller company needs to have that. But it would benefit from having people who are clearly accountable for data governance, just like you have people clearly accountable for technology.
And in companies where they are using third parties as technology providers or developers, how do they think about it?
Yeah, that's going to happen in a number of different cases. The important thing there is that you bring those technologies into your firm sensitive to a data model that brings all those disparate places together. If you just allow every data model to be different, then it's going to be hard to speak with one voice as a company.