AIエージェントの台頭：TIME COO マーク・ハワード氏に聞く「エージェント向け広告」とAEO

The AI Today podcast, powered by the Project Management Institute, cuts through the AI height to deliver you proven strategies, real-world examples, and actionable insights that drive AI project success. Tune in to learn from PMI hosts and guest experts. This episode of AI Today is brought to you by Tempo. AI is changing what companies set out to achieve, and Tempo changes their ability to deliver it. The world's most ambitious companies run on Tempo. They evolve strategic portfolio management, so people and AI agents work from one view of strategy, execution, and value. Tempo, ambition delivered. Learn more at Tempo.io slash PMI. Hello, and welcome to the AI Today podcast. I'm your host, Kathleen Walsh. And if this is your first time listening to AI Today, then welcome. And I'm so excited that you found the podcast. And if you've been here before, then welcome back. We are now well into Season 10, which is wild to think that AI Today has been around for a decade now, still focused on AI today.

Now that we're a part of Project Management Institute, our podcast is on PMI.org, and that's really exciting. So you can go there or you can go to your favorite podcast platform and find it as well. As we continue to integrate into PMI, we'll be interviewing and engaging with various thought leaders, AI practitioners, and project professionals to highlight both the AI transformation of project management and the project management of AI transformations. I always love the chance when I get to interview folks in the industry and have really wonderful conversations, so I'm so excited to have the opportunity to speak with Mark Howard, who is COO at Time. So welcome, Mark, and thanks so much for joining me. Yeah, thank you so much for having me. I'd like to start by having you introduce yourself to our listeners and tell them a little bit about your background and how you got to time. Great. So I've been at Time for almost three and a half years overseeing the finance, tech, product, data, consumer marketing, print, time for kids, and a lot of the go-to-market strategy as we think about our business and the evolution of traditional media

and what that even means in today's world of the Internet and AI and the agentic web. And we'll get into a lot of that going forward. But it is absolutely a time of tremendous transformation, as it is for every industry. I'm a longtime media exec. I did a stint at a company called Penske Media, which is much more the entertainment and lifestyle verticals. And prior to that, I spent nearly two decades at Forbes in various commercial roles, including almost a decade as the global chief revenue officer. So I spent a lot of time in the media space, cared deeply about journalism, and for quality, independent journalism to be able to survive and hopefully thrive in this new world. And I spent a lot of time today thinking about bridging the old way of creating quality content that only people can do, that people are still required for being able to get to the facts and the information that is going to be so critical in this world going forward.

However, realizing that the way in which people individually are finding information, having information assembled for them in various form factors is all changing very radically. And while we like to think each stage of transformation is the most significant that you've ever experienced, there's no doubt that the pace of change now is far greater than what we saw, even with going from digital to social and mobile or prior to that, print to digital or, you know, any of those major moves. So the change is happening quickly. it's very dramatic. And like every transformation, it's questioning all the business models and, I believe, creating new opportunities. Yeah. I mean, that's a really great background. You've seen a lot as well working in the industry for now a few decades. So I like how you can bring that perspective. And one thing that you had touched upon, right, is just the ways of working

is changing and what everybody honestly really needs to do. So Time was one of the first major publishers to launch advertising specifically for AI agents, right? A lot of people talk about AI agents, but now we're advertising to them. So what led to that decision and what changes in consumer behavior or technology convinced you that now was the right time to experiment with this? Yeah. So starting at the end and then I'll work backwards. At this recording, Time is currently the only media company that is actually serving ads to agents. And full disclosure, PMI is one of the two partners that launched with us and took that big, bold step when we realized the big opportunity. But let me go back a little and then bring us up to where we are today. This is part of a long journey. Three years ago, when I arrived at time, we started tracking the AI agents and bots that were coming to our website.

And at the time, it was a small number every day. And we knew it was important, though, to have that layer of measurement and analysis happening to see whether it was Claude or Gemini or ChatGPT or Perplexity or, or, or. And there are so many. And certainly over those three years of tracking and monitoring that, a couple of major things have emerged in the industry. The first is Cloudflare, which is one of the major CDNs, led a media industry initiative to say, hey, by the way, we shouldn't just be letting all of these AI agents and bots onto your site. They're crawling your pages. And in a lot of cases, they're taking your information. And that's very valuable IP, and we shouldn't allow that. And so we participated in what they called Media Independence Day, And we started blocking all the AI agents and bots with the exception of a few. In the course of two, three years ago, we started negotiating a couple of content licensing deals.

