Understanding Use Cases & the Ethical Implications of AI in Marketing [Q&A with Raj Balasundaram]

0
65 views

Share on LinkedIn

Global business concept.

I sat down with SVP of Artificial Intelligence at Emarsys, Raj Balasundaram, to learn about all of the ways marketers are using AI today and where AI is headed for the future.

AI is sparking a current of change among marketing teams — helping all e-commerce organizations who use it create proactive customer experiences and communicate with contacts before it’s too late. In short, I learned that AI enables truly predictive or proactive marketing.

When I asked Raj about the specifics of this new and general capability of AI, what he shared reshaped my perspective of how companies are working “hands-on” with artificial intelligence. If you’d like to get a deeper dive, you can check out my two extended/full-length interviews with Raj which we ran on the Marketer + Machine podcast.

As you’ll learn in this Q&A, AI has what Raj called a “huge runway” of potential, but moral and ethical considerations can’t be forgotten.



We discussed three major points:

  • Examples. How are real companies finding real results with AI?
  • Content. AI, algorithms and data are the foundational layers. But how will marketers be able to create enough high-quality content to feed into the machines for use?
  • Ethics of AI. Should algorithms be free to raise prices of fire extinguishers even in the midst of a massive forest fire outbreak? What’s the best way to mix the marketer and machine?

Without further adieu, enjoy the conversation.
_____________________________

Q: Offer three modern (2019) use cases of AI marketing — focusing on this whole “reactive to proactive” angle?

A: Let’s look at three wonderful use cases that are top of mind for me: BrandAlley, City Beach and another brand I like, Air Asia.

BrandAlley is a flash-sale oriented e-commerce company (and Emarsys client). They’re doubling their revenue every year. Why? They’re treating data like gold. They know their biggest asset is their consumers and the data they collect about them. They focus on making their data clean, usable, trainable, and predictable.

Over the past eight months, they’ve been predicting which customers are going to churn before they actually do. They’re using AI to make really accurate projections about who is likely to churn based on purchase history. So, they can know for a fact “Michael will churn after a 45-day period of dormancy” where as maybe Raj has a much bigger window within which he buys, like twice per year. If I deviate from my pattern, that’s completely different from you and your propensity.

Every customer deviates from their buying patterns in different ways.

So their solution, which worked wonders, was to essentially message you 30 days in advance of the time period you’re likely to defect. They can catch you before you’re about to fall off the wagon! How? By giving you content that’s prepared with some special treatment, incentive, or personalized content for you — as an individual — urging you to come back and make a purchase.

When they moved from a reactive strategy to the proactive approach, the same campaigns began giving them 4x in revenue within half the time period.

It’s a mistake to only communicate with customers once you realize when they are deviating from their patterns. BrandAlley epitomizes this whole “getting ahead of the curve” idea and obviously it’s paying dividends for the forward-thinking company. I love this example.

CityBeach is an Australian retailer that has a really cool way of connecting the offline purchases to their online audience. But that’s not even the best part.

They integrated AI throughout the entire customer lifecycle (in-store and online) to identify who’s going to buy, who’s not going to buy, and who’s actually going to churn. Results were phenomenal. In-store and online sales went up.

And, this is beautiful proof of what can be done when you can communicate to customers ahead-of-time using predictive technologies!

Lastly, Air Asia is an airline (I’m actually friends with their CTO) which does fantastic work around the ancillary sales where they basically predict what you’re likely to buy when you go on a certain route.

Let’s say you’re flying from London to Switzerland (this isn’t actually one of their routes, but just giving the example). If you buy in the months of, let’s say November to February, they can surmise you’re probably going for a ski trip if you’re traveling alone or if you’re travelling as a couple. How do they know this? They also look at:



  • Past purchases and previous trips
  • Time and kind of travel
  • Type of person you are

Armed with this information, they know the most important ancillary offer for you at this specific stage is an extra a snowboard check-in or a ski check-in. And they offer that at a very good price.

So, again, the predictive technologies can be applied at different use cases. But it boils down to one important thing — how do you translate these predictions to a business application?

Q: What does the future look like for AI and what do marketers need to be thinking about?

A: Let’s talk about the immediate future since it’s most relevant.

Marketers should be thinking in terms of strategy and content, and how they’ll incorporate all that AI can do. Here’s the thing: there are hundreds of use cases which I could highlight. But that doesn’t take away the heavy (but exciting!) “burden” on marketers who still have to figure out what content to put together for every stage.

If you really wanted to implement all of the use cases AI can offer, you would need hundreds of copywriters! So there’s a big runway — there’s a huge runway — for AI to improve itself with content.

Now, there are companies out there that can help with different aspects of content. Our partners like Phrasee and Persado, for instance, help with subject line. It’s amazing work, but the problem is making this stuff available to everyone. We need to make sure AI is stepping into the content shoes where it’s doing two things:

  • Reducing the burden of content on marketers
  • Properly interacting with the consumers almost like an interface or an intermediary layer which understands the marketers emotion and the consumer emotions.

The question becomes less about the AI, and more about producing content to feed the machine at a one-to-one level.

Q: Clearly, there are moral and ethical implications of AI. How should marketers be preparing for or managing these kinds of concerns?

A: Any algorithm has to be trained (when it’s written). When you train an algorithm, you’re actually training on certain data sets and certain biases of a person. This can have unintended results.

It raises the question: do we need somebody who’s training these algorithms to come from a completely multicultural environment? Or do we have to make sure that the machines are able to consider such subjective things such as what should be vs. should not be done? Sometimes accidental things can happen.

For example there was a feature in Wired magazine about Amazon’s fire extinguisher sales during the California fire. Because they have demand pricing and surge pricing, the price of fire extinguishers went through the roof. That’s exactly what an algorithm is supposed to do. The algorithm is supposed to jack the price up because the item is in demand. But if you think about the ethics behind it, I’m not sure this is the best strategy… it’s the most common thing somebody needs when things are on fire. So, is it ethical to let the algorithms do that? I’m not saying that the business did it on purpose. It is just an algorithm which has gone rogue.

How do we put guardrails on stuff like this? Is it is the responsibility of the brand or is it the responsibility of the vendors? Or is it the responsibility of the marketers themselves?

Not many people talk about ethics of AI because many don’t understand it. At some point — which I believe is not long in the future, let’s say the next three or five years — there will be some form of GDPR-like regulation(s) around AI. We’re not talking about some simple stuff here. We’re actually talking about deep data analysis about people. This has far-reaching implications. Obviously, for these reasons, I’m a very passionate advocate of ethics for AI!

_________________________



The hard results that companies like BrandAlley and CityBeach have seen when employing AI are hard to deny. The technology works when used right. And it’s helping marketers be more proactive when connecting with customers.

But not everything is as wonderful as we wish, just yet. Marketers are still facing hurdles that are arising as the domino-like effect of AI continues pushing us out of the status quo. Mixing content with algorithms is a persistent challenge, as is balancing the ethical dilemmas that come with inherent biases of programming a simulated intelligence.

With time, patience, and continued innovation, though, the future’s looking brighter by the day for marketers ready to explore AI.

Grab your free copy of our latest white paper called “Making Immediate Impact with AI” to learn more about how AI is helping marketers be proactive.

LEAVE A REPLY

Please enter your comment!
Please enter your name here