Rise of the Machines: Is the Media Buying FTE Model DOA

Watch on Youtube

Listen on Spotify

In this episode of Signal & Noise, Brett, Rio, and special guest Shiv Gupta (Founder & CEO of U of Digital) dive into the seismic impact AI is having on the advertising industry.

Listen as they explore Meta’s bold push toward fully automated media buying, question the future of the agency FTE model, and examine how AI could upend traditional workflows while opening the door for new roles, business models, and human-centered strategy.

---

Read the full transcript bellow:

Brett (00:00)

Hey everybody. Welcome to Signal and Noise.

Glad that you've joined us for episode number four, Rise of the Machines, is the media buying FTE model, DOA, dead on arrival. We'll be talking about the impact of AI on the agency world ⁓ with Shiv Gupta, the founder and CEO of U of Digital. And for those of you who don't know who Shiv is, he spent two decades inside AOL and Criteo.

He was a VP level sales leader at Critio before launching U of Digital in 2018. And since U of Digital was founded, Shiv and his team have trained thousands of marketing professionals at companies ranging from Google, Comcast, Disney, helping them decide everything, decode, excuse me, everything from clean rooms to AI driven media buying. He's an incredible guy, very funny, and a subject matter expert on all things advanced ad tech, martech, and AI. And so we really wanted to get his unique vantage point ⁓ kind of on the intersection of education, technology, talent, ⁓ to help really you guys guide ⁓ your thought process and decision making when it comes to navigating really what's becoming an AI automated future of advertising.

Rio (01:20)

This is a great discussion. I agree. I mean, I've known Shiv for a long time, really knowledgeable guy. I love the fact that he, in his role, he gets to work with so many different organizations, both buy side, sell side, brand side, you name it. So really good experience. And it's a cool topic. There's a lot of news items going on that I think make this really relevant, especially some of the recent turmoil in agency land, layoffs. We really felt focusing on education could be very important now. So it's a very timely discussion.

Some previous episodes. Stephanie Laser, Revenge of the Publishers, the great replatforming. She's, as many of you know, an expert at PubTech and on publisher monetization. Great discussion. Another one is with Ben Wild. He's a head of innovation at Georgian and they're a prominent VC firm that invests in SaaS and at Martech and AdTech. And we talk about agentic AI and a feature of the whole SaaS business. Brett, I thought that was a great discussion. And then last week,

Brett (02:12)

Absolutely.

Rio (02:14)

And then last week with Dr. Augustine Fou the ad fraud hunter, that was a great one. Got a lot of feedback, seen a lot of follow up both on X and on LinkedIn about that one. Really cool topic about ad fraud called, Are the bad guys winning? So this discussion, I think it's a good one too. It's no letdown. So I'm really stoked about this.

Brett (02:35)

And speaking about the agency space, I mean, there's been a ton of news and I've been trying to put my finger on the sort of current ⁓ impact of AI on jobs, which is something we talk about with Shiv and where is it going to have a transformative impact, a disruptive impact, a negative impact in some cases on humans' jobs, right? And one of those articles that stood out was the WPP News. I think AdAge put an article out called

WPP Sinks to Lowest Since 2009 on Bleak Outlook for Ad Spending And there was a whole piece in that, a whole paragraph in that article talking about one, economic uncertainty, but two, the impact of AI. And a lot of these agencies are plowing millions, if not tens of millions of dollars into building out tech stacks and building out agentic AI capabilities.

And so one of the questions I think you and I both asked was, is this really a sign of the impact of AI taking certain jobs and realizing that that model of full-time employees may not be sustainable? Or is it just the ebb and flow ⁓ of the agency ecosystem? I mean, we saw both that Publicist ⁓ and ⁓ Omnicom in their ⁓ earnings reports from Q2.

Both have very rosy outlooks with publicists stealing a bunch of very big clients from or conquesting we'll say from ⁓ WPP Paramount, Coke, Mars, they had a 5.96 % organic growth rate. so to me, if I had to conclude whether AIs were seeing it in the economics of these agencies, ⁓ I would say not yet if I had to prognosticate, but I think this episode,

Rio (04:10)

So big wigs

Brett (04:25)

certainly paints a picture of it's going to have an impact in the very near future considering how fast a lot of these changes are happening.

