I sat down with our SVP of AI, Raj Balasundaram to learn about all of the ways marketers are using AI today.  

About two years ago, I began learning about this supposedly revolutionary new technology that everyone seemed to be talking about, hyping up, and, apparently, starting to adopt: AI.

It could provide new depths of business intelligence and new dimensions of our databases previously unknown. It could scale, automate, and make decisions about products, pricing, placement, and promotion, about content, channels, and customers, all based on behavioral data already owned by a company.

But I’m a truth seeker, and the hoopla just wasn’t cutting it. So, I began to peel back the layers to find the truth about this dynamic technology. After all, I had to if I were to do my job. So, I reached out to the experts — people out there actually using artificial intelligence including our internal SME, Raj Balasundaram. And that’s where my learning curve suddenly spiked.

One of the reasons I love working with AI experts like Raj is because they have their finger on the pulse of the industry. And I’ll be the first to admit — I don’t now nor have I ever understand the inner-workings of the machines, the code written into the software, or the intricacies of how AI does what it does. As a marketer, I’ll gladly leave that stuff to the coders, programmers, and data scientists who are smarter than me. But people like Raj… their expertise situates them uniquely in the middle of the marketer and the machine.

Related Content: How Marketers Can Thrive in a Consumer-Driven World: Raj Balasundaram [Video]

I sat down with Raj to explore these topics deeper. I wanted to know just how much AI has changed the game, and how marketers can get started thinking about AI. Here’s what I learned.


Q: Two years ago, we know that many marketers and most consumers were uncertain about AI and the true benefits it offered. What’s the public opinion like now?

A: I was speaking to a journalist the other day and she was asking me why AI seems to be portrayed as having “gone wrong.” It’s this kind of hype, sensationalism, and misunderstanding about AI that gives it a bad name.

The reality is that AI has revolutionized everything – the way we see things, do things, and integrate our seemingly daily routines into our lifestyle. Has it “taken away” a few jobs? Yes. But on the other hand, it’s created tremendous amounts of new opportunities. The reality is that we, as humans, are in a process of evolution. AI is merely another computational evolutionary step.

Q: Data is everywhere. How does channel proliferation and the influx of data empower and enrich AI systems?

A: Think about, let’s say, even 10 years ago. The amount of data available on you was few and far between! The amount of data available on me was probably non-existent (apart from a few records here and there). But now that’s not the case — every single second of your life is recorded. Every single microsecond of your heart beating is recorded in the cloud (I’m exaggerating, but you get the point). That’s kind of scary, but also really neat.

Tech, tools, and appliances are ingrained within our lives as the IoT continues its rapid growth. Apple watches, Fitbits, iPhones, and more are as much a part of us as our own brains! Do you know what these devices are fueling? Algorithms! Whether people like it or not, that’s reality.

The moment you put an electronic device near you, you’ve already announced to the rest of the world who you are and what you’re doing. So the data explosion or the availability of data points on an individual has literally gone like maybe one thousand X compared to the last ten years.

twitter “The #data explosion (or available data points on an individual) has gone 1000x in the last 10 yrs,” says @RBalasundaram      CLICK TO TWEET

Data combined with computational power — that’s the real genius of AI systems.

Q: Intrigue around AI has never been higher. Okay, let’s settle this debate — is AI truly “happening” around us as much as some people seem to speculate? Give me a few examples of AI in action.

A: As I described above, what AI is really doing (or trying to do) is offering you a better lifestyle.

In most cases, you don’t even know it’s affecting your life. Here are a couple examples of ways everybody can touch and feel and sort of experience AI in their day-to-day life.

Example #1: Like many of its kind, the Wimbledon highlights app is very customized for whatever you like. But there’s an AI engine working in the background to serve you that content. This engine is basically analyzing hundreds (and maybe even thousands) of hours of footage and creating clips and markers of different events. Then, it puts different tags on clips of these various video points. They (IBM) serve the the correct content you’re most likely to be interested in. In this case, the content is very much catered towards you.

Example #2: AI has been applied in many different diagnostic applications such as detecting cancer or detecting a sudden disease or whatever it is. If you think about it, it’s a beautiful application. AI is actually giving us a very good level of predictive capability in these ways, saying “hey you know you might be prone to this disease” or “you need to get treated now,” and things like that.  The broad applications have been saving thousands of lives. But that one life it misses… that’s sensationalized. And what we all have to understand is even though machines have a high level of accuracy, they’re never a hundred percent right.