We did four of them. We did an open AI licensing deal. We did an Amazon licensing deal. We did a perplexity licensing deal. And we did one with a company called ProRata. So we learned a lot about what the LLMs are looking for, the information they want. We learned about how to actually create feeds of all of the content that we were legally able to license them and that they could use. And then we gained access to executives at those companies. We started joining quarterly roadmap meetings. We would have annual meetings with C-level executives. So we got an opportunity to learn what was important to them by having a seat at the table with direct access and direct conversations. That became important information for us in terms of just understanding what mattered for their businesses and how they saw the evolution of their products. So we had the infrastructure. We had the licensing deals. We started blocking by default.

Then we took it one step further and we said, well, actually, we don't want to just blindly block everybody. We actually do need certain agents and bots to be able to come into the site. So we switched to an allow list. We now have about 70 agents that we allow into the site unobstructed. All the rest, there's almost 3,000 of them that come in the course of any given month. We route to a paywall, a digital paywall. None of them pay, even though we asked them to, and we would grant them access to the information and the usage of that information if they did. There's an arms race going on. They have better technology than we do. They can circumvent our blocking, and they still come through, but that's a whole other thing. But we allow about 70 agents and bots to come through, And these are the ones that we deemed as either the good ones who show up at the CDN. They tell us, hey, I'm open AI. I am here for training or I am here for AI assist, which means real time information retrieval.

Or I'm clawed and I'm here for indexing and training or whatever it may be. And so many of them, the good actors will show up. They'll declare what they're doing. We understand who they are. We see the pages that they're consuming and what they're using that information for. That created an opportunity, which then got us to the point where we were able to introduce ads. We started separating the traffic. So when humans come to the site, they get the full HTML experience that all of us are familiar with. If you go to time.com, when the agents come to the site that are on the allow list, we send them to a machine-readable version of the page called the Markdown page. And the Markdown page removes all the JavaScript, all the images and the videos and the ads and everything that the agent isn't interested in. And we give them a very clean version of the content that includes the metadata for the page up at the top,

makes it easily accessible, and then word for word exactly what the content of that page is. And so we built this by design It makes it easier for the agents to be able to get to the information It less token usage for them to be able to get that information So we believe it's a better, more optimized experience, and we believe over time that will be valuable to them, and hopefully we believe that they will favor those types of experiences. What that also introduced is a new surface, And that new surface, we believe, is equal to or even more valuable as we think about the agentic web than the human version of the page. And there's a lot of reasons why we believe that. But first and foremost is when we think about marketing to humans, the objective was always figure out how to intercept attention, use images, videos, stories, narratives, emotion to make an impression on a person to get them to

remember your brand or to have positive feelings about your brand or be up to speed about the offerings of your products. And so that's a model. None of that applies to how agents learn. Agents are 24-7, they have endless patients, and they have no emotion. And what they want and what they're ultimately looking for is structured data delivered in data tables and FAQ formats for easy digestion. And once that has happened, they're able to store that information either by caching or indexing it so that they have it available to be able to be presented whenever any kind of query or search or prompt happens, that that information could be useful to a valuable answer. And so that was the genesis of agent ads. This idea of we have machine readable versions of

our content, and now we've created what we call verified brand facts presented in structured data formats on those pages for the agents to consume the same time they're consuming our content. Yeah, you know, I think that this is, I mean, super exciting times, but also something that traditional advertisers, traditional marketers really need to wrap their head around. I think even people in general, right? I know that you've talked a lot about this growing divergence between the human web and the agentic web. And that's an important distinction, right? Even as you're looking to set up content, now you have to set it up in two formats, at least for now. Who knows if there'll be more in the future? And so that even looks different. How do you see these two ecosystems evolving? And then what should organizations be doing today to ensure that they remain visible and relevant in both? Yeah. So we saw the diversion happening slowly at first, and now it seems to be accelerating.