Rio (04:32)

I mean, I think we're starting to see some of the impacts. We talk about it in the session with Shiv. all the uncertainty we're seeing, whether it's tariffs or interest rates or, I mean, there's a lot of stuff going on. So there's definitely an impact there. And then you look at what's the perception of the industry, right? The perception is Publicist has made some great moves. They've got some great technology. They've got great...audience and data that they're using to power their business and they're winning a lot of deals. The perception is WPP is making some interesting moves, but maybe he's been struggling a little bit. then, you know, the Omnicom IPG, the perception is that, you know, it's a massive merger, right? That it's going to be some distraction because of that. So, I don't know. So we talk about this. And interestingly enough, I just heard this a couple of days ago.

Maybe some of the big change in agency land is not over. I mean, I heard, I don't know if it's true or not, but apparently Accenture has been talking to WPP. I don't know how serious it is about some kind of acquisition merger. I mean Accenture is 16 billion. So it would be two massive companies coming together, two very different companies, right? Accenture is a consultancy. I worked in management consulting for almost 15 years, right? They operate very differently. There's some similarities. Obviously Accenture song was kind of becoming a, kind of like a hold co.

But they still operate differently. So that'll be interesting to follow in the coming months to see if there's anything there.

Brett (05:47)

Mm-hmm. Yeah, and it does seem like an interesting in natural alignment. I mean, I think in the episode we talked a little bit about ⁓ agencies moving up the value chain as some of this sort of the grunt work gets either automated or potentially outsourced. And they're offering higher value sort of consultative inputs to their clients, whether it's creative, media mixed strategy or whatever. It seems to me like that's a natural blend and potentially a blend that we can see across the industry.

Rio (06:20)

Well, you think about it, if they need to get better AI, which you mentioned earlier, they need to get better tech, they have to have better technical chops and better and more technical resources. Accenture would give them that in spades. I mean, they're basically, I mean, you could argue these big SIs are consultancies or really tech companies. Some other news items I think worth mentioning, MediaOcean announces H2 2025 market report. And I think not surprisingly, they show the 58 % of marketers plan to raise CTV budgets. I mean, that's where all of the incremental spend is going. CTV used to be the hottest spot for sure. And they also ranked, surprise, surprise AI as well as identity is the other top priorities in CTV. Related to that, there is an announcement with LG integrating with Viant. Viant continues to, I think, make some interesting moves, Brett And I the premise here being that...

Brett (06:50)

Yep.

Rio (07:10)

programmatic identity-based access to the 45 million plus LG Smart TVs in the US is kind of interesting, right? And this continues kind of the story of OEM data, right, from the manufacturers plus household IDs. Keep in mind that CTV doesn't have cookies, right? So it's targeting a little differently. Is this becoming kind of the new currency of CTV buying, really, using this deterministic data, So I think that's interesting.

Brett (07:34)

And it just shows you how important the television glass remains, however the content is delivered and the advertising is delivered, which is obviously programmatic or digitally delivered in these new ecosystems. The television glass, that sort of sound, sight, and motion of a giant screen in your family room continues to be a dominant engagement medium. I don't see why that's, it seems to me like a natural trend for sure.

Rio (07:59)

100%. You remember last year, Walmart acquired Visio for lot of the same reasons, right? So I think it's very interesting. There was the CloudFlare announcing its quote unquote paper crawl, right? And this is a feature to let publishers charge or block AI crawlers. then on one hand, this reflects, I think the urgency, we talked about this in the Steph Layser one, right?

Brett (08:02)

Yep. Yep.

Rio (08:22)

The urgency they're feeling to monetize their content as AI search starts to really, I mean, we're seeing already a steep drop off in referred search traffic to publishers from Google as AI searches take over, right? I think that's so, what to do about it. Now here comes CloudFlare, right? And in CloudFlare Brett, they're one these companies that if you don't work in tech, you probably never heard of them, right? In fact, I think many people in tech probably haven't heard of them, depending on where they are, that they are a content delivery network or CDN.

