Saturday, July 13, 2024

Data Analytics: Key Trends

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

It’s now clear: data analytics runs our world. Success in any competitive market requires efficient and effective use of data. The problem? Data has never been easy to harness, and the sector grows only more complex as applications and markets themselves become more complicated.

To provide insight into getting more from your data in 2021 and beyond, I spoke with three top thought leaders:

Andi Mann, Chief Technology Advocate, Splunk

Mike Kavis, Managing Director, Deloitte Consulting

Mark Thiele, CEO, Edgevana

Among the topics we discussed in a wide ranging conversation:

1. Where are we now with data analytics? What is your sense of how most companies are succeeding/failing with data analytics?

2. How important is Analytics at the Edge? Will this rule in 2021?

3. Will AI become a minimum success criteria for analytics in 2021? Or is this idea mostly just hot air?

4. What advice would you give the C suite – or to staffers – to use data analytics more effectively in 2021?

5. Future of data analytics, 2-4 years from now? And how can companies prepare for it today?

Download the podcast:

Watch the full video:

Selected quotes from the full discussion:

What’s the current state of data analytics in business? To what extent are companies succeeding with their analytics practice?

Mann: My view is that the future is here, but it’s unevenly distributed.  Data analytics is absolutely working for some businesses. Especially in retail, for example, you look at the dynamic pricing, especially like dynamic pricing on airlines on products for online stores, these sorts of things.

You look at some of the activities within my world in the IT operations and DevOps world, using analytics to understand customer engagement levels and anomaly detection for problems or security breaches, that’s absolutely working. There’s a lot of analytics is working by the business level and technology level, but there’s a lot that’s not.

We’re in the revolutionary stage, I think, and I’ve always had this theory that first comes the revolution, then comes the management. And you can talk about that in real terms, political terms, but you can talk about that in technology terms as well. We’re at the revolutionary stage here, we’re trying to figure out what works and what doesn’t. We’re making experiments, some things are working beautifully, some things are failing, it’s unevenly distributed.

It’s uncontrolled to a large degree as well. I worry a lot about data ethics and how we put ourselves and our biases into analytics, machine learning and AI, but I think there’s definitely results coming out, but it’s all very uneven.

Kavis: I think at the product level, sometimes it’s succeeding, so certain services or products are backed by data, but as corporate strategy, not so much.

First of all, where is all my data, is it any good? Different information, different silos. This goes back to when big data was first a term, and everything was on-prem, people were buying all this stuff, but they didn’t have use cases for it, and so there was a lot of failures.

Now, stuff’s cheap on the cloud, consumption’s cheaper, storage is cheaper, and now I think products and services are really taken it just like, they’re really taking advantage of this technology as part of their products. So I see it there, like you mentioned retail and pharmaceutical, in some of these they have great use cases where they’re leveraging the stuff and really making a difference, but I just don’t see corporate enterprise-wide strategies going anywhere.

Thiele: I would say that one of the key problems most companies have is not being able to one, identify metrics, or two, identify data that they think is important to those metrics.

So you can have all the great metrics you want, you could have instant turnaround with information on customer interaction with your company’s products as an example. But if it takes you two weeks or two months or whatever to react and respond to that data, then you’ve lost most of the value associated.

What I’m seeing from a success standpoint are those that have a fairly large target of opportunity for set analytics, and where I’m seeing it is in two places, other than what was mentioned already. Retail is a great example. You can say Walmart’s been great at analytics for years, even in the early days of big data and analytics, they’ve been good for years. But where I’m really seeing it is on the factory floor and in large critical infrastructure. People are using analytics against expensive equipment and making real-time decisions about repair cycles and even refueling schedules on truck routes and things like that, that lead to greater efficiencies and overall improve performance or safety.

How important is analytics at the edge in the coming year?

Mann: From my position at the center of the analytics world: analytics everywhere. Data is coming from so many places. As Mark said, some of this stuff is like, is my turbine gonna fail catastrophically, or can I spend a day shutting it down, doing some preventative maintenance, lose a grand, but get it working again? When that turbine is sending me signals every millisecond that says, “I’m fine, I’m fine.” Do I really need to use the bandwidth to transmit that to a central location and realize that turbine’s fine?

But I need that one data point that says, “Not so fine.” So I think it will happen everywhere. I deal mostly with centralized enterprise-style analytics, I think the edge will be hard for analytics as well.