We're now in a position where many days we'll see more agents reading our content than humans. It's not every day. And there are certainly big days where we release a very important interview. We had one last week. We not only introduced the interview but also the transcripts of the interview. And what was amazing is, of course, the interview itself was picked up by news outlets all over the world, so people were able to understand what some of the highlights from the interview were, we saw a huge influx of people who came to the website to read the transcripts. And we think that that is a really important signal in terms of what still matters for humans today is that original information transparently being provided for them to be able to go and make their own conclusions, verify for themselves the information. that they're reading. And I think that that's a strong signal. What's really interesting is

that transcript page did not have a huge volume of agents reading that content. On the flip side, we've released several franchises over the course of the last couple of months, Time 100 Creators, Time 100 AI, most recently Time 100 Art. And we do see a big surge in agents coming to those pages on the days of launch and subsequent days shortly thereafter. And so it's just really interesting to study the variations in the behavior. But Cloudflare is now saying that over 60% of the Internet is now made up of agents and bots versus 40% human. They have some pretty astronomical projections of the agentic web 1,000 X-ing in the next four to five years, which would essentially make human web traffic a rounding error in terms of the size of the overall web. And so we are moving into a world where the agentic web is the Internet.

These agents, certainly we've all experienced it in the last couple of weeks with the introduction of Muse and Instinct and GrokBot and Work. And soon there will be in Dots most recently and then there will be others. And we're all trying to figure out what that means. But what it really means is we're all going to have personal agents that 24-7 are out doing things on our behalf, and they're going to be hitting websites, and they're going to be retrieving information, and they're going to be building a knowledge layer that brings that information back to the people that they're working on behalf of. And so being able to think about this agentic traffic now as a new form of consumer is where so much of this is going. And we believe that being early in terms of figuring out models to treat it like a high valuable audience, as opposed to the way that bots were treated on the Internet for the first 20 years of the Internet, which was they were a problem. They inflated all of your metrics. You needed to weed it out. In fact, you needed to figure out how to block them.

Now we're moving to a place where not only do agents and bots form opinions for people and present it back to people, but soon they're going to actually be transacting on behalf of those people. So they are going to be a new form of consumer, and I think we need to start treating them that way. But it really is a bit of a mind bender to start thinking about what powers agents, how they operate, how they obtain information, how they store information, how they display information, and how radically different that is from the way that humans process information. Yeah, and I think that that is hard to wrap your head around, right? And just the volume of them, like you said, especially, you know, maybe not today, but what does it look like three months, six months, one year from now? I mean, it's going to be totally different in ways that maybe we can't even expect. So for organizations that aren't forward-leaning in this yet, what should they be doing or what should they even be thinking about to ensure that they remain relevant and visible to both of these different webs?

Yeah, as you can imagine, as we've introduced this idea of agent ads to the marketplace, we've spoken at this point to dozens and dozens of different brands who are all at various stages of their own journey. And some of them have gone so far as to create their own markdown versions of their sites by creating an LLMS.txt file that the agents get routed to, which are presented very similarly to the way that we've changed the form factor of our content. Many brands are doing that. Some believe, the same belief that we do, that by making it easier for the agents to get the information and the form factors they want, that they'll more accurately represent them. It's not 100% proven yet, but that is a strong working theory, and we certainly support it. People are thinking about this across all the different departments. I speak with CTOs. I speak with chief communications officers. I speak with chief marketing officers. We talk to SEO teams.

We talk to paid media teams. And everybody at this point, PR firms, ad agencies, SEO optimization firms, Everybody realizes that this is a critically important moment in time to start formulating strategic approaches to how to be thinking about this. I think at this point, everybody to some capacity has an AEO challenge. We work with a media marketing ad tech firm called Mobian, who has been a deeply integrated partner of ours for three years on the digital media side. They were actually our partners in coming up with the agent ads concept and building all of the infrastructure and how we deliver them. And what they've done is they built a framework for brands that we approach every brand with that is called a VFA. It's visibility, favorability, and accuracy.

And what we do is we analyze, Mobian analyzes paid advertising for a brand. So it's what are they putting money behind to convey to people and to the world that their brand stands for, their brand attributes, their product attributes, their intended consumer, their key messaging. and we take all of those key values and then we run them against all of the major LLMs hundreds of times a day over a period of time to see how the AI platforms speak about those brands When asked a relevant question, does your brand show up? Visibility. When you do show up, do the AI speak favorably about you? And then maybe most importantly, are they accurate about what they're saying about you when you do show up. So even though the sentiment may be high, the accuracy of the information could be off. So we lead with this VFA framework, and Mobium powers all of that, and it's a door opener to every conversation.

And we have conversations with brands that just simply don't understand why they're not showing up as often as their peer set. But when they do show up, they're spoken about very positively with high accuracy rates. Or other times we spoke to a company today where they had a product recall about two years ago. They've since fixed the problem. Everything is going great. All the consumer reviews are fantastic. So they're showing up with high visibility, but the favorability and accuracy are extremely low. And so that's a different problem that they now need to tackle. And that may be even more severe because they're being misrepresented with something that was an issue in the past that has since been resolved and is actually something that is doing quite well for them from a product perspective today. So that leaves more questions than answers for a brand about what do we do and where do we go from here. And so we believe that we are on the cusp of a burgeoning industry because, sure, you could make an ad buy with OpenAI,

But the irony there in terms of what we're talking about is that is still display advertising for humans. We are talking about the actual chat-based experience, the AI answers-based experience that is being generated and actually impacting what comes out of the models on the back end. And so very, very early days. I actually, while we're first and we're excited about that, we believe that the entire industry is going to have a version of this, of what we're doing. and that if labeled properly and stick to the concept of this isn't marketing speak and hyperbole, but it's actually brand verified facts with sources and the ability to give it to the LLMs and the AI in structured format, that over time we will be able to improve the output from the models. And by making the models more accurate, it will be a beneficial outcome for everybody.

Yeah, so much to unpack in that answer. I loved it, especially when you talk about this idea of visibility, favorability, and accuracy. You know, and you talked about SEO versus AEO. So SEO, I think everybody understands, right, search engine optimization. Google kind of owned that space for a few decades now, right? But now we've moved to AEO, AI Engine Optimization, which is very new. And I call a lot with AI the Wild West, and I think that this still is too, right? It's still settling. And so how are you going to win? And one thing that I've seen is one thing that's nice sometimes about AI but also can be frustrating for brands is that it provides limited options, right? You can have pages and pages of Google. Now, yes, of course, you want to rank up at the top, right, the traditional old school Google. You want it to rank up at the top because people aren't going to page three and four, but it shows up. And what you wrote had a direct kind of narrative. But now with AI, right, it's trained on data.

You're not exactly sure what it's doing. There can be biases. And then as you were talking about, oh, well, there was some maybe favorability challenges due to recalls that happened a few years ago. Sometimes old news resurfaces, right? And now these brands can't get control of that. And what happens with bad actors? When they're doing bad things, it seems like it's from a trusted and verified source. And now they're, you know, trying to poison the knowledge in certain areas. So there's so much to unpack. We're all still learning, which makes this incredibly exciting, but a lot there. Oh, yeah. Right. And I'm like, I'm sure you've probably seen a little bit of it, but it still is so new that I don't even know that there's enough data out there really to see the impact of this long term. Yeah, we learn every day. And not only by having these conversations with brands, we've also now spoken to several of the LLMs directly and shared with them what we're doing,

because we believe that what we're doing is the most transparent, straightforward approach to wherever you sit on the belief spectrum about the foundation models, the frontier labs have stolen the IP of the internet to it's justified under fair use. And we'll find out with some marquee cases that are going to be determined here pretty soon. You know, we'll have the courts respond to that. Regardless, the frontier labs, the AI LLMs, the AI agents that are crawling the internet are using brand and media and creator IP to help train their models, improve their models, and provide better outputs. So that is the reality of what is happening right now. And it's definitely left us all with more questions than answers. I think every day that we're living in

this space, because don't forget, we live in it in two ways. We are also a media company that thinks about our own presence of our content and how often we're sided with time content in AI answers. And that is, that's our AEO and is critically important to our brand in terms of same thing from a BFA perspective. When there are queries and we've written authoritatively on various topics or subjects, and they're answering for people, are we in the answer? Are we one of, and similar to Blue Link SEO, where the higher up the Blue Link was, it was perceived to have been higher value response and higher value to the consumer to be able to click through to then go get the information. There are sources in the AI answers, and sometimes they're exposed, Sometimes you have to open them, but there are sources. And I think the thing that is the most eye-opening to most of the brands that we speak to is when we do this constant querying based off of prompts or questions that align with their brand values or product attributes,

When you see some of the sources that are factoring into this otherwise blended response, and you look at these very obscure websites or these tiny YouTubers that effectively have a very small following and somehow are making their way into this blended, summarized response, it's pretty mind-boggling. And so there definitely have to be a lot more ways in which brands are able to work with the LLMs and work with other solutions on how do they continue to have the best representation. And I think for us, we've really honed in on this idea of the most accurate representation of who they are. It's why we've doubled down on the product being brand facts, verified by the brand. every brand fact has to be correlated back to a source where the agent is able to actually, if they follow the URL back to the source, they can verify the brand fact that's presented in this

ad. And so when we presented this whole concept all the way down to that granular level of detail, the LLMs were actually positive about the fact that we were doing this because accuracy is critical to them and being thoroughly accurate is also critically important to them but they don't have the ability to actually index the entire internet to that degree and have all of that ready at their recall for every single query so how we present all of that as long as we stick to that approach and and you know the interesting part as you could imagine is we still sell this product to humans. And so you're asking humans to put on their think like an agent hat. And they're still very much as much as everybody is focused on this topic, or we find the people who are focused on this topic, you're still selling to humans at this point. And so it does require them to take a little bit of that, that leap of faith. But we put a lot of thought into how we

how we measure it. I gave you kind of the opening positioning. We also do a couple of different studies for our partners. One is a causal lift study, which is sort of a contained control exposed. If when we go to the LOM, same question, same set of questions, you ask them a question without exposure to the brand facts. And we just have them, whatever their mechanism is for returning an answer. Can they answer these questions? And then we do the exact same set of questions when we give them exposure to pages on our site that have the brand facts And when they have the brand facts how does the outcome differ between expose and control And so that is a baseline where we are seeing repeatedly that there's a lift that happens when they have that exposure. The second is, over time, a model impact. And that will be the thing that will be the most important over time.

is can we, by repeatedly putting these brand facts in front of the LLMs and the AI and the agents, can we show that over time they are, in fact, with repetition, learning that information, making it better presented when appropriate to satisfy a query or a prompt, and can we show statistically significant improvement over time? Truthfully, we're only two months new to this whole thing, so I don't think we're anywhere near the amount of time that needs to exist. The same way no brand would launch an ad campaign and think that in the course of two months, they're going to dramatically change any of the major signals of impact to marketing to humans. So there's a little bit of this, hey, some of the attributes of traditional advertising still apply, but a lot of the attributes are very new because agents just are not people

and they shouldn't be treated as such. Yeah, I mean, one thing that I'm hearing from this is I'd like to see how this changes, how we teach advertising, right? What new roles are going to come of this out of colleges and universities? Well, I guess right now it would be a combination of folks at Time and Mobian who would have to teach those courses. We're the only ones. But there are more publishers that are signing up for it. And almost every major brand that we speak to is asking for other publishers because they would like to take a broader approach and have more of those touch points, more surface space to be able to do this kind of thing. Yeah. Yeah. I mean, we'll have to see how it plays out, right? Nobody knows. So this has been such an incredible discussion. Thank you so much. I always wrap up my podcast by asking the guests the same questions, and some get very philosophical. Some really reflect on their industry. Others reflect personally. So there's no wrong answer, which is why I love this question.

As a final note, what do you believe the future of AI is in general and its application to organizations and beyond? tough tough time at this exact moment in time um in terms of where that goes with with a lot of i think the the concerns that have rightfully been surfaced about the the signals we're hearing in the in the message warnings that are coming out of the frontier labs um it seems like the more that that conversation is happening the more i think we should all be conscious of that and and certainly concerned. I feel like it wouldn't be one of those things that's just not going to be, I don't think we're going to move on from that quickly. I think the careful application of the development of these models and what they ultimately mean to society and humanity, which is now being called into question, which is really frightening. I think it's really easy to

see every organization is going to be at a different point in terms of how they've deployed AI and whether they have any measurable ROI or just know that they need to be learning new muscles and deploying it. That was sort of how we took the approach. We started rolling out the ability to build agents first with the support of folks on our tech team. And now we have tools where through natural language, you can build them. You can stand them up pretty quickly. We've actually got more agents than employees at the company. I think that's probably pretty common because most people are going to have several at this point doing all kinds. Some of them are remedial tasks. Some of them are more sophisticated. That's just natural. But it was more important just to get everybody to touch it and to get some experience and to realize that this isn't about job replacement or displacement or elimination. This is about how do we train people, skill them,

how do we give them opportunity at their pace currently to be able to learn how to embrace this technology responsibly within their job and be able to do some interesting things. And we've certainly seen some incredible examples already today. So from a professional standpoint, I definitely am very excited about what this is going to mean in terms of the pace of how we can operate, but more so what it means for things to be possible, whereas so much was impossible because of resource constraint or just no way to actually be able to build new products, features, experiences, workflows, whatever it may be. That's all possible now. Everything is possible. And so that is tremendously exciting to me. And on the personal agent side, I think everybody is going to be thrilled to have an agent out there working on their behalf.

He knows their interests, knows what they like to purchase, whether they're in experiences or products or whatever it may be, and all the logistics that we all handle. And I really, really hope that it's going to enable us to do the in-real-life experiences that make people all the happiest, whatever those may be, and take a lot of the remedial tasks off the plate. So I just see a couple of parallel paths. One, watching all of the things that are happening at the more macro level with great concern and fear. And one of the things that we do at Time is we've got a crack team of reporters who are covering it. And we are covering it very objectively. And some of our features have been really powerful in terms of holding those leaders in the industry to account. Professionally, super optimistic about what this is going to enable across the organization and all businesses. And then personally, I'm really excited just to have that additional support as a parent of two kids who their activities themselves could be a full-time job,

just even doing signups and drop offs and pickups and even remembering where and when to be all the time and all the school details and everything that comes with that and trying to maintain of course your own your own hobbies and interests in personal life it's a lot but it feels it feels genuinely like the help is on the way and we're all going to harness these tools and and i do see and and i'm totally optimistic about what that's going to enable for people yeah i love how you ended on a positive note. I was like, oh, no, where's this going? But hey, I love every answer and it always varies. But I like how you also ended on a personal note because I think that that's so important. I have two kids as well and their schedules are crazy. Now they're in two different schools. They play on, you know, multiple different sports teams, all the activities. And so now I just have my agent go in and update the calendar. And that's been a game changer. I also travel a lot like you do, right? And in a number of different time zones. Now, I mean, it's good and it's bad. I was in India recently and I was like, I don't want to do the math and figure this

out. And so there was an invite that came up and I go, add this to my calendar, Eastern time zone. Right. And it did. And so I loved it. It was so efficient. And then I'm like, okay, I can see how our critical thinking skills are starting to go down the drain. Yeah. Yeah. My kids are in elementary school and I think a lot about that and making sure that that remains a core component of their education and how they learn going forward. Because that's going to be what matters going forward. Because we're all going to have the entire knowledge of humanity in our pocket, in our devices. And so that will hopefully be the key skills. And hopefully there'll be a return to appreciating what the real world does offer and personal relationships and experiences and all the beautiful things that you can go do if you can take so much of that time-consuming work of managing our crazy lives off the plate. Yeah. Well, again, I want to thank you so much for joining me for this really wonderful discussion.

Yeah, it was so great. Thank you for having me. Thanks for listening to AI Today, and thanks to our sponsor, Tempo. Only 23% of enterprises can measure the return on the AI tools they have adopted. Tempo was built to close that distance. They evolve strategic portfolio management for the AI enterprise, so your people and your agents share one view of strategy, execution, and value. Did it work? Find out at Tempo.io slash PMI. Tempo. Ambition delivered. Thanks for listening to this episode of AI Today. Please make sure to rate this podcast on Apple Podcasts, Spotify, or your favorite podcast platform. and if you want to lead both the AI transformation of project management and the project management of AI transformations check out resources from the Project Management Institute including the CPMAI certification our various e-learning courses and the PMI by AI blog until you join us next time keep innovating