← The Digiday Podcast
The Digiday Podcast
Mondelez overhauls its $3.5 billion digital commerce strategy in era of AI search
00:35:16
Loading summary
Transcript55 lines
- [00:00]
A
Foreign.
- [00:11]
B
Welcome to another episode of the Digiday Podcast, a show for anyone who has asked ChatGPT for best cookie recommendations. I'm Kamiko McCoy, senior marketing reporter here at Digiday.
- [00:22]
A
And I'm Tim Peterson, executive editor of Video and audio at Digitay Media. Have you asked ChatGPT for best cookie recommendations?
- [00:29]
B
Given that the cookies that I cover are Google's cookies? No. What about you?
- [00:36]
A
Not for cookies. Cookies. I'm pretty solid with my chosen cookies, but I have asked Gemini for help finding a barber. Recently ended up going to a hair salon, which was a first for me, but I really enjoyed it. But it's the kind of thing that does make me wonder, how is Gemini ChatGPT all this? How are they picking out which products or businesses, services to show me?
- [01:03]
B
That is very interesting that you asked that, because this week we actually had a very special guest, Andrew Lederman, who is the VP of Global Digital Commerce at Mondelez, and talk about what that looks like from Oreo's perspective. We also touched on their Cliff Health bar. But navigating this space right now is very much the unknown. I think you sat in a presentation not too long ago where another one of their executives talked about how they had bots blocked for Oreo search and then had to unblock those same bots to get Oreo to show up in the search results. So they've been testing and learning this for a little while. So I had the opportunity to connect with Andrew about what that looks like for a couple of their brands, starting with Oreo and then fanning out from there.
- [01:45]
A
So you're talking ao, geo, SEO, all the eieios?
- [01:50]
B
Yes, all. All of the acronyms that you can imagine for this space and how that kind of feeds into what, like a paid approach to ChatGPT, OpenAI and things like this then look like for them, given that's also on the table now as well.
- [02:03]
A
Got it. So not blocking the AI bots seems like kind of step one. Anything else that you took away of interest in terms of, like, what their strategy or their primary tactics are for showing up in these A searches and chatbots?
- [02:17]
B
Well, this is something that we've been talking about for a while because the paid and organic teams are actually having to work closer together than they have historically. Part of the bot blocking issue was, you know, you've got bots that are crawling all over these websites to take things back to the research results. But what the learning was is that you've got to play nice with them to be able to show up at all. So these teams are working together between paid and organic. And then also, you know, you've got to, you've got to come to the table to play ball. Publishers, I think, have learned a similar lesson that brands are also catching up on now.
- [02:53]
A
And it also feels like a lesson that's being relearned because that reminds me of SEO. And do you know the acronym SEM?
- [03:02]
B
I know a search engine, something marketing. Yes.
- [03:06]
C
Yeah.
- [03:06]
A
So which is basically the idea of like SEO is organic search, SEM is paid search. And the two are usually used with a slash in the middle because they're just like conjoined. So I guess we have A, O, G, E, O. And maybe we're gonna need aem, G,
- [03:25]
B
E, M, you know, for all of these. I know as journalists and reporters we're often very like cynical about what the future of all this is gonna be. But in talking to Andrew, he made the case that it's not a matter of if, but when. People are going to start shopping for products and things like this online. So even though the water seems very murky and they're still trying to read the tea leaves about what this means and how exactly they're supposed to show up, the fact of the matter is that AI is the future. I hate to use this corporate speak, but the safer bet is to be there than to not.
- [03:59]
A
Sure that makes sense. I also don't know to what extent AI is the future anymore. So much is very, very present.
- [04:08]
B
Fair enough.
- [04:10]
A
I'm excited to listen to the conversation.
- [04:18]
B
Andrew, welcome to the show.
- [04:20]
C
Thanks for having me. I appreciate it. Looking forward to chatting.
