MEDIASCAPE: Insights From Digital Changemakers

Navigating Tech Transformations: Saurabh Kumar on AI, Data-Driven Marketing, and Building Meaningful Connections

• Hosted by Joseph Itaya & Anika Jackson • Episode 44

Share episode

What if technology could be your ticket to navigating life's biggest challenges? Join us for an inspiring conversation with Saurabh Kumar, Senior Manager of Data Science at Kraft Heinz, as he walks us through his remarkable journey from a web developer in India to a tech-savvy leader in the U.S. Saurabh shares how his early career at Wells Fargo, amidst the 2008 financial crisis, set the stage for his success by integrating tech expertise with business strategies. Listen as he unpacks the secrets behind leveraging technology to forge impactful career paths, with insights that resonate well beyond the boardroom.

Get ready to explore the cutting-edge fusion of data and brand marketing. Saurabh reveals how data-driven tools are reshaping marketing strategies for companies, from industry giants like WhatsApp and Meta to the ambitious smaller businesses. Discover the importance of innovative brand tracking and the art of linking campaigns to product metrics. Be inspired by how Kraft Heinz managed to save a whopping $35 million by optimizing their marketing mix model, illustrating the undeniable power of strategic budget reallocation. We also share practical tips for small businesses aiming to make a big impact with targeted marketing and A-B testing.

AI isn't just a buzzword—it's transforming the marketing landscape in ways you might not expect. Saurabh breaks down how AI and machine learning are playing pivotal roles in everything from Uber's driver activation to creating dynamic podcast content. Hear about how generative AI is revolutionizing customer service and marketing creativity, and how customer segmentation is more crucial than ever for business success. But it’s not all algorithms and data points; Saurabh also emphasizes the importance of kindness and relationship-building in the digital age. Learn how a proactive, genuine approach can lead to lasting personal and professional growth.

This podcast is proudly sponsored by USC Annenberg’s Master of Science in Digital Media Management (MSDMM) program. An online master’s designed to prepare practitioners to understand the evolving media landscape, make data-driven and ethical decisions, and build a more equitable future by leading diverse teams with the technical, artistic, analytical, and production skills needed to create engaging content and technologies for the global marketplace. Learn more or apply today at https://dmm.usc.edu.

Speaker 1:

Welcome to Mediascape insights from digital changemakers, a speaker series and podcast brought to you by USC Annenberg's Digital Media Management Program. Join us as we unlock the secrets to success in an increasingly digital world.

Speaker 2:

This is Anika Jackson, one of the hosts of USC Annenberg's MS in Digital Media Management podcast, mediascape Insights in Digital Changemakers. I am very thrilled to have Saurabh Kumar here. You have a storied career in multiple large organizations that everybody knows about and uses probably on a daily basis large organizations that everybody knows about and uses probably on a daily basis and we actually didn't get to meet at the AI4 conference, but I hunted you down because you spoke about AI and marketing and this isn't talking about Gen AI. This is talking about data. This is talking about a lot of the other side that we're using artificial intelligence for. So thank you very much for being here. Absolutely, annika. It's a pleasure.

Speaker 2:

Now, what I think is interesting is that, of course, the Gen AI conversation is what got a lot more people talking about AI. Of course, ai is a little bit bigger and something a little different. Gen AI is just this little piece of it, and when we're incorporating it into our classes with my grad students now, I try to incorporate different AI tools in every course. So if we're looking at buying measurement analytics right, we'll talk about those. If we're talking about creating content, we'll look more at the Gen AI side. But I'd love to hear a little bit about your background and how you became interested in data science and AI, and when the big companies really started working with AI.

