Episode 215 / Amit Thard / GE Healthcare / Director of Omnichannel Strategy
Podcast: Supercharging Data Science in Marketing with AI
For Amit Thard, Director of Omnichannel Strategy at GE Healthcare, we’re only just scratching the surface of what AI and machine learning can help us with. Especially for making data-driven marketing decisions, Amit considers AI a shiny new object worth learning about.
We’ve already seen plenty of “AI experts” offering to show marketers and business owners how to boost their results using the new technology. However, in reality, the recent AI “craze” can be compared to the beginning of when the internet first came in, according to Amit. It creates opportunities, but you first need to learn about it, explore the depths to which it can provide a competitive advantage, avoid being intimidated by it, and also being “cheated” by it.
Amit believes that “there is so much to learn about AI - if you start learning now, you’ll know how to use it as a powerful tool, rather than wait for it to affect you.” This means being honest about what AI can do for us today and about what we don’t know in equal measure.
At GE Healthcare, Amit can already take large amounts of data about health infrastructure in a country, about upcoming developments and policies, about numbers of doctors in training, and so on. Overlaying this information with traffic data he already has from his marketing insights, he can make better predictions about where company budget is best spent. All of this data is impossible to consume and work through at a human level, however.
This is where the power of AI comes in. As it constantly learns more, its ability to pinpoint data points that a data marketer is missing increases. It will give increasingly better insights and predictions. And, “as we become more intelligent [about AI], AI systems will also become more intelligent. And that creates a need for different data points. And wherever there’s demand, there is always a source of supply.”
Hear more about how Amit is using AI for data-based decision making and listen to his top marketing tips for students and for seasoned marketers, on the latest episode here.
Transcript
The following gives you a good idea of what was said, but it’s not 100% accurate.
Amit Thard 0:00
Focus on the revenue. Every three to five years, there are restructures, there are budget cuts, you want your marketing to be tied to revenue generation and you've got to always have plans for the future. That makes you indispensable to the company and it also means that you're helping the company to grow.
Tom Ollerton 0:16
Before we crack on with the podcast today, I'd like to give a shout out to FUTR Europe, which is a retail marketing and commerce event on the sixth of June 2023. The FUTR conferences focused on fresh thinking and progressive change in the future of retail marketing and econ. You can get tickets at https://futr.today/london/. The FUTR guys have been longtime supporters of this podcast, so please check out the website, get a ticket and get down there.
Tom Ollerton 0:56
Hello, and welcome to the shiny new object podcast. My name is Tom Ollerton. I'm the founder of automated creative, we are a creative effectiveness adtech platform that helps brands remove the guesswork from their marketing. But we're not going to talk about that we're going to talk about the future of advertising. This show is a weekly podcast where I have the absolute privilege of speaking to leaders in this business about where they think the industry is going and this week is no different. I'm on a call with Amit Thard who is director of omni channel at GE Healthcare. So Amit thanks for coming back to the podcast. Could you tell the audience a little bit about yourself if they didn't get to hear that first episode?
Amit Thard 1:35
Sure. Good afternoon from Singapore. So my name is Amit Thard. I'm the director of omni channel strategy at GE Healthcare. I've been in marketing and advertising for a couple of decades now. So I started my career in the agency side in the US and in public relations in Singapore. And about 10 years ago, I moved to the brand side. So started as the Digital Marketing Manager at Philips Lighting and then slowly transitioned into E commerce. So I lead the E commerce at Lenovo, Stanley, Black and Decker. And now I'm leading the omni channel strategy, which is three departments, digital marketing, e commerce and inside sales. So I lead that at GE Healthcare for ASEAN, Australia and Korea.
Tom Ollerton 2:23
Fantastic, right. So varied career, you've worked for church and state agency and brand side, what is the advice that you would give to a student who's looking to make it in marketing?
Amit Thard 2:37
I would say, well, first of all get started. But don't restrict yourself by saying I want to specialize in something. Because things change so fast, that you've got to be flexible, right. And I mean, when I started, I started in traditional advertising, where we were still doing newspaper ads, this is before we started online advertising. And, you know, through happy accidents, I've now come into E commerce. And you know, where digital marketing still is important. But all that knowledge I started getting 20 years ago in traditional advertising still applies today as it did 20 years ago. So I would say get started. Be curious, ask questions. But don't restrict yourself and to a certain part in your 20s is to learn and experiment. So you know, take advantage of that.
Tom Ollerton 3:31
So that's a great tip for careers. But what is your top marketing tip? What is that one bit of advice you find yourself sharing most often?