- [04:23]
B
Yeah, I'm super excited to have this conversation. I feel like we have been talking about AI nonstop, but it's been a non stop stop roller coaster, I think that's fair to say.
- [04:33]
C
No, I think that's fair. A lot of excitement. Sometimes good news, sometimes bad news. Lots of ups and downs.
- [04:42]
B
Back in 2024, Gardner, the research firm, came out with some research that predicted that search and volume, search engine volume, excuse me, was expected to drop by 25% by the end of 2026, which is this year. And I feel like that gave us maybe the industry Chicken little syndrome. It was due to chatbots and virtual agents. I think publishers have been feeling a lot of the effects of that and digital has pretty extensively covered that. But this looks different for brands. Right. And CPGs like Mondelez, which is why I'm excited to have you on the show. What are some of the biggest impacts that AI and the changes in search AIO and GEO have made for you guys?
- [05:25]
C
Yeah. So look, I think it comes down to customer journey. Right. So we just understand if you think about our business, we're in the business of making sure consumers can find our products when they're looking for them. So whether that be in store, on a shelf, whether that be in a Google search or on a Walmart.com or anywhere else, what we're seeing is there's a bit of a change in the way that users are solving their problems or deciding what SKUs to buy. And we just have to address where they're gonna be and make sure that we're showing up the way that we want to be shown and that we're solving the problems that consumers have. So it's really to me more about just being consumer centric. And personally, if consumers choose to use LLMs to find their products, that's great. We're happy to be there and help they choose to find their products in store. That's fine with me as well. And if they choose to use Google, we just have to be prepared for all of it so that consumers can go where they feel most comfortable.
- [06:23]
B
Absolutely. Now this is a report that came back in 2024 and I feel like we've been seeing the effects ever since. But when would you say that your team kind of noticed? Like, hey, there's a difference in our traffic. We're starting to see Gemini, we're starting to see ChatGPT show up in addition to Google. When did you guys notice that?
- [06:38]
C
So really for us, at the end of the day, most of our business we, we do through retailers. So it really came to us through retailers. When retailers started saying things to us like they're expecting 30% of their traffic to be agentic by 28, we stopped and paid attention. This was about a year ago. Now that we've been working on the agentic commerce agentic search point of view directly from feedback from a customer. It's been about a year of a journey for us to start paying attention to this. It continues to grow and it's been as relevant as we expected. Where it goes from here is a great conversation. I think there's high speculation in both high and low directions, depending on who you ask.
- [07:27]
B
Absolutely. Now, you guys's global head of digital marketing and strategy, Jen Menes was speaking at an IAB conference earlier this year and my co host Tim was sitting in on that and she said something really, really interesting. That Oreo only gets cited in about 10% of AI chatbot responses. About Cookie recommendations partly because Mondelez had those AI chatbot crawlers blocked. Talk to me a little bit about you guys noticing that you guys had the bots blocked. I'm assuming that we've unblocked them. Talk to me a little bit about that.
- [07:58]
C
Yeah, so. So look, I think when it comes to digital commerce we've been very focused on PDP pages, the product detail pages on our retailers websites. But what, what LLMs change with all of this is that citation can come from many different sources. Owned and earned media retailer.com, brand.com so we actually to fully take a different look at our brand.com experiences and make sure that they're optimized towards AI. And this is not something that we had previously done. Our websites have typically kind of two primary purposes, one being for DTC to actually sell product for things like Oreo or Cliff. But on the other hand then we also have recipes and content places where people can get discounts and coupons. And we use it as kind of lead generation source so that we have good targeting data to make sure that we understand our consumers well. This is a different use case. So this is something new that we're building in the organization. And yes, you're absolutely right, we have turned on so that the crawlers can now make it through our website. So that is a great advantage that we've taken. But there's a lot of different technical infrastructures that we have to think about. Ultimately we break it down in three ways. We think about it as accessibility, which to your point, making sure that they can get to our site is a good step one, structure and then answerability. So that's kind of the way we think about our technology. And so we ensure that they can actually get to our sites through like clean site maps, proper TXT files, fast load times. Because to our point and to your point, if they can't crawl our site, nothing else matters. Secondly, we made content machine readable. So by having like schema at scale, FAQs, product article breadcrumbs, just so that ultimately the models can parse the information and get a good understanding of what they're seeing. And then third, we think about how we kind of reshape our content itself to be answer ready, so clear H1S strong intro paragraphs, a little like TLDR summary and formats like FAQs and comparison pages that LLMs can directly quote from. So this is kind of, it's a multi pronged approach to the way that we're even approaching the technology end of Our branded sites.