Speaker 3:

Absolutely so. My background I have an undergrad education in electrical engineering from India, and after finishing my undergrad I went into IT industry, as most of my classmates did. So I started for an American bank which was establishing their operations in India in a town called Hyderabad in Uttar Pradesh in 2008. So I started out there as a web developer. So I used to build web applications for Wells Fargo in India, and my web applications primarily catered to the employees of the Kattu. It was inward facing. So one of my first projects there was to build the performance management system to be used by the internal employees of Wells Fargo in India. I also built an e-commerce web portal for the employees in the company to be able to use that, to use the reward points instead of cash to buy products, to buy products that were souvenirs, that were branded with the Wells Fargo logo and such right. And I would have done this from scratch, like right from zero. There was nothing and there was a living and breathing website which processed thousands of orders over the time. So it was definitely really satisfying and that was my first sort of introduction to technology in a professional setting, and I learned how to design databases, how to make a web application, talk to a backend how do you deploy an application and all the good stuff of the technology?

Speaker 3:

And during the later half of my career, I was on the DevOps side of things, where I was primarily overseeing the governance of the cadence at which code should be released into production. What are the different stages through which the code needs to be butted before you can deploy the code? And this was for a bank called Vekovia Bank in the East Coast. Whilst we're going, 2008, the world was breaking down, financial crisis deploy the code. And this was for a bank called Vekovia Bank in the East Coast, wells Fargo. In 2008, the world was breaking down, financial crisis. They acquired a bank called Vekovia and I came to the US and I worked with Vekovia to try to integrate, embed those governance processes for a business called Wells Fargo Advisors, which predominantly were wealth advisors who were working with clients across the US. So that was my introduction. That's how I got introduced to the technology world. And then I wanted to go and trade stocks on Wall Street and make lots of money. So that's what my aim was, and I went to business school and I did not know at the time what is the distinction between you know, northern California, southern California, what is USC?

Speaker 3:

What is UC Irvine? What is UCLA? I had zero idea. I just knew that you would give GMAT and then you get a score, and then you apply to a few schools and then you just go there, right? So Irvine, uc Irvine basically extended me an offer to waive the application fees, the business school application, and that's the only reason why I applied there. I did not even know that USC, which was at that time used to be better ranked than UCI's business school, and I could have applied there as well. But I had no idea, and the internet was not as mainstream in 2010, right. So all of a sudden, then, I had a few admits, one from a business school in Canada, rockman School of Management, where I had been initially waitlisted, and the another admit was from UCR Ryan's business school, where I had a generous scholarship and I was not interested. I was. I got through straight up, right.

Speaker 3:

So then I came to the US and then I realized that, well, there are lots of Wall Street bankers who are unemployed at the moment, right? I remember Charles Schwab had a huge wealth conference for wealth advisors in San Francisco, and that was my first time coming to the Bay Area, to Silicon Valley and talking to a lot of very prominent investment firms, because I wanted to go work in an investment firm and learning that the action is not in the US so far as the investment world is concerned, but in the emerging markets, and this I'm talking about 2010, 2011 timeframe. There are a lot of investors who had come from Europe, from Britain. Tony Blair was one of the guest speakers in the conference. He was just huge right. For a student who has just come abroad, it was a great exposure, right.

Speaker 3:

So, anyhow, all of a sudden then I slowly realized that maybe if I pivot my experience in tech to stay in tech, then that will be much more easier for me to get a job right. And I was introduced through one of my alumni from my school, from UC Irvine, who was a senior director at Peta. She met with me at a career get-together. We used to call it the Bay Area Career Track, right? So we would drive a bunch of students driving from Southern California to come to Silicon Valley and visit multiple tech companies, and so I was at eBay and Yahoo that day and Yahoo was having a new CEO and during that process I got introduced to this mentor of mine. She introduced me to someone and that finally converted into a job opportunity and that's how I got my break into data science.

Speaker 3:

And it was not called data science back then, they used to call it analytics, and my first title, I remember, was senior analyst. So I was on the North America. My first title, I remember, was senior analyst, so I was on the North America FP&A team where I was doing business analytics. I was supporting not only financial planning and forecasting for the marketing business unit in US and Canada, but I was also doing things like pricing, where I was determining what is the rate that we offer a merchant and what would that mean for our margins as PayPal.

