‘Empathy at Scale’: Mondelez’s Thomas Beetschen On Driving Personalized Experiences In a Privacy-Driven World

A lot of consumers are committed to look for local solutions, local products, or something sourced in their own countries. These trends are becoming prominent. Again, they want to know where the products are sourced and where they can find it. Lastly, a huge shift, I call it the era of brand shifting. Loyalty has always been at the front of a brand and marketer, but even more so today with the shifts of brands that happen, not just during the pandemic but continue to happen.

They are mainly driven by millennials and high-income millennials who have loyalty programs and personalization scale. A one-to-one relationship with the consumer is becoming more important to keep the consumers connected and loyal to the brand.

Guffanti: Wow. Thomas, as Giusy just outlined, shoppers are likely to be more omnichannel than ever – online and offline shopping is more connected than ever, people are more conscious about who they buy from, and consumer behavior and loyalty is constantly shifting. As a consumer good brand at Mondelez, how do you address all these major shifting trends?

Beetschen: It's not that easy, that's for sure. If I may, I'll just take a minute for those who might not know Mondelez because the name is not the consumer name, it's the company name. Mondelez is number one in snacking. The brands known to most people are: Oreo, Chips Ahoy!, Ritz, Cadbury, Milka, all of those brands. We do chocolate biscuits, gum, and candy, and in some parts of the world, some beverage and meals business. This is a lot of what we do, but the reason I mention it is because it is also a challenge in itself.

There will be a different CPG organization with different types of product, but when we are in the food and snacking business we see two types of change in terms of where the consumers are going. There's one type, which we're not going to talk too much about: what do they think about food? … That's what Giusy was starting to talk about. We see the same thing. The biggest acceleration we've seen coming back to shopping online is a barrier to e-commerce and DTC.

A vast majority of people had never been online, and therefore, how would they get through the technology barrier of ordering through one of those things. The pandemic has actually lowered that barrier tremendously. For us, that's something we've seen. The other thing, which is interesting, is despite a lot of people in lots of locations having been in lockdown and at home, the role of TV is changing a lot and the way we market and push communication, that's a very important shift that we have to account for.

At Mondelez, we look at this in a number of different areas, but there are three main buckets. We have a way of doing marketing that we call “empathy at scale.” What that really means is finding the balance between very deep personalization, which is not that easy when you deal with food products. Not everybody just wants to buy every single bar of Oreo with their name on it; there's occasions for that, but not all of them. We have to be finding a balance, getting into a deep contextualization, and making the consumer feel that the communication that is in front of them is fitted for them.

This is more about segments, cohorts, and things of that nature. We call that “empathy at scale,” which we'll talk more about later. The role of e-commerce, especially the digitization of brick-and-mortar retailers towards online, is a big shift in not only the way we market, but also the way we have to understand the online shape to make sure the product is there as it is in stores, and how we understand that information. Then, DTC, where we have some element of online direct selling for specific occasions, which we've been accelerating as well.

Guffanti: Thank you very much, those are great trends. Let’s drill down a bit further. Giusy, we've laid out that everyone's moving online, digital experiences are significantly up and people are looking for more engaging experiences. With that, you also have to talk about privacy and how data is used. Can you talk a little bit about what that means for brands?

Buonfantino: First of all, consumer privacy and growing the consumer with personalized messages is one project.

This is going to be a marriage for a long time and brands have to consider both. We know that consumers have become significantly more savvy digitally, even the older generation. They know what's going on and figure out how to protect their privacy, how to manage their privacy. Those consumers that manage the privacy settings effectively are also the ones that are willing to give brands their data for personalized experiences.

Of course, it's not just the personalization of experiences, it's using the right data, AI, and ML. Automate these impressions, advertising communication at scale, especially brands like Mondelez and Oreo, of course they asked to scale it because these are not niche brands. The brands need to know how to do more with less data, especially in the future.

Guffanti: Thomas, I love the phrase, “empathy at scale,” it’s a great way to underscore, as an industry, how to view this new world that we live in.