- [10:25]
B
Absolutely. I'm curious, when Mondelez originally had the bots crawlers blocked. Right. What was the rationale behind that and what was the change of heart to unblock them?
- [10:37]
C
I think that when you think about the use case of how we were intending to use our websites, we came from an answer of does that help us in any particular way? And the answer was no. I think that there's been a look, I think for many in kind of the media side of the world as well, there's a conversation around ownership of information and data. And we are coming from a place of, hey, let's start with not sharing and then let's understand the use case and how it can be valuable to the company and then we'll open it up from there. And this is a clear example of where you see the value and you go, better off having them crawl the site. So I think it's a little bit of coming from what will this industry be? What is this going to look like? How do we have to set it up properly and ultimately finding the right use case to open up the traffic.
- [11:34]
B
Got it. So Mondelez has unblocked the bot crawlers. That's a wordful. Unblocked the bot crawlers. Now, at this point, talk to me, what have been the results? You guys have put a plan into place. What have been results of the strategy now?
- [11:50]
C
Well, I'll start by saying I think blocking the crawlers or unblocking the crawlers, sorry, I should say, is just one piece of a fairly complex plan. So just to give you kind of the overlay of the strategy of what we're thinking about is, so we're trying to build a knowledge infrastructure for AI so not just optimizing content. And we think about it in a few different ways. So first, we ensure consistent, structured product knowledge across every surface, so retailer PDPs, brand sites and earned media so that AI systems can accurately interpret and trust our brands. So first is this kind of holistic piece across. And then second, we're, we scale AI native content across multiple formats. So we're looking at long form answer ready content that's particular around occasions that matter for our customers, our FAQs, our nutrition. Like we kind of mentioned a bit before, we show up in a full range of consumer prompts, not just traditional cookies or snacking type of search words that come up. And then we measure. And I think, to me, this is the most important piece of the whole equation because I know a lot of people are putting out content and getting started and going different directions to try to get more citation or more visibility from the LLMs. But our first move was to make sure that we had a measurement system in place so that we can treat it like a performance marketing channel. Like we can have. We have five KPIs that three based on visibility, one based on citation, one based on sentiment. We have those in place. We put occasions down very particularly that we're going after, make sure the content on the websites, the retailers and earned and owned media is all in the right place and measure again, see if we're actually making progress in the right direction. So I think when it comes to even the crawlers or anything like this, it is one piece of a larger strategy that is highly holistic and consumer centric to make sure that we can provide the right answers to the LLMs so that they can bring them to consumers in the proper way.
- [14:13]
B
Absolutely. That being said, has there been an increase in how many times a brand like Oreo gets cited now? Have we. Have you guys managed to move beyond the 10% sighting?
- [14:24]
C
Yes. And I will add that as we've advanced our capability and understanding. So we're actually highly, you know, we're coming up pretty highly now for Oreo particularly. It's a very, very big brand, so we come up quite strongly depending on the particular occasion. But again, to my point around occasions is it's not just one size fits all. Are we showing up well or are we not? It's actually broken down by individual occasions. And that's where I think we have opportunities still. So if you think about things like particular holidays or celebrations for Oreo or endurance energy for Cliff, there are particular areas in which we still have opportunity to grow. And the way that I've been kind of looking at it is how does our share of voice in these particular occasions compare to our market share? They should be somewhat similar in my mind in order for us to have fair share. So the quick answer is, yes, we're doing a lot better than we were originally. The longer answer is, but we still have work to do in particular occasions so that we can present our products properly and that means we need to create more content. So it's something we're in the middle of. I don't think this. I wouldn't declare victory just yet. We still have plenty of work to do.