Speaker 3:

Right, interestingly enough, I was involved in the negotiations when WhatsApp used to charge a dollar every year to renew the subscription for users and it was a microtransaction. It was a very small fee a dollar. So how do you do pricing for that? If you charge 2.9% plus 30 cents on dollar transaction, then that's a lot of money. You just took 30% off of the dollar that they're earning. So they came up with a unique pricing model called micro pricing. So it was 5% plus 5 cents. So it was definitely much, much attracted.

Speaker 3:

So that was my introduction. And then I got introduced to marketing analytics. And then the industry itself started evolving with the tools that were becoming mainstream right. Industry itself started evolving with the tools that were becoming mainstream right. So, for example, r and Python and the way to access data itself started getting democratized when Facebook started launching Presto and Hadoop and Hive was the rage in 2013 timeframe, and gradually this field evolved, right. So then data science and then you started learning about the classic statistical packages, which can be clubbed within artificial intelligence if you will, but no gendered API yet right. So it was more around. How can you do forecasting better? Right. How can you do optimization better? How can you allocate your media dollars better using media mix models? How can you do regressions better? How can you predict if a particular person is going to default on a loan or not? What are the factors that are undermining it? It's all self-learned on the job and with the help of a lot of Coursera courses back then, and that journey has continued over the years in my whole career Amazing.

Speaker 2:

Thank you for sharing your background so robustly. I really appreciate that story because right now in the program I have students in India, students in Pakistan, I have students from Africa, I have students in Asia and China. So it is a very global program and I love hearing people's stories so that other people can see themselves and see, oh, it doesn't have to be a linear path. Sometimes one thing will just lead to another and you'll continue learning and then you'll find yourself in this other space completely and you were at right. So you started. Well, you're at Lending Club, uber, whatsapp, meta and now you're at Kraft Heinz. So a little bit different of a consumer product.

Speaker 3:

Yes, totally, but a legendary brand which is like 200 years old brand right because it still stayed relevant in the lives of the consumers.

Speaker 2:

Yeah, and how does data help brands stay relevant?

Speaker 3:

There was a time when you used to do brand marketing, data was not the primary thought on top of mind of people, right, there was a saying that they say that X percent of the marketing works and the Y percent of the marketing does not work. But I do not know what percent of it works or does not work right. But that has quickly changed. With the democratization of the tools, open source tools and access to so much data, it has become possible to track how your brand is performing. There are things like brand track how your brand is performing. There are things like brand trackers that have become common, where what you do is you consistently send out a survey to a sample of your users every few days, every few months, every few weeks, whatever, and you pull them on a set of standard questions and you see how the response the median response or the average response and the confidence intervals around it are shifting over time. So what is your brand's perception over time? So that is a very classic way to track the brand itself, irrespective of whatever you're doing in the marketplace so far as your marketing efforts are concerned. So that's the brand tracker stuff that companies have used, have been using successfully, and I myself have been part of building in the past at WhatsApp and Meta. Secondly, whenever you do brand marketing and you do like this huge splash media campaign on television, newspaper, digital and you run a 360 campaign and you spend a lot of money behind it, how do you measure it right? What is the objective of doing it right? That's very, very important. So now, brand and performance goes hand in hand for any company, so there are ways to tie the effect of one on another right. So, for example, the hypothesis is that if you're doing brand advertising, then you will be on the top of mind of your consumer. And how do you show this? You can look at variability matrix. You can look at product metrics themselves and try to tie changes to the product metrics for the timing of your brand campaign. So those are some of the innovative ways.

Speaker 3:

Some companies are more successful at doing it themselves. For example, the likes of Facebooks of the world and WhatsApps of the world have access to very granular and ginormous data sets where they can survey thousands of users at various stages of the flight of their brand campaign, while other companies may not have access to that type of data, so they rely on third-party companies which recruit users, although the numbers are smaller. So, believe it or not, when we were at WhatsApp, we used to be able to recruit like 9,000 people who complete a survey, at least one question on the survey. But even that was not good enough because it was not possible to survey the population of the sample from the population for WhatsApp, because WhatsApp just has a phone number for people. You cannot text a survey at that time or you cannot surface a survey right.