Thiele: Edge is an incredible opportunity to collect new data and use that data in new ways. Most of the customers that I’ve worked with, friends that I’ve worked with who are doing things around edge, my initial efforts around edge have indicated that data volumes tend to be a lot bigger than what most people originally anticipated. And once they start solving for one problem they realize they can solve for additional problems, which only magnifies the data collection and reuse opportunity.

And to Andi’s point, simply said, it doesn’t make sense in most cases to send that data somewhere else. In real terms from an ROI standpoint, in many cases, it’s just not financially viable to send the data somewhere else.

Edge historically has been defined as a latency-driven opportunity, or opportunities that benefit largely from latency. Latency in many cases, will actually be a positive by-product of the fact that in order to get value out of the data and do it at the lowest possible cost, you’ll wanna crunch the analytics against it, right there, where it’s created.

Kavis: I’ll use the example of wind turbines. So wind turbines have all these sensors on them and they have actuators on them. The actuator will help shift the angle of the blades to maximize capacity of energy.

In that model, you set up, you set it up with intelligence that says, “Hey, if the wind changes by this much, go change the blade.” So in that case, you don’t need real-time AI, you already have an instruction set there. What you do is you trickle back irrelevant data to go back to the data center to figure out the why and the how, and then you go back to the Edge and you change the instruction set. So in that case, it’s a pretty static environment.

Now you look at smart traffic. You have to look at what’s happening now and make decisions and there’s a component of machine learning there over time, your algorithms get smarter. Definitely, in real time there’s no need to go back. You may wanna go back to do some analytics or something, but really everything has to happen on the Edge there.

There’s no real Cloud in that real-time solution. So this is a challenge to architects. It’s not a binary thing. You have to look at each thing, look at what makes sense and what you have enough money to do too, because there’s money involved either way.

What advice would you give to managers who want to use data analytics better in the year ahead?

Kavis: I always go back to focus on business value. Too often, it’s technology for technology’s sake. Find a use case.

Because what happens is when there’s a use case and someone at the end is expecting a deliverable, decisions get made, data magically gets cleansed. Action happens. And so technology for the sake of technology, those are projects that never end and spend a lot of money and don’t get a lot of results. But leverage it for business outcomes so that it could even change business models or drastically improve productivity, those types of things – I would focus on value.

Mann: We found that the more data you collect and the more you operationalize that data, the more you share that data, you actually get better outcomes, empirical outcomes, revenue increase, cost reduction, innovation, speed, agility.

And it actually makes sense: The more data you have to make a decision, the faster you can make the decision, especially if you have a good analytic tool set. So I would say collect more data.

The other thing that it does, it reduces that bias. The more data you have available to you, the more likely you are to make less of your own personal opinions and biases. And so you can look at the data and say, “the data says this, and so I can go with that because I know I have a complete data set.” So yes, beyond absolutely agreeing with Mike, I would say, “Collect and share more data to make better decisions faster and more effectively without bias.”

Thiele: But data, as Andi pointed out, data can be vital, and is vital, but too many of us make the assumption that magic will come from data just because we’ve collected a lot of it. And really the most important piece is knowing what you’re looking for and being able to ask the right questions of it, and then I tend to agree that adding more data to the pool will often help reduce the risk of bias and potentially give you deeper answers.

If I had to pick a concern relative to our conversation today is that I would say that all three of us, as panelists are old enough to remember each of the most recent, the last 15 years worth of data trends. And it seems for the most part, is it 70/30, 80/20 of failure to success? That is, before we actually finish a data warehouse and make it really successful at every level of business or data lakes or data analytics or whatever it is.

By the time the bottom 70% of enterprises, [in terms of adoption] are attempting to adopt, it’s already forgotten and we’re on to the next thing. So it’s a dangerous place and it’s an expensive place to go without having a really, really directed vision for what you’re trying to do.

Most people are looking at data in a big way along with customers and efficiency as part of their digital transformation journey, and that both of my co-panelists would agree with the fact that everybody agrees that digital transformation is the right thing to do. But anybody that’s ever tried to do digital transformation would also agree that it’s the best way to screw over your organization if you don’t have a clear vision for what it is you’re trying to get to, right? And that applies to how you plan to use the data, who’s gonna be responsible for it, how are you gonna maintain it, how are you gonna validate it, all those things.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles