Home Episode 11

Podcast Episode 11

Building Better Customer Experiences with AI and Data

with Heidi Bailey, VP of Futures and AI at The Integer Group

In this Automating the Enterprise podcast episode, we sat down with Heidi Bailey of The Integer Group. She shares her valuable insights on how to add value, make an impact, get people excited, and ultimately Build Better Experiences with AI and Data.

Full Transcript

Dayle Hall:

Hi, you are listening to our podcast, Automating the Enterprise. I’m your host, Dayle Hall. This podcast is designed to give our organizations insights and best practices on how to integrate, automate, and transform their enterprise.

Our next guest is the VP of Futures and AI for The Integer Group, which is a data-driven marketing and brand activation company. I am very excited to have Heidi Bailey on our podcast today. Welcome, Heidi.

Heidi Bailey:

Hey, Dayle. Thanks for having me. I’ve actually been really looking forward to this conversation for some time.

Dayle Hall:

Excellent. So have I. Actually, this is a very different type of discussion that we’ve had on some of the others that have been more- we’ve had some colleges, we’ve had some customers, and a couple of nonprofits, but this one today I’m really excited about just because of the marketing and brand discussions that we’re going to have.

So thanks for joining us. Before we get into the meat of it, why don’t you give us a quick overview of who you are, how you got into this type of background, and who The Integer Group is, and how they’re helping people?

Heidi Bailey:

Yeah, absolutely. So The Integer Group, we’re part of Omnicom, the Omnicom family. And we’re one of the world’s latest growth company, and we’re focused mostly on commerce and retail. We ladder up into the Omnicom family through Omnicom Commerce Group. And then I specifically am the VP of Futures and AI, which is part of our growth sciences team.

So growth sciences at Integer is a full-service analytics offering. We have a suite anywhere from performance reporting, media modeling, attribution modeling, and then what my team focuses on is machine learning, AI, internet of things, those types of activities. I got in my role because I’m always looking for what is new, what is next, what is nobody doing yet.

So I have been in the agency for the past seven years. What my team really does is we sit at the center of the organization, and we sit between our internal client teams, but then also, we sit in between the clients, and we have these meetings in order to find the gaps, what are business challenges that they’re trying to solve for, but also opportunity areas, something that somebody might not have thought about before, like how could we actually solve this using technology data and AI.

We have a lot of different tools in our tool set. We work across conversational AI type skills, voice skills, Alexa, Google Home, that kind of thing. But then we also have data science capabilities, where we build our own solutions and products. So that’s what I do. Yeah.

Dayle Hall:

Wow. That’s really interesting. So it’s various types of technologies, but also you build solutions as well as advising clients, which I think is unique.

When you have customers, potential clients approach you, you just mentioned trying to solve business challenges, how does it typically work? Do they come to you and say, we have this challenge, can you help us with AI to solve it? Or do they come to you and say, look, we have a bigger thing we’re trying to grow in our market, or we’re trying to solve a bigger market challenge, and then you help to show them how AI can help it? Which do you find is a usual approach for you?

Heidi Bailey:

Sometimes they don’t even come with a business challenge. We might just be sitting in a meeting with a client. They said, oh, if we could only do this, or if we could only do that, right? There are definitely times where we are briefed to solve for very specific challenges and business objectives that they have. But there are also times when you just come up with ideas. You just sit there and you’re like, oh, wait a minute.

You could actually do that if you have this data set. And if you don’t have this data set, we actually have some technology that can help you gather that data set to where we could then eventually address your challenge or build something proprietary that will address that challenge.

That’s how we sit there. So I actually have my team sitting across multiple different meetings. I actually have my data scientists in creative reviews and different things like that, that we can find some of these opportunity areas and come up with new ideas, but also to just open up their mind to the possibilities of what can be done.

Dayle Hall:

Yeah, no, I love that. I love having data scientists sitting on creative meetings. I can imagine that’s a very unique perspective. So let’s talk a little bit about that. Let’s talk a little bit about AI using data to drive growth. So you said sometimes ideas come up in meetings. Sometimes there’s specific challenges that a client is potentially trying to solve for.

So you don’t have to name any names, but give me some examples of how you’re seeing certain technologies that you’re using to potentially identify trends or develop insights that help you give better recommendations to the team. So let’s look at the technology piece first. What are you using and how are you using it?

Heidi Bailey:

There are several different types of technologies that we use, obviously, AWS, Snowflake, all of those types of things, in order to automate the data more in real time. But on top of that, we work with DataRobot very closely.

And just to give a quick introduction of who DataRobot is, is they’re an AI cloud provider that really has allowed for us to democratize AI to our teams and clients. And what we really like about them, the reason why we’re working with them, is because they’re really an end-to-end automation tool when it comes to data science. They’ve automated everything from the development work. So they have an auto data prep tool, which then once you prepare the data- and as you know, data scientists spend about 80% of their time just in preparing data.

So they have automated that. You basically hit a button, it goes into auto ML function, also auto time series, where we can instantly start to build these models, all the way through to then also the deployment monitoring and managing of these models.

What we’ve actually found really, really successful in using with them is their no-code applications. So if you think about an AI project, it takes months. And then you spend all this money in development to get to some sort of application in order to run the AI. And then you start to realize that there is something wrong in the code, there is something wrong in the model. You have to go back and do the whole thing again. And so, at this point, you spent all this time building something and spending all this money without showing any kind of gains for the company.

With the no-code applications, you can instantly deploy your models into and experiment within a few days. So this really allows for us to scale our production of AI, but then also, it frees up a lot of time for the data scientists to be able to sit in those creative meetings and different things like that.

So that’s one tool that we use quite a bit. The other one that’s interesting is one called Q. This one was actually built by one of our sister agencies called sparks & honey. And what they have done is they have partnered with a- I think it’s like six K, maybe seven K publishers, where they’re collecting about 90 billion signals a day from around trends, cultural trends, different things like that. And they sit on a ton of historical data.

And so what they’ve done is they’ve built a platform that not only shows you- if you think about social listening or different things like that, not only shows you the trends that are short term or the things that are happening in market today, but you can also then- because they’ve looked at the velocity of trends, you can actually predict two years out.

When we’re putting growth strategies together for clients, we use tools like this alongside DataRobot and some other tools that we have in order to then understand what should we be focusing on today, but what is the lasting power in the market two years from now.

Dayle Hall:

Yeah. Those are interesting. What we see in our business, potential customers and customers’ approaches, they’re looking to do something similar with using our tool to get better access to data. It’s very similar in terms of a local. So they’re usually solving an immediate challenge.

And you just mentioned showing them things, which are potentially a couple of years out, how do they take that? Because usually, they’re like, well, we know we have to solve this now. When you show them what’s possible, or you show them some insights into a few years in the future, no offense, do they buy it? Do they think it’s all made up? How do you make sure that they really can appreciate or buy into some of the projections that you’re showing?

Heidi Bailey:

Truthfully, it is definitely a lot easier with some clients than with others because some are more open minded than others. But I’ve always found that if you’re very transparent in regards to how these tools are built versus just saying, hey, here’s another AI tool for you to buy, or for us to use, the way that we set it up is we are super transparent in how all of the products are built on the back end, the data used, the technology that is being used.

When we explain it in that way, then it makes it much easier to then say, oh, okay, now I see how you can predict that trend two years from now or how you can forecast my budgets two years from now. That’s how we approach it.

Dayle Hall:

Yeah. One of the other podcasts that I’ve done for this series specifically, one of the discussions we got into was don’t necessarily trust a vendor or a services organization, or whatever, if they won’t tell you how they’re using AI, and how it’s built, and the data that goes in.