Speaker 3:

So you're relying on people who are using both WhatsApp and Facebook and then you are surveying them on Facebook as opposed to surveying them on WhatsApp, and you also rely on third-party companies hoping that people who are not on Facebook, who are just on WhatsApp. You try to get a sliver of them to respond to your surveys in-field, and it is, of course, extremely expensive and gets extremely complicated extremely fast. Having said that, it does not mean that we should not measure our brand marketing efforts. Brand marketing can be measured. Brand marketing and performance marketing are related. That's been proven using scientific data and there is a lot of value addition that can happen if you think through a good plan where your brand and your direct response sort of work as a well-oiled engine when you're doing your marketing.

Speaker 2:

Yeah, my experience is more on the smaller side. I work with a lot of small businesses and entrepreneurs and at one point, I had a PR agency and what was fantastic was that I was able to measure how much traffic was coming to the website from the hits we were getting, which is not something people used to think about. Right, they used to think about everything as siloed, but I love this marketing mix and integrated marketing strategies and being able to really track every aspect. In the class that I'm teaching right now, we just talked about attribution, right? Are you doing first touch, last touch? What's the best way to figure out what attribution model works for your product design, and does it also depend on what stage they are in the funnel? So, are you at the awareness stage? Are you you know? Are they interested? Are they actually in purchase mode? One thing that I really thought was notable for you was that at Kraft Heinz, you helped save $35 million based on creating a framework for them to really be able to measure right their marketing mix modeling. That's incredible, totally.

Speaker 3:

It was a very pleasant surprise when we started looking at individual channels, or individual levers as we call it in our marketing mix, or individual levers as we call it in our marketing mix, and started assessing the effectiveness of each of these channels in an integrated fashion and realized that promotions that we were running where probably there was scope for improvement. So to the tune of the amount that you just mentioned, you can just reallocate that much money and invest it into something that is more incremental to the overall media mix.

Speaker 3:

So, yeah, if done right, these models can be incredibly rewarding. And you don't need extra budget. You can just use the existing budget more efficiently to drive better bottom line impact, and that's something else we were talking about.

Speaker 2:

We were talking about, for instance, if you do ads on Google or Facebook or any other meta platform, and how you can get data from other people who've been served ads to the Facebook ad library. But then you can also do some A-B testing and see what's really working and corresponding to different audience segments, and then you know where to reinvest your spending. Where to reinvest your spending. So are those the kind of findings that you were getting is, oh, we need to put more advertising into this bucket, whether it's TV streaming versus just strictly online websites, social media versus other efforts.

Speaker 3:

Yeah, so it definitely varies from company to company, right? So I've had experiences working with MediaMaxMari during my Uber days as well, and the Uber media budget used to be more bigger, right? So you're spending close to a billion US dollars across the globe Wow, performance marketing alone, which included some brand marketing. Yes, now, there was this notion at the time before the Media Mix Mori, to measure everything using last click, right. And when you start using last click, then channels like paid search look incredibly lucrative. Paid last click right. And when you start using last click, then channels like paid search look incredibly lucrative.

Speaker 3:

Paid branded search, right, where somebody is searching for drive for Uber or just the word Uber, and then they click an ad and they become a signed up driver and then they convert and take a first trip and you say, oh my God, paid search is this amazing channel. We should allocate all of our money to this bucket, which was true for our case. We were putting a lot of money into branded search. Almost half of our media mix was in branded search paid branded search on Google, basically, right, that was half of it. But when you start peeling the onion and you start building a media mix model, you suddenly figure out that channels like programmatic display are not going to show up as easily as efficient just on a last-click basis. But when you start looking at the media mix, the story becomes different. Right, so there are approaches where you should model the paid search differently compared to other channels, because paid search is inherently a demand-capture channel. The intent is already there and that's why they're searching for your brand. So then we realized that, oh, programmatic display was extremely undernourished in the last-click model and, based on the media mix model learnings, there was opportunity to significantly increase the media spend on programmatic display. But those were some of the interesting learnings.

