Marketing 911
Marketing 911, is the podcast where we tackle the toughest marketing challenges at the executive level.
Whether you're navigating complex strategies, trying to reach your target audience, or facing shifting market dynamics, we're here to provide you with actionable solutions.
From digital transformation to customer retention, if it's a marketing crisis, we're here to help you solve it—before it turns into a full-blown emergency.
Marketing 911
AI at Scale: Personalization Beyond Automation - Live from Marketverse
Scaling personalization across 150,000 SKUs worldwide sounds impossible, but for Fabio Ranieri, Head of Go-to-Market Innovation at HP, it's exactly the challenge his team is tackling with AI. Speaking from Austin, Texas, Fabio reveals how HP has moved beyond basic automation to create truly personalized experiences that resonate across vastly different global markets.
"Content needs to be less of a seller and more of a helper," explains Fabio, highlighting the fundamental shift in how HP approaches customer communications. Rather than simply translating marketing materials, they're developing sophisticated digital twins of customer segments to understand how the exact same product might serve completely different needs across regions. A high-performance laptop marketed to gamers in the US might be the perfect tool for developers in India – requiring entirely different messaging despite being the identical product.
The conversation takes fascinating turns as Fabio explains HP's pragmatic approach to implementation: "We went bigger by going smaller." By focusing on super-narrow AI use cases rather than broad applications, his team minimizes hallucinations while maximizing impact. Equally impressive is their collaborative relationship with legal teams, built on transparency and education rather than confrontation. "It's not a fight, it's a partnership," Fabio emphasizes.
From machine-to-machine communication that knows when your printer needs ink before you do, to the careful balance between innovation and responsibility, this episode offers rare insights into how a technology giant navigates the AI revolution. As personalization moves toward true one-to-one experiences, Fabio's measured approach demonstrates how thoughtful implementation can drive remarkable results while maintaining brand integrity. Ready to rethink what's possible with AI in your marketing? Listen now and discover the strategies that could transform your customer experience.
Hello everybody. This is Brian Backstrand for Marketing 911. We're down in Austin, texas. I'm here with my co-host, richard.
Speaker 2:Yeah, brian, it's great to be here and we're doing a series of these podcasts with a variety of guests.
Speaker 1:What's great is we've never got a chance to actually meet our guests in person right, you and I have never been in the same room when we've done a podcast In our life. That's right.
Speaker 2:Yeah, we've known each other quite some time. But yes, was that on purpose, brian? Yes, well, we're kind of excited today because we've had a chance to meet a lot of fun people and Fabio we've gotten to know a little bit better. Sorry, you're going to introduce Fabio. We've gotten to know him a little bit better than our normal guests A hundred percent.
Speaker 1:We are on the stage today doing a little panel discussion and Fabio is terrific. So we are here with Fabio Renieri. He's the head of go-to-market innovation for HP, a major global company. So welcome. Thanks so much, and today we want to talk about the world's favorite topic, AI, which we were talking about today on the stage. I think you said a number of things on stage today that were really interesting around how you're using AI within HP. Why don't you just touch on a couple of those?
Speaker 3:Yeah, so in HP, we have a challenge, which is the scale right. Again, delivering something that is relevant to people everywhere for all the products that we have is a challenge on the volume itself and some of the things that we're doing in HP. Again, delivering something that is relevant to people everywhere for all the products that we have is a challenge on the volume itself and some of the things that we're doing in HP. Again, we use AI as really an answer of the marketeer itself, but also we are using a different way, which is, again, how can machines, in a nutshell, improve what is standard automation of content creation with AI? So, before think about it, you could do the dynamic content optimization in a way that is just basically bringing one ingredient and adding coffee on top of it.
Speaker 3:We are right now looking to how you can do that with AI and in a way that it can be done delivering personalization to someone in India, someone in Brazil and, in a nutshell, kind of speaking their language, not just from a translation perspective, but again aligning that the product is perfect for them, explaining and really being more relevant on once they're buying a product, guaranteeing that they're buying the right product for them.