If they call it the secret sauce, generally, you’re not going to get customers or clients bought into the data that you’re showing because can they trust that your black box is actually going to be the right output? So I love the transparency with AI. I think that’s something that some of the companies that I’ve worked with and for that have this kind of tool, the more transparent they are, the easier it is to get clients on board and actually stick with the findings. I’m sure you find something similar.

Heidi Bailey:

Yeah. And with some clients as well, especially some that are still a little skeptical, we actually do involve them in the process of building models if we’re building something ourselves. In terms of we put the training data set in front of them, they get to review it, they understand the dimensions that are in it, so that ultimately, they also understand what the AI then delivers in the final end. That’s been super effective.

Dayle Hall:

Because of The Integer Group and the focus on marketing, branding, customer experience, that kind of thing, what kind of work are you doing? What kind of considerations should a marketing, or someone that’s responsible for customer experience, what should they be thinking about as they’re thinking about using automation, or AI, machine learning to solve key challenges?

I know we touched on this a little bit at the start about how people approach you, but typically, when you’re sat with, I don’t know, a CMO or someone that’s responsible for customer experience, what baseline do you set with them? What guidelines or roles that you say, okay, so this is how we’re going to begin. How do you start that conversation?

Heidi Bailey:

It all depends on the client and obviously what they’re trying to sell for. But in terms of using AI, automation, the way that we approach clients on how they should start thinking about stuff- and if you think about it truly, automation, AI, machine learning, and the technology being used in the marketing industry has always been pretty advanced.

Think about programmatic media. You set your parameters instantly, these ads are being served to consumers wherever they may be. It’s not a new thing to be able to use some of these technologies in the marketing industry, but it is new in the way that you’re starting to think about it even from a personalization perspective, if you think about it.

It’s not new, it’s not a new concept. It requires a lot of data. And personalization has been around for a long time. If you think about, you still have to segment customers, you still have to understand who your target audience are, so you can go out at scale and target them. And then you could use dynamic creative optimization platforms to then deliver all these additional personalized banner ads to people.

However, we’re approaching it from a different perspective because we’re saying, instead of letting them machines actually make those decisions for you, once you’re executing, you should actually already know what is going to drive success before you execute it.

So instead of letting a dynamic creative optimization platform make all the decisions for you, you should already have a really good understanding that if you target this audience, these are the types of conversions you’re going to get. And if you hit them with this type of creative, if you change the color from red to blue, for example, how is that going to drive additional conversions, or is it not? We’re trying to actually get ahead of the- before you even hit those platforms, to have a really good understanding of what you need to execute.

Dayle Hall:

That’s interesting, because that then makes me think that using these technologies helps to augment what marketing is doing and what they should be doing, and not necessarily replace what they’re doing. And I think that’s a very key point. But when I do these podcasts, I try and make sure everyone can take something away and not just be like, well, that was interesting, but I don’t know what to do with it.

But I think that’s a really key point, which is if you’re in marketing, you should know the things like things you want to serve up, what kind of conversions to expect. And then you use AI to help you scale, become more efficient, or whatever, but don’t go into this with a, hey, we have a problem, AI is going to solve it for us, which I think is a very key difference.

You’ve seen the chart like I have, how many MarTech tools are there now? 7,000, 8,000, right? You’ve seen that. How many of them do you think mention AI now? Probably a lot.

Heidi Bailey:

All of them.

Dayle Hall:

All of them, right?

Heidi Bailey:

Yeah. They mention whether they’re doing it or not. They do mention it for sure.

Dayle Hall:

Yeah. Yeah. For sure. For sure. But I like that differentiation, which is you understand your own processes, and then you use AI to potentially augment that. So I know you work with a lot of clients, and I’m sure very, very impressive retailers. Do you have some examples of how they may be using- some of these people are using that AI in tandem with some of the work that they’re doing to potentially develop some insights, or better engagement opportunities to augment what they’re doing in their process?