Speaker 3:

Definitely, it's hard to show the incrementality of TV if you're running a national TV campaign and you do not have the patience to run a small test before rolling out with TV nationally, which happens a lot in big companies because there is a rush to try to spend the budget at times. So how do you sort of come up with a plan to take a few steps back and measure TB at some point? Yes, you do use your media mix model but, as you rightly said, it's important that you also do follow-up A-B tests to try to validate or invalidate your hypothesis that you're seeing from your media mix model results and, yeah, no one result is the gospel. No one result is absolute truth. It's important to stress, test it. It's important to use different methodologies to triangulate on the data and make sure that whatever media mix model is saying is indeed robustly tested for you to make confident decisions in your day-to-day marketing mix.

Speaker 2:

And this work for, particularly in the company you're in now, which has multiple products, right? I mean, of course, uber has multiple products, meta has multiple products, but I think of the number that you must be working with now, hundreds, thousands. So does this technique work when it's a brand specific campaign that is time sensitive and are even in a big, large company? Are you able to make quick decisions and reallocate just like that? Because I know something that people think about in large companies it has to go through all of layers of approvals.

Speaker 3:

the process can be onerous, totally so, yeah, at Facebook, we used to have this concept of a family of brands. That's what they used to call it, if I'm remembering it correctly, right? So basically, you have Instagram, you have WhatsApp, you have the Facebook classic app, and then these days, you see the devices business where they are the Oculus virtual reality and Orion is what they have launched recently. So you're doing marketing differently for each of these assets. How do you show up as one brand? That was a strategy that Facebook took, right that we want to make sure that everybody knows that everything that we own in our family of brands is owned by Facebook. So you would have seen, within WhatsApp or Instagram, whatsapp from Facebook, or something like that. They started showing that because, when they had surveys, they figured out that people did not know that WhatsApp had already been acquired by Facebook and the CEO of the company definitely wanted to be making sure that every app that he had is associated with the mothership brand that controls those brands. That's the strategy that Facebook had, or even Uber had. So you had Uber Eats, where the name itself is embedded into the product, and then you had Uber Lives, where the app itself was called Uber. Right, so there's no question about it that the brand is owned by Uber, right, and you have Uber Freight, so everything is Uber, right, uber is a part of the brand. But when you come to the CPG company, for example, for hours, or even a company like Procter Gamble or the others in the industry, you'll notice that they have tens of brands. They have like 200 brands probably, and they would have probably the top 10 or 20 contributing to the majority of their budget. Right, so, like this, classic power law applies everywhere, right? So it's also true for these companies.

Speaker 3:

And yes, it is tricky, it is hard. You cannot just snap a finger and make a decision and change everything on the ground, because budgets and plans get locked and then it takes time for you to propagate those changes. You've already signed deals with XYZ, tv stations or media companies and you're going to just randomly tweak it live, unless it's a digital portion of your mix. It's much easier to tweak the digital plans than to tweak the non-digital pieces. If it's a newspaper or if it's a TV deal that you have struck beforehand, then you have less control over it. Having said that, yes, there is a process, though to change it to make sure that the learnings are systemically applied. It may be a little bit more fast versus a little bit more slow, depending on the channel that you are trying to tweak, but it is still doable. So people are trying to make the best of their worlds, depending on the organization that they are in.

Speaker 2:

Yeah, oh, fantastic. Thank you. I want to hear about artificial intelligence and how it plays into the world of data. Of course, if you're working in Gen AI, we know that that's a whole other ballgame. But I've spoken to people who work in finance and say artificial intelligence has been running the majority of our financial systems for many, many years without people realizing it. So I'd love to hear your perspective. You know how it plays a role and what you're seeing on the horizon.