Amit Thard 3:39
Focus on the revenue. Right? So when you're working in the corporate side, especially in multinationals, right? What I've seen is that every three to five years, there are restructures, there are budget cuts, then there are external factors, right, it could be a recession, or as we see now global war, leading recession or COVID. Anything can happen, right? Typically, in a company, when budgets get cut, the first budget that's cut is marketing. So what you want to do is you want your marketing to be tied to revenue generation, you've got to show your company that the leaders of your company, the finance department of the company, how much return on investment they get for every dollar that they put in, and you've got to always have plans for the future where you say, okay, look, if you give me another dollar, here's how much more I can get for you. That makes you indispensable to the company and it also means that you're helping the company to grow
Tom Ollerton 4:40
Spoken like a true ecommerce professional. I absolutely love that advice. Brilliant how to keep yourself future proof by tying yourself directly to the money
Tom Ollerton 4:49
Right so we are going to talk about your shiny new object now which is Data Science and AI in digital marketing quite how can I do this in 15 minutes? I don't know. But I'm really curious to know why this is your shiny new object? And what do you specifically mean about data science, AI? And in digital marketing?
Amit Thard 5:14
Okay, so look, AI has been around for a while, right. And the perception is that AI has been around in some laboratories, and now it's breaking out into the mainstream. But that's not true. Right. So if you think about it, chatbots virtual assistants, your Amazon's dynamic pricing, which is based on the inventory levels, image recognition from like a simple Google search as a Google or Facebook, or LinkedIn, and targeted advertising, your Alexa voice assistant, these are all examples of AI, which are already in our life. Right? Now, with ChatGPT Coming in, suddenly, you know, newspapers want to sell stories. So they're talking about all the negatives of AI and trying to scare everyone. But like the internet, when the internet first came in, I'm young enough to remember that, or, like social media, when it first started, it creates opportunities, right? But you have to learn about it. So you don't get intimidated by it. And you don't get cheated by it either. So you know, anyone who started in, when digital marketing started out, remember that, you know, suddenly, a whole lot of experts came around said, I'm an expert in digital marketing after doing it for a month. And they initially earned a lot of money, because the people giving them the money didn't know any better. Over time that industry evolved. It's the same thing for AI, there is so much to learn. But if we start learning it now, then we know how to use it as a powerful tool, rather than, you know, wait for it to affect us. And data, I think, look we have access to more data today than we have ever had. Right? So I'll give you an example of how I am currently learning. So I haven't I haven't figured this out yet. But I'm learning how to use AI for data analysis, right? So at GE, I lead the digital marketing for nine countries. But I have a very small budget. So if my boss comes in tells me okay, you know, here's 20 grand, which country are you going to spend it? When are you going to spend it? You know, what's the ROI that we can look at? At this point, I would use my instincts, a little bit of my experience, to maybe make a suggestion. But if I started looking at the data, I could look at a lot more things. So when someone comes to my website, what am I looking for? Immediate data point how much traffic is coming to my site? But what if I look deeper? How much of the traffic is staying on my site? What's the bounce rate? What's the average time spent on the site? How many pages are these people clicking on? That tells me quality of data. Now I take that across the nine countries that I lead, I know where the quality of data is better, where I get better ROI, then I take that over a one year or three year or five year period. I now have seasonality data. I know if I spend my $20,000 in January versus spending my $20,000 in August, what the return will be which country the return may be better. Now if I overlay that on more data, like I'm in GE I'm selling medical equipment, right? So a place which has more hospitals or clinics is where I want to be. So let's say I have data about infrastructure developments that's happening, you know, new hospitals coming up, new clinics coming up, or which country has the most doctors in training. So graduating doctors, now I suddenly have data on where I need to be before my competitor gets there. I overlay that with the traffic data I have with past campaign data I have, and suddenly I know the best use of my $20,000. Now that's a lot of information, right? And so AI can help me give that prediction. But all of that information is something that I cannot humanly consume and come up with an answer for. So that's where AI comes in. And it analyzes everything. And as it learns more the beauty of AI is that it's not like a Google search where you just type in a parameter and you know, get an answer. It's machine learning. So as that learns more, AI will also tell me what data points I'm missing what data points I need to add, and will start giving me better predictions. So that's where I see it being used for digital marketing purposes.
Tom Ollerton 9:59
This pisode of the shiny new object podcast is brought to you in partnership with MADfest whether it's live in London or streamed online to the global marketing community, you can always expect the distinctive and daring blend of fast paced content startup innovation pitches and unconventional entertainment from MADfest events, you'll find me causing trouble on stage recording live versions of this podcast and sharing a beer with the nicest and most influential people in marketing, check it out at www.madfestlondon.com.
Tom Ollerton 10:29
That is pretty exciting to hear. Just tell me how you would do that. So I've got an idea how I approach the same task? Like, are you? Are you using proprietary systems internally? Or are you using the kind of go to, the ChatGPTs of this world? So you're talking about a bunch of different datasets, you've got data coming back from your paid media, you've got an investment opportunity to infrastructure plays, and so on and so forth. Seasonal data, if someone's going to copy your approach and a different sector for different brands, how do you actually go about getting that data clean and into the right tool to help you?