Heidi Bailey:

A great example is, and we’re just riffing off of what we were talking about before as well, but it’s like how to use AI to augment what you are already doing. I look at it as AI fighting AI, if you will.

And so as an example of that, it’s like you have all the big media giants, you’ve got Google, Facebook, Amazon, walmart.com, I could go on. But all of these platforms have their own proprietary AI tools that they’re using in order to execute media. And they give you very, very little insight into how it works. Just as an example, you will execute a social campaign one week, and it does amazing. You have really great engagement rates. You have lots of shares. You have lots of likes, all this kind of stuff.

Maybe a couple months down the road, you’re executing a very similar campaign, similar audience, very similar type of creative unit and it tanks. And it does horrendous. And you spent probably more money on that one than you did on the previous one because you were expecting success. There’s really no understanding of why that’s happening.

What we did is we said, okay, well, we’re sitting on- we’ve been executing Facebook social media campaigns for years and years and years, could we actually use our historical data to have an AI find the patterns of what might potentially happen so that we can actually predict the success of future posts? That’s exactly what we’ve done.

It’s all that we don’t have fundamental understanding, or a calculation of their algorithms, or understanding of how they’re doing this, we can now have the AI look at future posts going live, and then actually give us a prediction score on whether or not it’s going to be successful. So that’s how we’re looking at using AI to fight AI.

Dayle Hall:

I’m sure we can have a whole separate podcast around AI fighting AI. I like that. I like that terminology.

Heidi Bailey:

And you typically see that in IT and security services, because it’s becoming so sophisticated that you almost have to execute AI, so in that capacity to fight all of the people trying to get in. But I have never heard of it being used like that in marketing. So it’s really fun.

Dayle Hall:

Yeah. So as we move forward in thinking about- that’s a really good example. And obviously, there’s a lot of- whether it’s in marketing or specifically customer experience, I think pretty much anything that has a workflow or process, that can be automated.

If I’m sat there, I’m thinking like, okay, this is a great discussion that those two are having. But how do I go back to my organization and say, how do I identify these processes? How do I apply? How do I identify which tasks can be automated or where AI can help? If you were going to advise someone fresh out of the gate, where do they start? Do they start with marketing? Do they start with any process? What’s a good first project?

Heidi Bailey:

It’s really having a good understanding of who to talk to in your organization in order to understand where you’re going to start. And that’s basically just having conversations with key stakeholders within each one of the different departments.

And then you just start to interview them. It’s just a natural conversation, like, hey, what is it on a daily basis that you’re always having to do that’s driving you crazy, that you wish you didn’t have to do and you could spend more time on something more valuable? And so it’s those kinds of conversations.

And just to give you an example of one that we had, and this one was actually- it just came up. It was one where we weren’t actually trying to interview different people inside the organization, but I happened to be on what we call- it’s our HR department, but our people team inside of Integer. And I was on a call with them. And the person was like, man, I wish I had more time to spend on more emotional, the conversations that actually matter. And I said, well, what do you mean by that? And she was like, all day, all I do is answer the same question over and over and over again, like where can I find this form and where can I find my pay stub and stuff like that?

So a simple solution to that then is, okay, well, these are all non-sensitive questions, can we quickly create some sort of HR chatbot or people chatbot? You can just ask it and say, hey, where can I find this form? And it automatically responds with the link. That’s the most simple kind of example that I can give you.

But then those are the types of things that now free up their time about five hours a week, where they can now go and spend more valuable time with people, really understanding them, spending more time addressing sensitive emails and different things like that. And so that is a big efficiency gain.

Dayle Hall:

Again, one of the other podcasts in this series, I actually talked to someone who does people analytics. So I was amazed at how many- that the AI and automation and the data analysis that they’re looking at now for people. He came from one of our customer organizations and just joined a massive event company called Freeman. And they have thousands and thousands of people.