They're really a critical player in what's called internet plumbing. And they pretty much sit in front of around 20 % of the web, right? And they can reroute, secure, accelerate traffic. When there's a big distributed denial of service, DDoS attack, they tend to deal with that by routing traffic in different places.

Brett (09:10)

Yeah, definitely in the security realm for sure, right, in terms of protecting site security. How do you monetize a paper crawl? I mean, think that's that, like, how do you actually price that? What is the value of a crawl is to me the hardest question to answer. And it'll be interesting to see what comes of that, if anything comes of that.

Rio (09:15)

Yep. Yeah. Well, I mean, guess publishers were just sick of, like we have this binary choice now, Either we let them all crawl and scrape our website for free, right? And we at least get in the AI model or inform it, or we block them all. And we're not in the AI models, So that's kind of been the choice up to present. I mean, I think it's interesting to Cloudflare. And funny, they call it Content Independence Day, this announcement. And that's...I mean, look, it's a good idea. needs to be done. I don't think anyone knows. There has to be some kind of micropayment strategy, potentially.

Brett (10:05)

Yeah.

Rio (10:06)

And I think that, you what's interesting for CloudFlare, they kind of become the merchant of record here, If this actually takes off, they're the kind of referee, the one handling the metering, invoicing, payout, You know, I'm assuming there'd be a take rate for them. So this is, if this works, it would be great for CloudFlare. If it works, it'd probably be good for pubs as well. But, you know, I think this also begs the question of like, are we, is this new model emerging finally? Like as kind of the old model goes away and, you know.

And I mean, they framed it as this is the first step, and I'm going use quotes here, in building an agentic paywall, Where intelligent agents handle these things, negotiating budget, licensing, all this. As publishers figure a way to let these models use their content to build better models and inform it. So, I don't know, we'll see what happens. It's an interesting development. I thought it was worth covering here. And then I guess the last news item.

Related to this, but I think also related to the discussion today in a big way. Chat GPT had their big announcement yesterday where they launched Chat GPT agent. I found this is fascinating Brett, because I mean, you look even back three, four months, I was listening to a lot of AI podcasts and people say, this agent second actually go book an airline ticket, do very sophisticated tests that require thinking feedback, working with other agents. That's going to happen, but it's several years away. Well, it's not several years away.

During the session yesterday, they had this Chat GPT agent planning a wedding. you can actually ask it other things while it'll do it in the background. it was planning a wedding and they asked it to make a sticker. And it did that. I mean, they even said that they'd seen three or four hundred percent advancements in just a few months. So AI is accelerating, I think, quicker than anyone, including the people building these tools, had expected, which is really, really amazing and a little scary too,

Brett (12:00)

I was just talking to our producer. I just introduced a Chat GPT team, which is secure within our organization, but within my team. And there's a bunch of agent type capabilities to build PowerPoints, to build spreadsheets, to build Word docs or Google docs. so...we're seeing that it's still not perfect, but it does give you scalability and speed to least speed to draft. That's what I find. And I think if you talk to any creative leader, yeah, speed to draft, speed to an initial either idea, ⁓ outline, version of a creative, and then it takes human intervention to get it to ⁓ something that's publishable, right? Whatever the content type is. And I would assume that that type of conclusion probably applies for a lot of these agents. It's going to build me a wedding plan. Yeah, it's going to build me a draft for a wedding plan. And then I'm have to go in and say, no, this is wrong. Tweak this, tune that. But the draft oftentimes is the thing that takes the longest amount of time. So it adds value and speed to market, no matter what you're doing with it, is what we've seen as a team. So certainly interesting.

Rio (13:12)

Totally. then, and you look at the announcement yesterday, I think they were saying they're different plans, I mean, available at different tiers in ChatGPT but I think the pro plan gets 400 queries per month with ChatGPT agent. And I mean, my opinion, this is a big upgrade.