Amit Thard 11:10
Well, so that's the new part, right? So we are still figuring it out. So my company has not started doing this yet. So this is a project that we are in the planning phases for. And that's because honestly, no one quite knows how to execute it just yet. I predict that at least for the first year or two, it will be combination of software like Chat GPT. And you know, a whole lot of other I mean, there are there are hundreds of AI software already out there. It'll be a combination of that and us building these algorithms manually. And over time, what will happen is that either we will start building better algorithms or more likely, someone a lot, a lot, a lot smarter than me will come up with a new system, a better version of Chat GPT in order to do this, so the market creates that need. So I don't think that's out here yet. We are kind of in the MySpace version of AI, the Facebook has yet to come out.
Tom Ollerton 12:18
And what do you think about the datasets that are available? I was listening to a podcast, and they were saying that there's an issue with where the training data is coming from. Right. So certainly, from a chat GPT perspective, it's coming from with a California lens, what they were talking about was do we need local data? That is open source to make sure that we're going to get better answers from these types of services. So for example, if you said, oh, right, write me the LinkedIn posts that target healthcare professionals in Singapore, for argument's sake, right. Unless you have Singapore healthcare, professional specific data, you're going to get a skewed wobbly California version. And actually, what would be more useful is on a country by country basis, you have all of the publicly available data, it's all open, it's all verified, it's all maintained to a degree on a wiki. So then that training data was very accurate. Whereas currently, what we're looking at is an approximation from a global perspective coming out of Microsoft, essentially. So I'm curious to know, yes, this is powerful. Yes, we can combine the datasets that you mentioned that what safeguards need to be in place and what data Best Practices needs to be in place for marketers to know that what's coming out of these services is reliable.
Amit Thard 13:44
So look, a lot of the data sets already exist. They may not be public domain, but they already exist, right? So for example, the GE traffic data that's already there, I have access to that. Now, I don't have access to Lenovo traffic data, because I'm no longer Lenovo but I have access to GE's data. And if you go to even a something as basic as website traffic data, there are a lot of data points there, which we typically don't look at, right. So when we are assessing a campaign, we look at how much traffic came in how much you know, sales happened, we just build that very basic funnel. But there are a lot of other data points, right. So traffic quality can be assessed by bounce rate, amount of time spent on site number of clicks, then, depending on what you click, then you build a heat map. And you can see where the customer the visitors looking at. So the all that data set does exist. If you look at databases have dropped, as you said healthcare professionals in Singapore, those databases do exist, we have to buy them. And there is of course that element of if they don't opt in to receive communication from us because of PDPA laws. We cannot immediately use that data to reach out to those healthcare professionals. But the data set already exists. Infrastructure data that's already there. That's that's public domain, right? We know where Indonesia is spending its money in infrastructure development. We know how many medical schools there are in Philippines, we know on average, how many students there are per batch in that you can get that on Wikipedia. So it's a matter of putting those data sets together, which we have not necessarily done at this intricate level in the past. But that's where we also have to change our mindset. And I believe that as we start doing it, a lot more datasets will become public, right? We don't even know what we're looking for yet. So as we become more intelligent AI systems will also become more intelligent. And that creates a need for for different data points. And wherever there's demand, there is always a source of supply.
Tom Ollerton 16:00
Unfortunately, we're coming to the end of the episode. And we need, I think we need to have a regular follow up on this topic. And it's fascinating to hear you talk about with such passion and conviction about what is possible and to share the fact that you're thinking like this, and the fact you haven't sorted it out yet that you're in the process of learning, which is more than a lot of people on this podcast would have the guts to say, so thank you for that. So if someone wanted to discuss this with you, where can they get in touch with you? And what makes a really good outreach message?
Amit Thard 16:30
Look, I'm on LinkedIn, I'm very active. And my last name Thard is a very rare last name. So it's easy to find me on LinkedIn. And look, if you're in Singapore, I'd love to meet for a drink to discuss this further. I mean, I'm currently doing some courses as well with Singapore management university with the MIT to learn more about data analysis and to learn more about AI. And the people I meet in these courses are from all countries all walks of life. And you know, it's amazing the kind of discussions we have, because we start with asking questions, and then go into these random brainstorming sessions. And that's where the best thinking happens. So, yeah, we'd love to include you in our group. Just look me up anywhere.
Tom Ollerton 17:18
Thank you so much for your time.
Amit Thard 17:19
Thank you. Nice talking to you.
Subscribe to the ‘Shiny New Object’ Podcast on Apple Podcasts, Spotify, YouTube and Soundcloud.