And how they’re using the data now to identify whether people are responding to internal surveys. Are they engaging more online? What are they doing? To try and identify things like, are they likely to churn out of the business, and how much can we save if we could identify that and stop the churn?

And that’s all just as you pointed out, which is really looking at the processes, the things that drive you insane. And this goes back to what you said at the start. That may not be a business challenge, but it’s definitely a role challenge. It’s something that someone is tasked with every day that they’re driving them insane.

Heidi Bailey:

Yeah. Because if you think about it, most people leave because they’re doing uninteresting work. If you’re no longer doing interesting work, you have a tendency to leave. And what is considered interesting work? It depends on your role, obviously, but everyone wants to not sit there and have to do the same thing every day and be repetitive.

Dayle Hall:

I wonder whether in the future and you interview for a new job, whether you are asking what kind of AI they have or automations in their organization to let you do your job better. That could shape whether you join that company or not. Are you doing everything manually? Or do they have those technologies to help you?

Let me ask you a question, because, obviously, when we see different people that we work with, different customer profiles, but typically they’re either within the IT or within the line of business. So they’re either trying to solve more of the bigger IT challenges and they’re looking at scale, or there’s a process, a role, or something within a line of business, HR, marketing sales, finance, whatever it is. And then they approach us and say, look, we’re trying to get better access to this data.

In your experience for the work that you’re doing, do you find that it’s line of business that approach you? Is it IT? And then when you actually start with your projects with them, do both groups come in? Is it becoming a bigger challenge where more people from an organization are involved in solving these challenges? Or does it usually start and end within a specific group?

Heidi Bailey:

No, it is multiple people, and especially from a marketing standpoint. And especially because we’re a global agency, we’re headquartered out of Denver, what you see is you have various different people from various different clients and backgrounds coming to us. It does not start at the IT level. Especially because our IT is centralized at the Omnicom level, you’ll have these different functions coming.

So one day I may be talking to our people team. The next day, I’m talking to our account team. The next day, I’m talking to our creatives. The next day, I may be talking to our media team. So it definitely varies. And then from there you have all the various different projects that come up, and then you also have the availability to have lots of different data from different people, which is interesting.

Dayle Hall:

That’s interesting. We’ve talked about how to use it, potential processes. You’ve given us some good examples. Let’s talk about how you implement these things and how you guide customers on implementation.

Because a few years ago, maybe longer than that now, but there was always this tendency to pontificate or think that using AI was going to put people out of work. When I think as you described just a few minutes ago, what it actually does is it frees people up from the mundane, allows them to be more creative in their role, but it doesn’t necessarily mean that their role goes away. In fact, they can actually work on bigger, more creative tasks.

When you start to work with some of your clients, how do you approach those challenges, and how do you approach things like change management? Because change management is not a dirty word when it comes to customers, but not often do people think about it ahead of time. So how do you approach that?

Heidi Bailey:

Yeah, absolutely. So I think first and foremost is you have to have a good strategy in place. Because a lot of the AI projects, you have a massive vision. People are like, oh my goodness, here, I’m going to have this AI chatbot, I’m going to do these things and whatever. And it takes a long time to deliver those types of projects.

So how do you add value very quickly? How do you create almost like a crawl, walk, run type scenario for clients, and say, okay, within the crawl phase, these are the things that we’re going to deliver? But make sure that- and what we look at is what are the things that can bring the most impact right away and that are flashy and fun? Because people get really excited. That’s how you drive. It’s almost like storytelling with your products.

We do it in that way, but we also, like I said before, is we involve them in the whole process so that they have a really good understanding of how the models are trained, what the output is going to be, and then also how they can work with it, and how it’s going to help them.

And because they’re involved in it, just to give you an example of the social use case I gave you before, is the social team was involved in the entire building of that process. We actually took their input in how they would use it to help them do their job better so that the final output ended up being a tool that would help them understand what day should we be placing our social content, on which platform should we be placing it, what should the image be, what should the text be. It’s optimizing all of that. So even though they’re still writing the copy and text, and they’re still creating the images, they now have a tool that allows for them to look really, really smart in front of their clients.