- [15:43]
B
Absolutely. I do want to dig into that, but I want to pause and circle back to if you could give us a number for how often a brand like Oreo or Cliff now gets cited. Right. And what Tools are Mondelez using to track that.
- [16:00]
C
So it'll be very different by occasion. So it can be very, very small to very, very large. If somebody's going after a specific cookie based query, okay, we're gonna come up in the 70 percents, but if you're looking at a broader topic, like maybe snacks for a road trip, we may be coming up in the 30 percents. So it really does depend on the occasion. And this is where it's also important to make sure we choose where we want to win and where it fits with our brand, because it's not. We don't want to present our product in a place where it's the wrong answer. We only want it to show up when it is relevant to the consumer. In terms of tools, there are some amazing tools out there. We use multiples, we don't have one and we don't lock into any long term contracts right now. It's an ever evolving space. But BrandRank, AI or Profound are great tools that we're using today. We've also used Scrunch and a few others. We're still understanding everything that we have to offer and many of these companies are still fairly new and growing and changing. And many of our existing vendors are also creating new products to help us understand from another perspective. So as of right now, we're looking at a bunch of different items and we're paying attention to a few of them. But the tools have been incredibly powerful so far and very impressive.
- [17:40]
B
Take me back to the content production arm of this. Right. One of the things that I've been reporting on that I find incredibly interesting is that on the one hand, yes, brands are required to produce more content, whether that be blogs, FAQs, things like this. Right. But that's not all that the LLMs are pulling in. They're also going to third party sites like Reddit and even TikTok in some cases. Right. So how much emphasis do you guys put on owned and operated content versus some of the third party press releases, earned media and that type of deal?
- [18:10]
C
Yeah, it's a, it's a great question. I think this is an area where we still have work to do, but I'll at least give you an idea of how we're thinking about it. So I think about we need to shift from impression based thinking to citation based influence in the way we think about our kind of third party earned media type of an approach for earned media. It's no longer just about reach, it's about credibility. Source that AIs pull from that means more structured and factual expert led content that's actually answerable. Things like comparisons and FAQs and all these things that we talked about before. But I think the idea is that we have to worry about some different KPIs in the way that we think about our third party media and have again this holistic strategy. Because one thing that I have found is answers are highly fragmented. There'll be 30 different citation sources in order to create one answer for a fairly simple question on ChatGPT or Gemini or wherever. So oreoid.com may be one source. SnackWorks, another one of our websites may be another source. Walmart product detail page or Target product detail page or Amazon product detail page may be another source. But then there's also going to be Reddit, YouTube, blogs, all sorts of different sources. So it's about having a fairly broad approach to the way that we're working on the content in terms of where it shows up. We want it to have everything singing from the same song sheet, but additionally it's about a different form of content. We are very much used to doing video, brand emotional commercials, things like this. We actually have to build the muscle internally or through our partners to make long form text a big piece of our content business. So it's something we're working on. I wouldn't say we've solved it all yet, but we are publishing and pushing it out into the market today. And there's also an operations component to all of this. Large companies like Mondelez are structured in a way where people are responsible for particular areas of this. And one thing that we're looking at and thinking about is how do we restructure for the future? Do we need to have all of our digital content sitting under one house so that we're creating multiple pieces of content that's coming from the same brief and those pieces of content will live across retail.com brand.com and owned and earned media. So this is at least the way that we're going about it today and what we're thinking about so far.