Turning our attention to predictive marketing, the holy grail of a lot of companies is to be one step ahead of the consumer and optimize predictive marketing efforts. In order to do that, you need to be intelligent, have the right data, and use it in the right way. Can you talk a little bit about how you utilize AI and ML in a predictive marketing strategy?

Beetschen: The short answer is in very automated ways sometimes, and sometimes it's very much in analytical ways. Automated ways in the sense of a lot of the technology and software platforms that we use have embedded AI and machine learning capabilities.

Think about the goal of the programmatic media buying that we do is driven by this, a lot of the A/B testing we do on our websites, the backend of that is about AI and ML. That's an important use, it's important not to think that we always only do AI and ML when we have a cohort of scientists driving brand new models that didn't exist before.

A lot of it is already there and that's important to remember because for CPG, one challenge is automating back, whether it's the next ad we place, the next page that we do on the website. We can't have marketers on every single one of those steps taking that action, we need to be able to move towards the edge, to the AI and ML capability. We have a lot of that and that's something which is more or less industrialized in the way we operate on the marketing side at Mondelez, mainly around the media buying and some of the websites, some of the CRM.

We then take and create some models, not necessarily to predict what the consumer will do, but to predict whether some messages will be liked by consumers. We are using AI and ML for the address that we create, and put them through models to get a view of how likely they are to be seen, after how many seconds are people likely to drop off, etc. It's a tactical performance management level, but quite useful for us.

Then, we also get into some of the more internal unique models, bringing data from some of the consumer experiences into R&D, helping to work out what trends are likely to happen, some of the backsides, some of those activities.

I like to try to demystify a little bit of AI and ML – it's not just extremely complex deep learning capabilities sometimes, but it's embedded in a lot of technology. Having worked on a lot of the Google platform, that's been useful because there is direct integration for the technical people out there. That direct integration is important; it helps speed up the automation of the result of the AI and ML capabilities.

Guffanti: That leads me to my next question, Thomas. You mentioned how AI and ML is embedded in pretty much everything you do, not just the more complex deep learning capabilities. Giusy, Mondelez has it totally embedded into their infrastructure, what do you say to companies who are on the webinar right now saying: we’re not even on step one of embedding AI and ML into any of our communications or anything. What is your suggestion and advice there?

Buonfantino: First of all, use your partners like Google to understand your maturity journey or digital maturity by brand and by country. CPG is a complex world, multi-brand, multi-country. Not every brand is on the same digital maturity journey. They are finding very foundational use cases and marketing campaigns – it's their mandate, that's what [they are finding] as more use cases emerge. Companies are using their own data and activating this marketing campaign across digital channels and starting to do some of the ROI.

Mondelez is more multi-moment, of course, everywhere around the globe, they optimize across all the channels and all the consumer journeys on the fly. These are mature. However, no matter where your maturity journey is, break the data silos, bring the consumer data all in one place, start to have a strategy of first-party data, build the infrastructure of first-party data in view of the third-party cookies becoming less reliable.

Then start to do tests and learn with AI, ML, and data analytics with a platform that can be democratized across the organization, which is a fundamental component. If data is not available to everybody, then it's going to be very difficult to scale it.

Guffanti: We've been talking about how to embed artificial intelligence and machine learning throughout your organization to get more predictive, get closer to the consumer, optimize messaging before the messaging even goes out, things like that. Thomas, can you give us some examples of what that looks like when you're implementing these best-in-class strategies and engaging with consumers at a high level. What does predictive marketing look like? What are some of the successes that you've experienced?

Beetschen: CPG is everything but monolithic. If you think of Mondelez, Mondelez is a 1,000-brand country combination. In there, there is a war of the variation of size of brand, size of campaign, size of consumer demographic, that needs to be adapted for. To come to your question, what does it feel like for the consumer? Most of the time, nothing different, which is actually the interesting piece because when the communication has been done well, it's the right message at the right time, in the right context. To some extent, as a consumer, you just felt that it was just right.