And then, at the same time, we provide them with the marketing materials to then be able to take that and say, hey, client, look what we did. And it makes it much easier. Yeah.

Dayle Hall:

I love the concept of we’re trying to solve a problem for a certain group of people, obviously, bring them in throughout the process. That seems like elementary, but I’m sure sometimes people higher up in the organization just feel like we’re going to make this happen regardless.

And actually, I really like what you said, which is the quick wins, the flashy wins. It might not be the ultimate use case, or it might not be the bigger plan, but getting those immediate- I don’t want to say flashy, but getting those quick wins that potentially have big visibility, whilst it might not be the most complex thing you’re solving, but I can imagine that would bring people along on that journey. And you mentioned it yourself, the quick wins, the crawl, walk, run. We’ll get some wins in the crawl phase, then you’ll find the organization supports it.

Heidi Bailey:

That’s how we’ve also seen people who initially didn’t want to work with us. So we’re a little bit skeptical about AI. You start to see them gain interest over time when you do it in that way too, just because they’re like, oh, wait a minute, that’s not what I thought this was going to be.

Because typically, when you think about it, data is not the most interesting topic. People think of Excel spreadsheets and, oh, they’re going to deliver another dashboard. When you can show them flashy things, or have them along in the process, and they actually see the power of data outside of just the number itself, then it gets them really interested.

And then going back to change management as well, the other thing that this then does- and this actually happened naturally, I didn’t even- the first time we did this, it wasn’t intentional. But what ended up happening is as we were bringing these teams along, we actually upskilled them. We start talking about AI, machine learning. I mean, obviously, they’re not experts, but they have the vocabulary now to when they’re having additional conversations with people, we can’t be in the room or something like that. But then they have a fundamental understanding, can still answer all those questions. And so it’s almost like you’re upskilling your team members and staff and others by doing this as well.

Dayle Hall:

Yeah. When we talk about the social media example, I talked about lines of business, and then there is the more complex types of AI and machine learning that you use, which, to some extent, requires a little bit more expertise, data scientists, for example, because they want to do something more complex.

Do you feel that what’s happening with AI, machine learning, automations, do we not need data scientists anymore in organizations? Or do we need them less, or do they focus on other things? What are you seeing with clients? What are they using the data scientists for if you are helping some of these business problems to be solved?

Heidi Bailey:

Maybe not all people think this way, but if you think about AI enhances what we do, fundamentally. Every tool that I build, every tool that my team builds, I ensure that there’s always going to be some sort of human aspect to it.

So if you approach it in that way, what happens to a data scientist- and there’s a lot of stories that have been out there recently that in 10 years, data scientists, data analysts are going to be obsolete. Because you look at tools like DataRobot, which have automated the process and are super easy to use. You also see things like ThoughtSpot, which is more analytical tool, but it’s searchable data. Do you need an analyst anymore if an AI is creating the graphs and analysis for you automatically just with you searching something?

But the thing is, the jobs are just going to shift. You still have data scientists, but instead of the data scientists actually writing code and creating the applications and all of that kind of stuff, they become interpreters, or they become the people who ensure the AI is sound from an ethical perspective. What is the AI’s output? Is that the right output that we’re wanting? Why is it making these kinds of decisions? Can we go back and look at it? So you start to move into more that explainability role for a data scientist, because you still need those skill sets in order to interrogate the models. You have to have those skill sets, even if there’s an automation platform that’s doing it, somebody who can get behind the code and look at it.

So it’s just a shift, like a creative person still going to be a creative person, but they’ll have AI tools to enhance what they need to do. As an example, just look at like OpenAI’s GLIDE tool. Essentially, you can take text and write, hey, I need an image, a couch sitting in front of a fireplace, and I need the couch to be red, and it needs to have a coffee table, and maybe a lamp, and all these different things, and books or whatever. Translates that, automatically gives you lots of images to choose from.