You look at Chat GPT, it's gotten better faster. This, I think, kind of completes it before you. You had deep research. You combine that with the operator, which is their like, Agentic kind of web browser. You look at these three together, it's very powerful. You can start to do some very sophisticated things. ⁓ so I think stay tuned.

Brett (13:40)

Yeah. And the Microsoft Office suite, add that, which I just mentioned, you add that to the mix and suddenly they're competing with a whole bunch of different companies. interesting.

Rio (13:51)

So interesting times to say the least.

Brett (13:54)

All right, gonna welcome Shiv Gupta to the stage. Stay tuned.

Brett (14:00)

Hey, welcome Shiv to Signal & Noise.

Shiv (14:02)

Thanks for having me.

Rio (14:02)

It's going to be a fun one. This is one we've been waiting for for a while. Shiv, really glad you could join us here. greatly appreciated. We've known each other for a while. And I was thinking of what would be a really cool topic we could talk about. And I think we landed on one. The topic here is rise of the machines. And the T2 reference is totally on purpose. Is media buying the FTE model that agencies have deployed for decades?

Is it dead? it DOA? I guess the of this was some of the Mark Zuckerberg announcements that we heard. And we covered this last episode a little bit, Brett, where we talked about how Zuck was saying, brands are just going to log into Meta. They're going to push build my campaigns for me and maybe they have uploaded a product image, maybe they don't, everything's gonna be automated. So it's a really bold, really audacious vision. He's continued to talk about this. And he recently went in that binge of hiring AI talent. I think that's probably part of the reason he's doing that, right? Is his vision of everything being automated. So it's a, let's call it his latest push in the AI automation race. It's a big, it's bold, it's audacious. And I think for the industry, especially agencies and hold cos it's potentially seismic, right? So he said he expects for those who missed it really all media buying to be fully automated. He didn't say 2030, he said next year is going to happen, which is look at the calendar quite literally just months from now. And if this super aggressive

Brett (15:21)

And it's a big step forward from what they've gotten in place right now, which is more like ad variation.

Rio (15:25)

Totally. mean, you compare it to like going to DSPs today. It's like really clunky buttons, controls. I mean, I saw the great, interview with Adam Epstein, right, from Gigi, how they talking about taking what's a really laborious, time-consuming manual task and streamline using AI, right? So, but, you know, going back to Zuck.

If this aggressive timeline holds, it really would signal a complete wholesale reinvention of how advertising works today and how it's worked for decades, right? And this will mean that algorithms powered by AI, not humans, will determine how billions of dollars of ad spend, where it's going, how it's being used. This would upend decades of, let's say, inertia and ways of doing business across agencies and marketing teams. This ambition extends far beyond media buying too.

What Zuck outlined was systems with fully automatic creative production, writing, copy, assets, even generating entire campaigns. It affect pacing. It would even affect the analytics. Super bold. Totally, right? And then for brands, you think about it, super compelling. Wow, like we won't, we can have fewer people do this, right? Fewer manual errors, faster optimizations, and definitely reduce costs. But for agencies, you think about it, especially holding companies that have for decades built their clients based on FTE model.

Brett (16:19)

Truly end to end.

Rio (16:37)

The implications are potentially existential, right? You look at WPP's recent stock plunge, like, so wow, this is great timing. Not that it's great that the stock plunged or they've had a bunch of layoffs, but what's happening here, right? Is it just because of AI, the stock hit new lows, you know, not seen since 2009. It's crazy, that was during the crisis, right? This underscores how vulnerable these big companies are, how vulnerable the traditional agency model potentially is this new era and as AI really reshapes the value chain end to end and the whole marketing workforce is forced to respond to this like let's face it they're bracing for disruption potentially greater than any one ushered in like certainly since programmatic right came in in came in more than a decade ago it is not just about tools this is about the future of work in marketing the future of work in advertising who's needed what they'll do

What will the machines do, and will the machines just take over? And what does this do to the traditional agency hold co-model? So really, really important. And help us impact this, what this shift means, who's going to change, who's it gonna impact, and what should they do about it? We're to talk to Shiv. Shiv, welcome. Great to have you here, founder of U of Digital.