A lot of the things we do is to make sure it's going to be just right. Again, it's about managing the performance of the platforms. When I say the platforms, it's YouTube, Facebook, and Twitter, and making sure we actually go and buy in the right place, position in the right place, put the right formats, etc. We've talked about the creative being in the right and best fit for purpose for each of those platforms. We're starting to work around segmentation and groups of consumers to understand what those groups of consumers are.

In FMCG and CPG, this is not just about demographics and location. In our world, this is about what people want in the morning, what they expect from a snack (dark chocolate or white chocolate). Those are the types of things that you can’t buy data out there for. It's about starting to have the data and looking at it to create some of those segments that are category specific so that we can then use them to create those types of groups. On one hand we have the functions of marketing against the media buying, the website, the CRM, we optimize those platforms. Then what we also try to do is manage it, as Giusy was saying.

A lot of that is about driving and understanding where the handshake or conversions happen – from an ad to a website, from an ad to a CRM or a specific promotion that the consumer wants to go in. That's quite classic, but it can be optimized a lot through ML. We've done some things that were quite interesting in a couple of markets where we've genuinely started to match sales at retailers with advertising pressure and understanding whether those things are working or not. Again, that helps to select the best adverts, the best platform, the best path-to-purchase. We've done that on a number of occasions.

When you get specific, take Oreo for instance, we've got things like My Oreo. When you move into a real level of personalization then you get much closer and have a lot of capability to expect what the next space is. We've done similar things with in the UK, for instance. It's a range, and in CPG this is one of the challenges that we have.

Guffanti: If I'm understanding what you're saying correctly, in an ideal world, you know you're doing something right when the consumer has a sense of a natural relationship with the consumer goods brand. They don't sense anything out of order, unnatural, or inappropriate. That's not an easy trick to pull off and it's great that you're accomplishing it. I would assume that in this privacy-centric world that we all live in, the importance of first-party data is top-of-mind for you. Can you talk a little bit about the different types of data and how you're thinking about what you can do at Mondelez with it?

Beetschen: I'll start by redefining or sharing what we mean by first-party data at Mondelez. In the industry, a lot of people see first-party data as equal consumer data. We've taken a slightly different view at Mondelez. We called first-party data, data that we own and that's actually broader than specifically consumer recalled if you want. Quite often, in the CPG industry, the CPGs don't own as much data as they could. Look at it in a simple way: every time there is an interaction between the brand and the consumer, in terms of the messaging, the brand has paid for it. The brand has paid to produce the ad, to buy the inventory, to create a website. If we've paid for that, we've made that investment, it's only fair that the performance signals that are created are ours.

We've done a lot of work. It’s been a 3.5-year journey to systematically go after the data that is actually not consumer-centric, but created at the time of the consumer interaction and make sure we can actually own it. Once we own it, we can bring it in, clean it, and use it to generate a lot of insights around the performance of campaigns and platforms. That's been one part, it's probably an opportunity for a lot of CPGs to go after that. It's there, it's ready to be used. There is some saying that data is the fuel, and this is a very easy way to fill up the tank in the CPG world. The IT organization has a lot to do with that, but that's very important to go and do. We sit now at Mondelez with trillions of data points that have been generated that we can then use and analyze.

Now, having said that, we've tried and we continue to try. We have a couple of areas where we're doing this. We're doing this in the pilot first because the main thing we want to do is learn two things: learn how, and then make sure that we take care of those ecosystems and the data in a very sensible way. That's very important, the legal requirements, the ethical requirements are there. We take extra care to do that. We have technical pilots in legal and others that have helped us do that.

Equally importantly, we try to understand who's going to use this information. We have been using different pilots across the globe: what happens if we give main access to the CDP to a brand manager versus a marketing agency? Who's in charge of driving the campaign. This is where we're trying those types of capabilities. The biggest learning for us is to have marketing campaigns that are cool and want to be engaging. It's not just about awareness, you have to be engaging and therefore it's important to underpin CDPs in those areas.