But as a creative person, that might scare you because you’re like, oh my goodness, this thing has just automated away my job. I’m not going to need to create anything anymore. But you still need the fundamental understanding of what is it that you wanted to create. Because you can’t just have a person sitting there saying, oh, I don’t know if I need a coffee table or not. I don’t know if I want books there. Because you still need that design and artistic aspect of it. So they just become tools to enhance what you do versus taking the place of what you do. But it is a shift in how you think, for sure. And I think that’s the difficult part.

Dayle Hall:

Yeah. And I like what you said, which is there’s always a human aspect to this. And we talked about AI, this is augmenting what we do. I see more and more examples of that in the work that we do, the partners, and the vendors that we work with. But have you seen any reticence to this?

Again, we talk about it like, hey, it’s so easy to understand. It’s just augmentation. It’s not replacing you. But have you seen companies that you see some of that reticence or like, I don’t need that support from my business? And how do you approach that? Or people embracing it?

Heidi Bailey:

No, it is definitely difficult to get some people to embrace it, but I think the more you talk to them- and again, I think it goes back to the involvement piece of it too, because they get excited. You educate them on what it can and cannot do.

Because right now, fundamentally, AI is built, it’s trained to deliver specific tasks. I mean, yes, there are companies that are looking at cognitive AI and different things like that, which would give it a more person capability versus what most of the AIs are meant to do today. If you look at Hollywood and all the different things that they’ve done to drive Terminator-like experiences, where AI takes over everything, in the back of their mind, that’s what they’re thinking.

So you have to overcome that. I think, again, it’s just involvement, education, transparency, and how things are built, and then just figure out ways to get them excited about it.

Dayle Hall:

Never let reality get in the way of a good story. I think that’s Hollywood’s response to it. You just mentioned something as we start to wrap up here, but you mentioned something specifically around AI and the ethics of it. We could talk about the ethics behind AI and bias and so on. It could be a whole separate podcast. But I’d just be interesting to get your take. Does that come up with clients? Is that a concern? And how do you potentially handle that if they have that concern around there could be some bias within some of this technology and the data?

Heidi Bailey:

I think it’s so important to get into explainable AI. This then goes back to how are you translating those models or how are you interrogating the AI that you’re building? And I think it starts right from the beginning, taking a look at the training data set, understanding exactly what is in that training data set.

And it’s a very interesting conversation. I actually had that with an internal person a couple weeks ago. But it’s also having to understand, if you have a target segment,  let’s just say you’re doing an advertising campaign and you’re saying, I want to hit this target segment. But underneath it, all you know is you’ve built a taxonomy of people that like to potentially buy coffee, drive this type of car, have this kind of house, of income, those type of things.

But do you really know what the makeup of those people are? Sometimes that level of transparency isn’t there. So we’ve been trying to figure out- we’re not quite there yet, but we’re trying to figure out, can you actually drill down into and get a really good understanding of the percentage breakout of who are you actually targeting? You may be targeting people who like coffee and have all these different attributes, but are you really getting in front of the people that you actually need to.

It starts with a training data set. It takes with full transparency of how the models are built so that you have an understanding of also what the output will be. So,being able to then go back and explain that I think is what has to happen.

Dayle Hall:

Right. Yeah. I’m sure that explainable AI is a good terminology for it because I think that way, we’ve talked about this earlier, which is then you get people bought into it. People believe it, which is let’s face it, we’re trying to do this so we can be more successful, not just help our lives and our processes, but we’re trying to be more successful in our roles. If there is an understanding of how these data sets are built and how the AI is used, it’s going to be much easier to get buy-in across the organization.