- [21:08]
B
My colleague Sam Bradley did a really smart piece on this maybe just a couple of weeks ago that the lines between a traditional SEO and paid ads are starting to blur because they're holding hands now at this point as far as strategy. So talk to me, who, who's in the room? Who's at the table as you guys are working through SEO strategy, right? Is the influencer team there along with the paid team, along with the SEO team, Who's at the table?
- [21:33]
C
Look candidly it works very different in different markets. I'm coming from the global perspective. So in some places it is well, more integrated than than others. And this is, you know, something we're always continuously working on. But you've got a lot of those teams that you just mentioned that are typically in the room. So we also will have any like digital commerce will also be in the room which is a separate team typically. So they're thinking about the PDPs and whatnot. So the SEO type strategy is not just on Google. It's also the SEO SEM approach that we're taking on a retailer's website. Retail media has become a very, very important kind of search engine for the growth of every one of our brands. So it's a very holistic approach that we do need to take. But I do think that the Agentix search gives us the opportunity to step change in capability as well. So it's not just about, hey, how do we integrate this into the rest of our SEO type process Today it's actually, let's start from the ground up like first principle thinking and rebuild the way that we actually have to go to market for Agentix Search.
- [22:51]
B
Let's actually do talk about retail media, right? Especially as a CPG brand. As owner of CPG brands, one of the things that I keep thinking about is the idea like if I'm a consumer and I start searching for something like, you know, best cookies for a Super bowl party or something like that, the likelihood of me going to a Walmart.com or a Kroger Albert sense, right. To look for these things. Instead of starting my search journey there, I may now start it on ChatGPT or Gemini. Right. So talk to me a little bit about the value proposition. Right. What does that then mean for retail media budgets for a company like Mondelez?
- [23:26]
C
It's a great question. I think the truth is we'll have to see as we see these kind of retail media components develop out of the open AIs and Googles of the world and Rufus just released their ad clients. So there's going to be more advertising opportunities inside these LLMs. But I'll tell you, the way I see it, at least I think retailers are a very important part of Mondelez strategy to reach consumers. We've done DTC work in the past. We have DTC sites today. At the end of the day, the vast majority of our volume comes from our retailers and they're incredibly important partners to us and we believe our products work very well paired with other items in a basket. So it's still a great place from the transaction and perspective to take place. However, I think things like the LLMs and even social commerce or even Last Milers like a Doordash or an Uber Eats has changed the user journey in a bit of a way. Where you have these consumer platforms that actually sit at the front end of many consumer journeys, it's important that brands have direct relation, understanding and input and influence on those consumer platforms before they ever direct anybody to a retailer. So I think the value chain has changed a little bit as we've begun to see and I think it'll continue to change in that direction. And there are brilliant retailers who are adapting to this extremely well and they're likely going to be the ones that continue to grow, share and do well. Brands like us are very focused on making sure that we are as close to the consumer as we possibly can be. Those are great opportunities for us to get closer to the consumers, to spend more times in those social commerce moments, in those agentic commerce moments and the last mile or so as well. So we're spending a lot of time there to make sure that we have a good understanding.
- [25:36]
B
Absolutely. Doubling down kind of in that space. Has Mondelez or any of its brands started testing ads within ChatGPT?
- [25:45]
C
Not at this time. I think there's a few opportunities that we're trying in some other LLM, other LLMs. It's a bit cost prohibitive for Mondelez at the moment, but it is something that we're exploring. I don't think that there's anything off the table. As you can imagine. These are things that are only a few months old. So we have planning cycles that go on that take some time for us to make sure that we're making the right choices. So we will have pilots at some point in time when it makes sense for us. In many of these platforms, some of them we've already begun, but not chatgpt specifically.
- [26:30]
B
Got it. Would you be able to tell us which ones you have begun? I know Perplexity kind of came and went, but Copilot and some of the others are also offering ad opportunities.