The CDP enables you to do all of the things you wanted to do, but couldn't as a marketer – talk to every single consumer that is willing to talk to you. It's the multiplication capability to do that. My big advice is for the technologies here, working with marketers, the number one question you have to ask is show me your campaign. 

Guffanti: Giusy, a similar question about AI and ML implementation. For brands that are just starting out and looking to implement the CDP, what are some of the challenges they run into?

Buonfantino: Firstly, it's a new technology stack and the challenges are all organizational silos. Of course, the CIO and CMO need to get on a unified agenda, which is very important, as they may have different perspectives. The other one could be globalization. As Thomas said, there's no one size fits all. That's why the CPG Google Cloud customer data platform is modular enough. Then, the usual implementation hiccups. My recommendation, and Google Cloud’s recommendation, is to work with partners that already know the organization, the company, and the technology that the company, specific brands, or enterprise are using. The integration will be much easier. There are more ready solutions and built app solutions on Google Cloud. … I recommend starting with few use cases and a small house.

Guffanti: Thomas, circling back on Mondelez, you're a global organization, global company, global reach. What do you think about global versus local when you're experimenting with different models and so on?

Beetschen: In the context of marketing – it's important to push it there – where does the scale get optimized? For example, when looking at driving API integration with the likes of YouTube or our platform website, which will be important when we then feed data to ML models or CDPs, that’s done once. Then, we ensure everybody in the organization does it the same way because there is no benefit in doing that differently.

A number of decisions are made at regional, for us a region is Europe, APAC, or North America. Here, we put a lot of visualization capabilities and high-level analytics capabilities because the big marketing decision-makers are at those levels. For me, the consumers are always local, especially in food.

When we look at the technology stack, that is very close to the consumers – the CDPs, the CRMs, the web optimization tools are all good examples –  they have to be local, it doesn't mean they all should be different, but to some extent implementation and use has to be local. For us it's important to be focused in that way.

Guffanti: Thomas, talk to me about the rest of the business. Obviously, marketing is one part, but as we know nothing operates or should operate in a silo. The impact of consumer behavior and a shifting consumer behavior impacts multiple areas of the business – supply chain, product development, etc. How are you thinking about this at Mondelez? How are you tying it all together and getting the organization to pull in the same direction?

Beetschen: I've been lucky enough to work with marketing, sales, supply chain, and a lot of those functions in my career. The CPGs have always been consumer-centric and looking at bringing that information through. The reality is what was happening before was aggregated information, things that are already formed into an insight and quite often could sit on a PowerPoint presentation. That's what used to travel from one function to another. Where we see things now is with cloud capabilities, ML capabilities, the fact that we can store massive amounts of information and deal with it relatively easily.

Now, what we can do is move some of the raw data between the silos and open up use cases we've never been able to dream about before. It gives an idea of: could we impact media pressure and salesforce pressure, how do we make sure we only send the rep in the right places based on where the consumer seems to have a demand? We can do things across both marketing and sales. We can share much more of the consumer feedback, even some of the non-classical feedback with R&D.

For me the big difference is not that we can't communicate, we always did, but now we communicate with a lot more granularity, and therefore we don't lose the richness of that granularity, where in the past we were. That's what we're trying to do. Using things like Google Cloud or big repositories helps create places where everybody can bring data and collaborate on it.

Guffanti: Shifting a bit, let’s talk about future-proofing our businesses. We're all on this webinar and consumer brands are shaking their heads and saying, “amen we'd love to get to this ultimate end state.” Giusy, the future is now. We want to future-proof our businesses and that means getting started on the journey right now. What is some of your advice for consumer brands who are not as far along on the maturity scale as Mondelez might be?

Buonfantino: The most important thing is to break the organizational silos. The consumers, to Thomas' point, are at the core of every CPG company. Everybody in the organization thinks about the consumers, but about the consumers together and the consumer data. Not just the consumer data, but the customer data, retail data, and enterprise data in one meeting room, one team. After breaking the organizational silos, break the data silos, bring all the data in the cloud, where AI, ML, and other tools like BigQuery can help you data and analytics, strategize, and make a decision in a matter of seconds.