Speaker 3:

Absolutely so. Yeah, the finance counterparts, or the people from finance who did mention that artificial intelligence has been in place for a while, they're absolutely right, right. So, for example, I can tell you, when I was at LendingTree, we were determining if we should give a loan to somebody, right? So you're using credit score, you're using a classic model the banks have been using see basic logistic regression model. We're trying to estimate the probability that an individual is going to default on using a bunch of metrics and credit scores being one of the vital elements of that decisioning system, to determine if you should give a loan to an individual or not. When I was at Uber, I remember I built a model to try to predict what percent of the signed up drivers in a specific week going to activate, meaning going to actually start driving for Uber within the next 28 days, because that was the sort of the time that it took typically to capture like 60 to 70 percent of the conversions. So we would run a model an XGBoost model every week on the signed up cohort and try to update the probability of them activating depending on the stage where that specific individual was in the conversion funnel. So you sign up to be a driver. Then you upload your driver's license and your vehicle registration. You go through a background check. Once you pass that background check you get the green flag that now you're ready to drive and you go and take your first trip. So there are those three or four or five steps that need to be finished Now. Every single step further down that you are in the funnel, higher is your probability of conversion. So then you can just add all the probabilities of conversions in your database for that cohort and you know precisely how many conversions you are going to get in the next X weeks. So now, sitting in week zero, when the sign-ups have happened, you have a pretty good idea as to how many activated drivers you're going to get in the next Y weeks. So then you can pretty much use that information to plan how much money do you want to spend acquiring drivers in that specific market? And Uber was highly localized. So if a San Francisco has to be planned differently from a Seattle, or a Los Francisco has to be planned differently from a Seattle, or a Los Angeles has to be planned differently from a Chicago or a Las Vegas. So you do this like at scale, using computers and programming to like run it for hundreds of cities simultaneously and have that know-how. But the basics, the underlying fundamental model, is an XGBoost model that was helping us determine the wants of somebody converting, so you can call it AI.

Speaker 3:

Personally, I do not even know what is the difference between AI, generative AI, stats, machine learning these are all words, these are all alphabet soup. Ultimately, you want to solve a problem and there are these tools available and we can pick the best tool and just go ahead and solve the problem. But what is more important is to solve the problem. To answer your question, as we are evolving, as we are moving from, of course, generative AI is unlocking a lot of efficiencies for a lot of companies. So you would have heard that there are companies in the customer services space that are right for benefiting from generative AI almost immediately, because computers can do reading the script and talking back much more efficiently than humans and they can do 24 hours without complaining.

Speaker 3:

But beyond that, even in our world, in the world of marketing, the way I have seen generative, the way I use cases being applied, a is the call center example. Marketing can drive a lot of traffic on the phone If you're trying to give away a coupon to people to call to activate some coupon, it can lead to a spike in your call center volume. You can use artificial intelligence and we have been using artificial intelligence in the past to try to determine what offer should we give to which individual customer. So, for example, if you're an advertiser and you are going to call your representative at Meta and you're a small and medium business, so they would set up a meeting with you on a specific day or time. When you have this meeting, you call them or they call you and then, as soon as the rep connects with you, there is an underlying model that is running under the hood, which is code each advertiser for the different set of ad products that should be pitched to this particular advertiser and whatever is the highest likelihood of adoption product that gets pitched to the end advertiser. So that's an application for artificial intelligence. So what is on the creative side? We're generating the entire ad field. There are artificial intelligence solutions, generative AI solutions available now where you can have characters which are not human and a pharmaceutical ad.

Speaker 3:

Recently I don't know if they had used generative AI or not to generate it, but you cannot tell right. So, for example, some of the developments that I've been seeing on the Google from the Google side. Google Notebook LM is what they call it. You can just take a transcript of your thoughts, of your blog blog post, and google can generate a podcast out of that blog post and it would seem very original, very real. So podcasts are a great way that marketers leverage podcasts to grow their brand. Specifically, if you are not as well known, it definitely helps. For example, I'm not very well known and I've come on your podcast. Now that I have come on your podcast, hopefully people will think oh yeah, I know Saurabh, he knows a thing or two about this topic.

Speaker 2:

Yeah, and I found you at the conference, so yeah, exactly yes.

Speaker 3:

And for also generating the creatives themselves that, for example, you have to wrap this can of Diet Coke in some Coke creative. Now you can come up with customized creatives where each can can be a creative, customized to the individual customer who is going to order that can. Who knows, that will be possible, right, and they had recently launched that as well. I don't know before generative AI or after generative AI. So those are some of the interesting use cases I've seen. And also in the field of writing. The creative itself is pretty good at writing raps and would like to create a pitch for a sales rep, to pitch a customer so that they would be liking to take a call. So a lot of work still needs to be done. It's still in these stages, but definitely very promising, yeah, fantastic.