- [26:39]
C
Yeah, I think there's also great opportunities in the retailers LLMs. So if you think about a Rufus or Spot. Yeah. So I think those are really logical places for CPGs to spend some time and money to make sure that we have good visibility there.
- [26:57]
B
Absolutely. Have you guys laid out what a framework for success metrics looks like as you're testing these?
- [27:04]
C
We have some basics, but again, we're nowhere near scale. This is pilots at the very beginning of the journey. I expect over the next year for this to start to materialize into substantial business and at that point have me on in a year. I'll tell you more details there.
- [27:24]
B
We'll hold you to it. Andrew, I want to talk future facing now. Right. Because one of the things that also Jen said at the February IAB meeting is that Mondelez is expecting 20 to 30% of product purchases to be made via I A I agent, excuse me, in the next year, year and a half. How is Mondelez prepping for that?
- [27:46]
C
Well, look, again, I think the, the ultimate goal of our strategy is to make sure that our products are being recommended by the LLMs. I think the next step of where those transactions take place is if it happens inside of an instant checkout type thing that ChatGPT may bring back, I've gotten rid of. We'll see whether that's a referral, link off to a Walmart and somebody buys on the retailer directly or if that happens through one of the app stores inside of the LLMs. I'm impartial to the way that that particularly goes. I just want to make sure that we're being recommended in the right place. When it comes to how we're preparing or how we're thinking about this, we're very much focused on that recommendation piece, making sure that the clarity of the technology is there, making sure that we're creating the brand. Our brands are long, deep, nostalgic brands in many ways. We need to make sure that we can translate what humans think of when they think of Oreo or Cadbury or Milka and into what AI thinks when they're going after these brands. So that's not an easy task. That's actually fairly complex that we have to find the right way to bring that data together. So we're very focused on the recommendation piece and making sure that we have the right structured data to get us there, the right content to connect to our occasions, and the right holistic operations to make sure that we get bring all this to tell one kind of story. And that's where we're spending the time today.
- [29:42]
B
Absolutely. Not for nothing. 20 to 30% of product purchases made via AI agents, that's a big bet. What is the likelihood of that reality? What signals indicate to you that that's what's on track?
- [29:56]
C
Look, I'll say this. We have no expectations of agentic commerce increasing the pace in which digital commerce adoption is going. But of those who have adopted digital commerce, the odds using LLMs is quite high. So this is more than a three and a half billion dollar business for us in terms of digital commerce. So 20, 30% of our digital business moving in this direction is a huge, huge, huge change. We can all debate on timelines internally, we'll debate on timelines, we'll have base bull bear cases of adoption rates. But even at the bear side of things, it is a substantial piece of the business that I really don't. There's not too much question around will it happen? The question just comes up of when it will happen. And that's really what we tend to focus on over here. Which means let's get organized, let's get ready and let's make sure that we're ready to meet that moment when it comes. If it comes next year, if it comes in three years, we are fully prepared to make sure that if we have 20, 30% of our volume moving through Agentic that we're, we're there to win.
- [31:14]
B
Absolutely. I appreciate you walking us through what this all looks like for Amanda Lease. Although it's early days, I will indeed circle back in a year and we'll see what this looks like. But thanks for joining us in the meantime.
- [31:27]
C
Thank you so much. I appreciate it. I will definitely come back in a year and you can hold me accountable for all the new work that we have to go do. But it's really exciting times. I've been doing digital commerce for a very, very long time now, and this is definitely the most exciting time I've seen yet. Agentic is changing the game and it's even reinvigorated my own excitement around the area. So thanks again for having me. I hope this is helpful to some listeners and look forward to chatting again soon.
- [31:58]
B
Absolutely. Well, that brings us to the end of this episode of the Digiday Podcast. Thank you to everyone for listening. And please don't forget to share this episode with someone who you think would enjoy it. You can even rate us and leave us a comment on Apple Podcasts. We'll be back next week with another episode of the Digiday Podcast. Thank you so much for joining us.