Then, democratize the data and analytics, the dashboard across the organization. It's important that the supply chain sees the same consumer data that marketing sees, that R&D sees the same consumer data that marketing sees, and integrates all this data with external data, your data, search data, media data, or public data. Events, location, geographical data are all publicly available, and integrate that  data to internal data for faster decision making. Break the silo.

Guffanti: Thomas, I believe this question is directed to you, regarding retail collaboration or data sharing. With the majority of products sold through key accounts who also have their own rich customer data, why are you not working more closely? Or maybe you are working closely with them, but how is that dynamic and how does that dovetail into the efforts you’ve described?

Beetschen: We are working and trying to find ways, but the first thing to remember is in a large number of cases, the shoppers at the retailer didn't give consent for the data that they shared with the retailer to be given to anybody else. It looks like the data is out there, but without the explicit consent from the consumer or the shopper, that data cannot travel, or at least cannot travel in its raw form. What we are doing is finding the aggregation level where the privacy issue disappears, because it's not talking about Albert and Giusy, but rather about a certain cohort of consumers. Then we can share this with retailers and manufacturers.

The other way, which is something we've done at Mondelez successfully on a couple of occasions, is to architect your campaign in a collaborative way with your retailers. Then create the opportunity for that data to be consented from the shopper and the consumer to both parties, and for both parties to use it. Again, that starts with having the right campaign architecture; this is not a technology challenge.

The third piece, and the trick because it's not easy to execute, is not to worry about the consumer record. There are an awful lot of things driven by the collaboration and what the retailers can share are performance signals. That is equally important to understand how the marketing works and whether the marketing choices we make as CPG are working or not. Retailers will be more keen to share whether a conversion works or doesn't work without necessarily sharing direct zero-party data. The challenge for CPGs is that the number of direct interactions are very limited. CPGs need to find ways of creating more of those interactions, there’s no doubt about this.

Guffanti: Giusy, building your own CDP can be very time intensive. Are there any as-a-service options that can help achieve the end result?

Buonfantino: Again, it depends on the digital maturity, the business size, the brand size, and the resources that a company has. There are many SaaS solutions that are easier to build and offer a more limited number of use cases, especially on the media side. There are solutions that can be more complex that offer the opportunity to use this data for use cases beyond the media. In CPG it's a mixed bag of both, and we help the clients and the customer make the right decision with the partner of choice in each location, including Mondelez.

Guffanti: Such a great conversation, it went so quickly – such great information. Thank you both for providing your insight. Do you have any last words on the topic, any parting words of wisdom for our audience? Giusy, let's start with you.

Buonfantino: Start the journey. If you didn't, it's not scary, it's very important. Third-party cookies are going to become less reliable and consumer privacy is going to be at the core of what you need to do. Break the silos and start to build a strategy, 

Beetschen: For me, there are two messages. The first being you have to build the silos and break them at the same time, because the risk otherwise is the CPG never gets to granular data because that's in the DNA of a lot of CPGs. Then, create the bridges at the same time – that’s a control approach to it.

The second one is for my technology colleagues. Put the sensors out there. The data will never come to you unless you actively go and decide to capture it on your website, on your ads, on the promotion mechanism, on your CRM. That is not a historical strand of CPG and it has to become one. If you don't have the sensors, the data doesn't come in. It's important to go and love that granular data. Go after it and be conscientious about how we deal with it.

Guffanti: Thank you so much to both of you. I appreciate all the information and I'm sure our audience does as well. A lot to digest, a lot of words of wisdom that we can internalize and start our own journeys, as well. I want to thank our speakers Giusy, Thomas, it was a pleasure having this conversation. Thanks to our audience members as well for taking time out to join us on today's conversation. Have a great day and we will see you all soon. Thank you.