Home » Q&A with Sukumar Muthya: How to Use Data and AI in Business
The future of business decisions relies on data and AI tools. From data collection, to analyzing and using that data there’s a lot you can do with data to grow your business.
Artificial Intelligence, commonly known as AI, is a hot button topic for businesses and consumers alike. As the collecting of data and usage of AI in businesses becomes more commonplace it’s important for businesses to strategically incorporate this into their business.
In the Q&A session, EarthLink’s Chief Data Officer Sukumar Muthya draws on his 28 years of experience in technology and analytics to demystify the world of AI and data analytics for businesses of all sizes. Muthya leads EarthLink’s data strategy, data science initiative, data management, and data quality efforts.
Let’s dive into the essentials with Sukumar Muthya.
Data collection analysis has become the DNA of every single organization, whether it’s small, medium, or large. The important part is to collect. Enterprise companies now have digitalization across the board, of all the functions, so data is collected every single touch point that the customers are interacting with you, whether it’s the website, call center, social media pages, etc.
It’s very, very important for enterprises and organizations to collect that data. The customers are number one. Number two, what they are doing with you and how they are interacting with you, and are they happy using your products and services? Are they dissatisfied? Are they complaining about something? Are they actually sharing the product information?
It’s very important. Once you have that data, you can do many, many things. You can use it for marketing, you can use it for elevating your customer on how they interact with you, and conversions, and so forth. It’s extremely important to collect that data and analyze it. But what you don’t know is if you’re actually already collecting data. It’s probably in different silos.
The important part is to understand where it is coming from and then put it all together and do an analysis of that data. Once you do the analysis, you can internalize some of the data related insights on how you run your business, and then you can also externalize that data, making it more customer centric. It’s very, very critical and important moving forward. It’s kind of a given for every organization to stay competitive.
In my experience, I’ve been working in data for the last 28 years. The common mistake that I see is, in the last 10 years: storage has become really, really cheap. Because of the cheap storage, people are hoarding, right? So, you have tons of data, and then all of a sudden, there’s a question about, “We have all of this data, so what are we going to do?”
Enterprises are trying to solve a business problem with the data that they have, right? I have all this data. They’re thinking, “What can I do with this data?” versus “I have a business problem I need to solve, what data do I need?”
They are not looking at it in an inverted way. They are looking at it bottom up, like here’s all the data I have. What business problem does it solve? Then eventually you end up solving the wrong business problem if you’re looking at it versus if you look at top down at the company strategy, company goals, KPIs. These are the things I want to measure. And for that, what data do I need? If you look at it that way, then you can solve more business issues and make a bigger impact on the business. That’s one thing I see.
And then the most common thing in analytics I see is people getting confused with causation and correlation. There are two variables, they are correlated, but it doesn’t create causation. When you make that assumption, you’re going to make the wrong decision on the data, the wrong investments on data.
I was reading the other day; it’s a very good anecdote: Data is a new oil. But it’s not really, because if you look at oil, it has a almost unlimited shelf life. But once you use the oil, it’s done. Now you come back to data, and then data has sort of limited shelf life, depending on how you’re actually going to going to be using it. But you can use it many times until that shelf life expires.
For example, email address that we collect from customers: You can use it any number of times until the point where they unsubscribe, and then they’re done. You have to know what data you collect, and if that is the right data, and then use that appropriate. If you don’t use the data, then don’t collect it.
Let me give you an example. If you’re receiving a huge number of calls from your customers inquiring their order status and calling your call center, or they may be working with your chat bot. Maybe you’re collecting and notice in the data there’s spikes in the question specifically around order status. So why are all of these people asking order status-related questions? Is there a problem with your inventory? Is there a problem with your tracking system?
This is just one example. Don’t waste a lot of resources based on gut. If you’re using resources on data-driven decision making, getting to the bottom of it, then you know you’re on the right track.
Most organizations have some sort of data. It’s all available in silos. Email tracking system has email data sitting in one place. Your order management system has ordered data in one place. Your financial system has financial data customer-related calls in all that data. This is assuming that enterprises are collecting data at different points of their functions.
If they’re just starting up, I would say you need to start thinking about going back to what I said earlier, which is, you must think of top-down versus bottom-up. “Let me collect the data and then figure out what to do with the data.” That’s not the right type of thinking. That’s not the right mindset.
The right mindset is, “What are my KPIs? What am I tracking here? How do I run my business? What are the important factors and variables to run my business efficiently?” Then you identify what data you can start collecting. Then it becomes easy. It’s not easy in terms of implementing it. Strategy and planning are critical. And once you know that, then you can start figuring out what solutions and technology can be put across.
AI is everywhere these days. I mean, you cannot walk 50 meters without the word AI anyway, right? It’s become a part of everybody’s life.
The biggest misconception that we hear is AI is going to take over all the jobs. In my opinion, not really — at least not right now, until the point where we have artificial general intelligence. What people are talking about is still AI, and it’s essentially, in my opinion, replacing a task versus replacing a worker.