Speaker 2:

Is there anything we haven't talked about in the world of marketing mixed modeling, data analytics, big businesses that you think that we should really impart to the audience? I think we did not touch upon customer segmentation.

Speaker 3:

I personally was not aware of using customer surveys to try to then segment customers. I had not done it in the past until I started working at WhatsApp in 2018. So what we used to do is we used to ask the questions what is your perception of Facebook? What is the perception of WhatsApp? What will happen if WhatsApp was taken away and stopped working tomorrow? Will you feel happy, sad, indifferent, whatever? Right these types of questions, and that meant there were like nine or 10 questions and you're trying to estimate on the spectrum of being neutral, positive or negative towards the brand.

Speaker 3:

Where does an individual lie? And I know the Department of Statistics at UCLA has a detailed how-to on a package called POLAR, so it's Polytomous, ordinal Logistic Regression, and there is a professor at university in the Midwest, I think University of Wisconsin, who used this particular package, polar, to try to identify which type of wheat is more efficient, is likely to be more yielding more output, right? So I was then able to apply that same methodology to try to segment WhatsApp consumers based on their survey response, saying what are the odds of an individual to move their perception from a negative perception to a neutral perception, from a neutral perception to a positive perception, from a positive perception to an extremely positive perception. Now, as you have that information, you can segment your audience and you can reach out to them differently. So I did not know that was possible.

Speaker 3:

So that was extremely interesting learning that I had. Yeah, there are other ways of doing segmentation, where you look at how much time people are spending in the app, how active they are, what are they doing in the app, how much money have they spent? When did they spend that money? How many transactions have they had? All that good stuff which most people know? But this was something unique that I thought was extremely interesting when I did that at the time.

Speaker 2:

Yeah, that is really interesting and I think that's like you said earlier. That's a great use case for AI, because then you get the the data, and then you can have these models help figure out how you're going to personalize and how you're going to create different funnels for each one to get them to the behavior that you want.

Speaker 3:

Absolutely.

Speaker 2:

Amazing. I've really enjoyed our conversation and the time you've given and just your career is amazing, and you have a lot to offer right now on this podcast that you've shared with us, so thank you. I think it not only gave me a better understanding of how big businesses are using these tools, but also our students, so I really thank you for that.

Speaker 3:

Absolutely. It's a pleasure. Thank you for inviting me.

Speaker 1:

And.

Speaker 3:

I hope that our conversation helps students at your school and even those who listen to your podcast outside of your school, if it is available to them.

Speaker 2:

Yes, it is available on all platforms. So thank you for that, and do you have any last thoughts that you want to leave the audience with? One piece of advice, perhaps for somebody who's looking at getting to a career in marketing digital marketing or marketing with data.

Speaker 3:

I would just say that you know, things are changing a lot and at a very rapid pace, but at the same time, there are a lot of things that do not change. It doesn't cost anything to be a decent human being. So just sticking to the classic basics sticking to the basics is also very important to be successful. Never underestimate the power of something simple and don't think that you always need the most complicated solution to solve a problem. So the one thing that I would say is that, you know, be a decent human being. It goes a long, long way in building a network, in getting job opportunities, and having that positive attitude to show up, to learn and help others without expecting something in return. It just makes you feel good at the end of the day, and I think that's more important than anything else 100% agree with you on that.

Speaker 2:

So this has been Mediascape Insights from Digital Changemakers, with co-host Anika Jackson, and I've been here having a great conversation with Saurabh Kumar, who is the Senior Manager of Data Science at Kraft Heinz. Thank you again, and thank you to everybody who's watching or listening to this episode. Don't forget to leave us a rating review and a follow.

Speaker 1:

To learn more about the Master of Science in Digital Media Management program, visit us on the web at dmmuscedu.

People on this episode

Head Shot

Anika Jackson

Host
Head Shot

Joseph Itaya

Host