Heidi Bailey:

Yeah, exactly. And in addition to that, obviously, you have to have diverse teams, and you have to have them part of the conversation. And this also goes back to not just your team itself, but when you are in the room, when you are starting to build out a use case or you’re starting to figure out how to solve a problem, you have to make sure there’s different functions inside that room as well. So just having different types of people and different types of functions with different types of perspectives is incredibly important.

Dayle Hall:

Yeah. Well, look, as we wrap up, I have one last question for you, and I’m sure over the last few years, you’ve seen a lot of developments. I’m sure there’s a lot of things that you’re very excited about. But if there was a couple of things or one thing that you’re excited about over the next, it could be 12 months, it could be 10 years, what is the one thing that- within this area of AI and automation, using data, and machine learning, what’s the thing that you’re super excited to see come to life or occur over the next 12, two years, whatever?

Heidi Bailey:

It’s interesting because if you think about marketing and advertising, from the beginning of time, it’s always been to change consumer behavior. I’m going to give you a coupon so that you go by my product, whereas you wouldn’t have bought it without a coupon, or I’m going to put this ad in front of you to change your mind that my product is better than somebody else’s product.

And so we’ve been exploring the idea that what if- you’re not trying to change somebody’s behavior because that’s very ineffective and not efficient at all because you have to bombard somebody with 11-time frequency in order for them to finally change their mind. But what if you can use AI and technology to have a really good understanding of who your customer actually is, your consumer? And then instead of trying to make them change their behavior, you are creating products by having that understanding to support them instead.

And so along that line, one of the technologies we’ve been looking into or one of the AIs is emotional AI, which is very interesting when you take a look at research and different things like that. Emotions typically drive a purchase of some kind, and almost everything that we do in life is driven by emotion. So if you have a really good understanding of the emotional aspect of your customer or consumer, but you also have an understanding of what kind of emotion is actually going to drive an action, then you can start to play in that space.

And so we’ve been working with a company called Cognovi Labs, who are doing some very interesting things in this space. So it’s not just about understanding emotions from a social listening standpoint, like net sentiment and passion intensity. But it’s understanding, is anger the place that we can actually drive the most action? But it’s a good anger, it’s not a bad anger. Or is it anticipation, or is it excitement? It’s not all about brand love. It’s not that you’re always- because if you’re happy with something, you’re not going to change anything.

And so that’s really interesting. And then, how do you quantify that? How do you quantify emotion and how it affects your inversions and revenue is something that we’re really excited about. And I think it’s going to have a pretty big impact in the next year or two.

Dayle Hall:

Yeah, no, that’s interesting. Definitely, as you mentioned, emotion is generally behind a lot of buying decisions, our buying behavior, and how can you leverage that for good, not said in a bad way, but to help us. Because at the end of the day, when we’re making these decisions, we don’t buy things that we really don’t want to. But if there is a way of encouraging us to do it, we’re already there, but sometimes we just need to see it in a certain way. That’s very interesting.

Well, look, I got a lot from this. We talked about making sure you have some good business challenges. I love the idea of actually having some of these data scientists, certain parts of the creative reviews.

We talked a lot about AI as an augmentation, which I think was one of the critical things that we discussed today, which was it doesn’t replace anything. It doesn’t mean if you don’t understand who your customer is or the profile that you’re going after or what they want, if you don’t understand that, AI won’t fix it. But AI can help you automate and get more out of it.

I love the concept around AI fighting AI. That’s a term I think I’m going to hear a little bit more in the future. And one of the things that I really liked what you said is around the augmentation, but the fact that there’s always a human aspect to this. Using AI and data can actually help us be more successful. And I can’t wait to see what you’re going to post around emotional AI at some point in the next few years.

So this was a great, great discussion. I’m sure we could- when you have something else to say around emotional AI, then maybe we’ll reconnect and put a whole podcast behind that. But Heidi, thank you for your time today.

Heidi Bailey:

Yeah. Thank you for having me.

Dayle Hall:

Okay. That’s it. Thank you, everyone, for listening. We’ll see you on the next episode of Automating the Enterprise.