If we are all working on certain things, then AI becomes a productivity copilot. It will help our jobs better, easier, faster, efficiently, so you don’t have to spend as much time. The misconception is, AI will replace jobs. It’s not. It’s going to be your copilot. It’s going to be a task replacement versus job replacement.
The second misconception of AI is that it’s not a magic silver bullet that you can use for everything, all kinds of business transformation. It needs adoption. It needs a gradual adoption of the typical business case and then expanding that business case across the enterprise, that’s number two.
And then number three is, AI typically says one side will fit everything. It depends on your use case. Using it for marketing, using it for finance, or using it for any other similar situation, not one size fits all. Also, it all depends on your foundation of data, team, data decision making and model building within AI.
And then finally, it cannot solve all data-related problems. So that’s another misconception. It can solve data-related problems, but not until you create developments, you create a very thoughtful approach to the data, all of that is done.
It is very critical. AI should not start randomly within the organization. AI should be a strategy. It should come from top down, so everybody has to go through a certain level of AI training. That way they understand what it can do and what are the efficiencies that you can drive with AI as your copilot.
It starts with a little bit of training across the company. For best practices, companies go through management training across the board at the employee basic level. There are millions of videos that are available to the public, but they should be more focused on your enterprise.
We need to understand what is free, what is available everywhere, and then how that can be actually incorporated into an enterprise.
There is so much garbage out there, so we need to understand how to leverage that. This requires a certain level of basic understanding of what is right for companies not using AI.
In my opinion, I think it’s starting small. Look at what business case that you want, as simple as your call center. “I want to make sure that with every call that is coming in customers have a great experience.” A customer has a great experience when the call center agent knows something about that customer so they can personalize the conversation about the customer. How can you make that personal? You can leverage data, right? It’s as simple as that.
And then starting small, using one business use case, and then driving some AI applications like that. Try to understand what software applications you’re using in the company, and then what sort of AI built-in capabilities that software has and then start using that.
A lot of times we never use some of the functionality within the software, right? Start thinking about AI-driven decision making, so whether it’s native to the tool or whether it’s custom built. Start leveraging the tool.
It’s important to understand what AI can and cannot do, and then use it for the right product, even though it may be a very, very small thing.
The other misuse that I see with AI is widespread plagiarism, right? So everywhere, whether it’s education, whether it’s the things that we are writing and claiming as our own, for blog posts and those types of things. So, there is misuse of AI-related data, and then I would say there are a lot of data privacy violations. That’s another misuse of AI and data.
Especially when it comes to Personally Identifiable Information and with widespread renovation, with GDPR, CCPA and so on and so forth, it’s even more critical for organizations to protect the data. Now you’re taking that data and training it through the models and using it for every problem in the series of data.
AI is used, whether it’s experimental stage or pilot stages, across all industries. It’s there everywhere. But in my opinion, there are the top two or three industries for AI driven applications. The number one is financial services. Financial Services, obviously, because of fraud and risk management and so on so forth.
The other one is also direct consumers, retail ecommerce, that’s a lot of personalization in collecting data. And think about the amount of data they collect.
Then just the performance and service delivery. AI is very important for those.
Coming from the data and analytics space, there are a couple of things that I’m particularly interested in. One is code generation. GitHub is very, very interesting, and can help you write code efficiently and things that you don’t know how to do. It can give you an idea how to go about writing it.
I’m a big proponent of talking to data versus the whole entire BI reporting. In my opinion, this is going to go away in two years. Because reporting is always about asking the user to do the interpretation, do the math based on the chart, based on the numbers in the table, versus talking to the data and getting your answers directly what you want. So, you know, that’s within BI data analytics space, tools like this, plus AI and so on and so forth. So, you can just read all the data and give you information.
Jasper AI can immensely increase marketing productivity, especially when it comes to copywriting, writing, blogs and text SEO, optimized texts and stuff like that, right? And of course, our productivity tools are there for your day-to-day things and tasks that we are using — like right now to transcribe our conversation.
AI obviously is a new thing. For me, the foundation of all of these things is collecting data in multiple places. Number one, bring it all together. Number two, find it, make sure it’s actually in a usable format. Number three is the data governance layer. Then, once you have this foundation, you can build many, many AI-related applications for marketing, transportation, logistics, finance, healthcare, you name the industries that we talked about. You can build many custom applications as well, as you know, native applications that are useful.
I’m a big fan of the worldwide web, but there will be enterprise-based internet. Google is one thing, but enterprises will have multiple applications. It’s all in different places, but there will be an umbrella application, which is an AI. With all of these applications, you can ask questions about anything. Take a CEO dashboard application, for example. A CEO can ask me, “What’s going on with my call center?” And that will give you a response back, because it is run by different applications, but the data is accessible. It’s an enterprise tool that is not too far away from becoming reality.
980 Hammond Drive, Ste 400
Atlanta, GA 30328