So, when it’s a life and death situation, you don’t rely only on the machines. And this comes from a person like me who actually works with AI on a daily basis. AI is not the end all be all. It’s more of like an augmentation to human intelligence, but it can never replace human intelligence. What it can do is take away the laborious things that we do on a daily basis.

Q: It’s one thing to talk about AI, but when it comes down to it, how do you actually do it? Are brands like Facebook, Netflix, and Amazon doing something the rest of the pack aren’t — or can’t?

A: The base fundamentals — the algorithms, the way the science works, the way the building blocks work — don’t change much. How you build AI systems changes based on what industry you’re in, and what you want it to do. And that’s where it gets really interesting. The answer to this question is really three-pronged.

Number one, you need to have a very, very, very good data set. Without that there’s no chance of success. Plain and simple.

Number two, you need lots and lots of data.

The third point, which is the most important (and which many people forget) is that there’s a person who trains the machine, and who programs all the algorithms. There’s a human who sets up the application of those algorithms and models. That person needs to understand that industry as well as the application of the AI solution and for whom it is intended. Not all AI is created equally — there isn’t one standard set of code.

Companies like Facebook, Netflix, Amazon, and others essentially find the perfect combination of everything — they’ve built the right kind of technology environment for everything to thrive.

So, without a good technology environment, you’re not going anywhere. You need to have good data sets, and you also need to understand the business application. This is what I always call the holy trinity needed for everything to come together.


“Without a good technology environment, you’re not going anywhere [with AI]. You need to have good data sets, and you also need to understand the business application.”

Raj Balasundaram •  VP of Artificial Intelligence, Emarsys 

twitter “For #AI to work, you need a good #data set, a large quantity of data, & a tech ecosystem to support it all,” says @RBalasundaram      CLICK TO TWEET

Q: For companies using AI in their marketing, which jobs or positions are usually needed to use AI? What does that internal team structure look like?

A: It all depends on what you want to achieve and how. So either you can establish a team of people who do data science, or leverage a qualified vendor. For companies that prefer to keep as much in-house as possible, they’ll need three things:

  • Two or three data scientists
  • another team of business analysts who can convert all of those findings into a business output
  • another team of marketers who work with the business analyst(s) and take that insight and can convert that into marketing output.

Often you have the hardcore data scientists who are crunching the numbers. They’re putting all the numbers together but then there is somebody else who does the application of it who typically comes from the business side of things. They understand the customer, and they tests and optimize. The combination of these two people is what makes AI work, and that’s what you see in the front end — recommendations and the like.

Related Content: How AI Marketing Works: Achieving 1-to-1 Personalization at Scale

But here’s the kicker — you need all of these three teams to work as a single unit. That’s where it becomes difficult. As we  know, marketers have their own way of doing things. Data scientists have their own way of doing things. The business analyst guys have their own way of doing things. And that is where the issues can arise. Whatever the data scientist team was putting together for analysis may not be used by the business analysts, and business analysts may not give the right output to the marketers and so on and so forth. And this is not a new thing.

Q: Talk about getting started with AI. For example, how do you choose between building your own, in-home software vs. buying? What about SMBs that can’t afford to invest millions of dollars in top-of-the-line AI software?

A: Don’t compare yourself to anyone else. Facebook and Amazon were the pioneers. They didn’t start yesterday. They started ten years ago. But now things have changed. Now, AI is provided as a service (AI-a-a-S is a thing!). Amazon it was doing it from an infrastructure perspective. IBM is doing it. Microsoft has been in the game for a while. Now any brand has the option if they so choose.

AI is not always  a magic bullet — it’s all trial and error. It’s the same way any algorithm was developed. It’s all trial and error so it always requires a business front. And it also requires a technology backend. And those together is how you get close to that magic bullet.


Industry experts like Raj really help drill down through the clutter about what AI is and what iot can do.

All this to say: there is a LOT happening with tech, data, and the melding of the two. Infinite amounts of data are being created each second on many scales, and the possibilities with what we can do with that data are virtually limitless. But I can tell you this, too: AI is a sort of figurative pair of glasses which marketers can put on to open new doors of perception, to get closer to truth, to knowing the reality of the story that data is trying to tell. ◾

Handpicked Related Content:


AI predicts buying probability, likelihood to convert & CLV. Learn more!

Download the Whitepaper