Speaker 2:I ask this question because, again, this is beyond the moments of our interview, because at dinner last night you've talked about it and the panel we talked about it. Fabio and you talk about this personalization, because it's fascinating, because we're not just talking about adding their name at the top of an email and personalization, to some degree, has been like okay, dip into this database, pull this piece out, dip into this bit. We've seen that in the past. As you refer to it, automation right, but you're talking about something at a whole different level and scale, that is, introducing a new concept and ability.
Speaker 3:Yeah, I think we know automation. We've been doing automation for years right, and, by the way, if you do automation right, it's amazing, it's cheap. It's cheap, it's super cost effective, but again, ai is allowing you to go. I would say, deeper. Right, it's understanding the persona and giving an answer to that specific persona. Not giving copy to them, but an answer. And I think that's the biggest difference when you say an answer is it?
Speaker 3:an offer or are you responding to a question? I'm guaranteeing that the content that they see answers the questions that they probably have in advance by doing the analysis before and in a nutshell. For example, I'll give an example on something that happens Not everyone really knows what is the perfect laptop for that. Can I guarantee that once I'm giving content to what would be the specific audience of that product, I'm answering the questions that we know that they have? For example, if we search or scrape user-generated content or if we scrape everything that we see on, can I guarantee that, once that product is out, answer the questions that, for example, rich Richard has once he's buying I don't know a PC for your house or a printer or something like that. We kind of have that answers already. We kind of know. The only challenge is it's really hard or it was really hard to do content creation at that level. That will answer to Richard's questions, not Bobby's questions.
Speaker 2:I can say so my HP printer talks to me all the time, okay, and it's doing it through Amazon. So Alexa is talking to my printer. That's telling it I'm low on toner. And then Alexa is now telling me hey, you should be ordering black toner because your HP printer just told me that it's running low. Now that's not necessarily AI, but it is a bit of an automation, but I got to tell you. I'm like, yeah, yeah, yeah, I already bought it. Okay, stop, I already bought it. But it's really fascinating how the machine started to talk to the machines and then inform us, keep us, because it's not trying to tell me to buy something else, it's telling me very specifically what I need. It knows what I need and it probably knows how often I'm using it, because it's watching that tone go down. There's all kinds of things that personalize it there. I know that's really simplified what you were talking about, but I suddenly was struck by the fact that, oh, yeah, it's a real life.
Speaker 3:It's a real life. It's incredible to you because you would hate to be out of ink. For example, right, right, for example, hp launched instant ink. Instant ink is a program that, in a nutshell, you basically get owner or ink delivered to your house again and you don't run out of ink, because usually what happens then people start to hate their printers is again, by the time that you need you are out of ink and you that it's on you. You did not buy ink. Once you need it, you said, no, I'm going to buy later. I'm not going to focus on this right now, but once it happens, it's something that does not make you happy with the product that you have it happened to me?
Speaker 1:the other day I ran out of ink. Oh, my mind goes to HP. Doesn't give me enough ink for what I'm doing.
Speaker 2:But also, Fabio, you brought up an extra point at dinner. You kind of mentioned it in our panel. We're not just talking about ink. How many SKUs are we talking about?
Speaker 3:Once we count all the SKUs, all the languages and the way they're structured 150,000 SKUs 150,000.
Speaker 2:So, while I think it's a simple process for them to figure out what I need they know my printer, they know my ink, they know my toner they're going to give it to me, but 150,000 different SKUs around the world different needs.
Speaker 3:That's what we're talking about Exactly. And that's the challenge, because if you think about, the same SQ that is sold in the US might be used by a different audience in Romania. Give us an example of that. I'll give an example with cars which we are used to. So the same way that we sell for example, let's talk about the audience of Corolla in the US right there's. And again, if you look at the same car I'm going to give an example in Brazil. In Brazil, that car can be a car bought by a executive. Same car, a little bit over spec in Brazil, but two different audience for the same product itself. And the same happens with a computer. So, for example, something that for in the US can be for a gamer, in India can be for someone that is developing code, because they're really high in processing power, for a legit price I will not say cheap, but affordable price. That makes sense. So how do you?
Speaker 1:gather all of this and continue to grab. Gather all this information on the millions of customers.
Speaker 3:So the thing that we're looking for is the creation of the digital twins, right? So what would be the four, five, six audience that I would have in a specific location? And what they're looking for? It can be user-generated content. It can be content that we have inside HP, but again, we are still not pursuing that, what we call one-to-one personalization, even because the e-commerce platforms are not ready for that. Think about content from a first perspective. It's one content at one country. We're trying to move to the next levels. How do I understand that? In your specific audience and I give content that is relevant to you and the nice thing about content that is a little bit different than, for example, advertising you can do that really, really easy. But content it needs to be less of a seller and more of a helper. The way that we see content today that really works is it's something that helps you get the right product, that makes sense at that time for you. But it speaks your language, not mine, because my language is stack-driven, but not necessarily you know what we speak.
Speaker 2:And we're not talking about Portuguese versus English versus Spanish. No, we're talking about my lifestyle, my focus how can you use words that resonate with me and my life and my industry and what I'm doing? That's what we're talking about.
Speaker 3:It's really a use case. Usually you're buying a machine because you have a specific or desired need to it, right? Um, for example, I'm going to give an example that happens at my home. I used to have a really uh, high-end laser printer because we used to work, uh, and my wife used to print like crazy because of her work and, again, she loves the concept of really using paper and doing annotations on paper, etc. Now we have two kids. The amount of ink that we use is out of this world and we move to a different picture.
Speaker 3:By the way, I think we are one of the houses that have one and one printer. We have two printers one for the kids, which is not common, but it's a different use case and sales is now don't even have a printer. So I imagine that, yeah, I have to adjust to compensate. So one of the things that for me is I'm one I'm looking at a specific use case where I've worked with for development, etc. Etc. Etc. Um, and the other one is I have kids and they preach like crazy because they love to draw and, yes, how do I speak that language? How do I speak the language of the benefit, not a language of the feature itself, you know. But again, it's a use case based discussion. How do I carry content for everyone?
Speaker 2:I gotta tell a story here, because I was speaking, the uh, the cio of pinterest was speaking to a group and I was in the back of the group standing next to his 10 year old son. So I talk about technology because we're all. We're all technologists, we all. That's our language. And I asked the son of the CIO of Pinterest do you know what your dad does? He's like no, and I'm kind of trying to explain it to him. And somehow I mentioned the concept of a fax machine. Do you know that 10-year-old's mind was blown when I told him what a fax machine was? He couldn't believe that you could take a piece of paper, stick it into a machine and that that piece of paper would pop out somewhere else in the world.
Speaker 1:He thought it was the same piece of paper.
Speaker 2:Yeah, he thought it was the same piece of paper. It blew his mind. As we talk about technology, it's moving so fast, it's changing so much, and one of the things that we've talked about in the past with you was AI, and how do you stay on top of that AI, how do you make that part of it? And so you've talked about how it's machine to machine, but there's challenges that you're facing, aren't?
Speaker 3:there. There are, and one of the biggest challenges that we have is that AI can do a lot of pain these days, but one of the really the biggest one is it tends to not look at really specific elements on a product that might generate returns, for example, or bad customer satisfaction. So, for example, if I say AI, grab this laptop and create another image out of this laptop, a product image with white background, etc. There's a big chance that it's going to tweak a little bit the product. So the way that we're doing today is what we call the super narrow use cases, and instead of having AI generating a whole picture, we ask AI to generate partial so it can guarantee that the product is 100% accurate. So that's how we were able to exactly move into a more scalable solution by narrowing the use case.
Speaker 2:That's interesting because you were able to go bigger by going smaller. The first, by keeping it focused, not having it just do anything but training it skill. But training it skill, preparing it and giving it the data that it needed to do a very specific thing yeah, highly specialized.
Speaker 3:all the use cases that we get out they're highly specialized and with that we reduce hallucinations, which are usually the problem.
Speaker 2:For images, text content yeah, across the board. I'm interested in something that you talked about legal, the relationship you have with legal when it comes to AI. Can you expound on that, because it came up at dinner and I found it fascinating, because it continues to be a challenge for a lot of companies, that the legal department is struggling to keep up with the compliance, the privacy that AI is introducing.
Speaker 3:Yeah, I think we are lucky at HP that we're not getting to know. We are getting the let's work together. And part of the narrow use case was exactly the conversation with legal. For example, can I go LLMA, can I go LLMB, can I go generation A, generation B? That was a long conversation with them to guarantee that it was what explaining them what is possible today, explaining them what is possible tomorrow and working with them to understand what is acceptable.
Speaker 3:Okay, and one thing that we do is to measure, for example, is if we're gonna go and launch something that is AI based, there's a big chance that we're gonna run tons of tests before we say we are first ready to go with those patients and, second, to understand how customers are perceiving that. So, for example, for lower-end products or products that do not have big volumes from a worldwide perspective, we try some stuff that we never did before that now are capable through AI, and we presented that to legal and said tell me what you believe that you would not recommend and those things we work with damage guarantee that it was a lot fine to the needs that were, and I like what you're saying because you experiment a lot, a lot, and you put the put it in front of the legal people so that they could actually take a look at how they are not, let's say, a gate approach.
Speaker 3:They're part of the process. That's the nice thing. I like that.
Speaker 2:And both Brian and I have seen both sides of that. We've worked with innovative legal teams that are part of the process and then working with legal teams that are the Department of no, and so what a difference it is when you're able to work with the legal I like that.
Speaker 3:But the most amazing thing was we brought them in. I think it was not a can I use this? It was no. Let me show what is going on on the market. Let me show you everything that we have. Let me show you the capabilities. So, for example, there are some capabilities that we presented, that we discussed yesterday that they say no, yeah, and we understood. So we were not pushing for things that we knew that would disrupt anything that HP, from a corporation branding its perspective, would not agree.
Speaker 1:I mean legal. The last thing they want is for something to happen and then they're in a position where they have to work hard to correct it from a legal standpoint and you're a great partner to bring it to them. That's why they're so thrilled you're a partner. Yeah, they don't want a surprise and they don't want to have to fix something. They'd rather just be able to.
Speaker 3:And there's a reason why they're there. I think this is something that is not a fight, it's a partnership, that something that is not a fight, it's a partnership.
Speaker 1:I like that. So one of the things we talk about here on 911 is you know, you've got a huge job at a global brand. What are some? What's a crisis or major issue that either has happened or within your scope of responsibility you worry about happening?
Speaker 3:I think the one that I worry the most is exactly some major hallucination out of AI that we did not plan to. So that's why everything that we're doing, we're testing and experimenting. Like Tracy before, I would say let's go live. So, for example, I think the one-to-one personalization, it's going to be a challenging one, because you cannot have real personalization like I'm gonna create a video just for you, um, I'm gonna create a video just for someone that is in brazil, in the middle of the country, trying to understand how to use that receipt or that printer to make their lives better. Right, and you have no human oversight. That is still question mark and it's still a crisis that is about to come and I think some people will, some companies will decide to accept that and there are some crises that we're going to see in the future. I like that 100%.
Speaker 1:Well, listen, really appreciate the conversation. Appreciate having dinner with you last night, on stage today, you did a terrific job. You did a terrific job today. So this is Brian Backstrand. We are in Austin, Texas, and this is Marketing 911. Myself and Richard wish everyone a good day. Thank you very much.
Speaker 3:Thank you guys, Pleasure to be here.