Shiv (17:47)

Thanks, guys. Yeah, thanks for having me. I'm thrilled to be here, thrilled to talk about this. I have so many thoughts. I have so many things that I want to hear. I want to hear your thoughts. And yeah, this is going to be an awesome conversation. Let's do it.

Brett (17:59)

Yeah, and let's, for those that don't know Shiv Gupta, ⁓ he is the founder and CEO of U of Digital. You spent two decades at AOL and Criteo combined in sales commercial roles before launching, I think it was 2018 that you launched U of Digital. ⁓ And John Hopkins, I've got an 18 year old going to school. ain't anything to, ⁓ that's impressive. You must've worked really hard in high school.

Shiv (18:27)

⁓ Well, so quick correction Brett, sorry to start on this note. It's Johns Hopkins with an S Johns, okay, can't no you said John that's a common it's a common mistake ⁓ and I was a tour guide at Johns Hopkins and so I know this Yeah, no, I know this and the reason it's called Johns Hopkins cuz like what kind of name is Johns It's because gal named Margaret Johns married Gerard Hopkins

Brett (18:32)

Johns! That's what, yes, yes. Isn't that what I said? I wrote it. I wrote it right. I spoke it wrong.

Rio (18:38)

Yeah. He corrected people all the time, I imagine, right?

Shiv (18:57)

where Johns Hopkins came from.

Rio (18:58)

I was going to ask were there

Brett (18:59)

You caught that, you caught that. I swear to God, I wrote it the right way. So tell us, yeah, tell us, I know he's like, like, he's like, I totally fell into the trap. So tell us a little bit about your sort of trajectory through the U of Digital.

Rio (19:02)

Fred, he was waiting. He was waiting for that. He fell into the trap.

Shiv (19:12)

Yeah, absolutely. So yeah, I spent a bunch of time at great companies in the space that molded who I am, molded my career, AOL. I actually started at advertising.com, which was early pioneer in ad tech and programmatic. We were doing programmatic before it was even a term. And that was, I learned everything there. I learned the nuts and bolts of how the whole thing works. And it was an awesome first place to work. There's a whole ad.com mafia in our industry. It's awesome. ⁓ that, know, yeah, there's, you know, all, they're all the different mafias. There's the AOL mafia, there's the Google mafia, there's the Yahoo mafia. So anyways, ⁓ I was part of that AOL tree, the ad.com tree. I was there for 10 years, mostly in sales roles, sales leadership type of roles. I was at Critio for a couple of years running their ⁓ sales organization for part of the United States. And then I started U of Digital in 2018.

Brett (19:42)

Yeah, it's like the double click mafia. Yeah.

Shiv (20:06)

And our mission, our vision as a company is very simple. ⁓ We wanna lift all boats, right? Through education, through knowledge. Like I'm sure you guys have experienced this. Everyone has experienced this. Our industry has so much information, ⁓ like asymmetry, right? Like you get companies at the table that are very similar, right? Let's say SSP and DSP. You get people from those companies at a table.

Brett (20:27)

Yeah.

Shiv (20:34)

and they don't even talk the same language. They don't understand each other. And those are like the two most similar types of companies you can have in our industry. So imagine getting people at a table from a DSP and a CDP. like one, one side's talking Japanese, the other side is talking French, right? So like there's so much inefficiency in our industry because of information asymmetry. And that is our goal, right? We want to try to make the industry better first and foremost by helping people get to the same level playing field in terms of knowledge.⁓ And then, you we want to help companies achieve better outcomes by making their team smarter. And we want to help individuals drive better outcomes for their career, right? So that is our hypothesis. That is our vision. That's what we're trying to do. We've been doing it now for almost eight years. And it's just been really fun. It's really valuable. ⁓ You know, we've seen, we create a lot of value and it's really rewarding, right? Like for me, I started it because I love seeing light bulbs go off. I love being in front of an audience and trying to explain something and seeing people get it, that is the most rewarding thing to me. And so that's why I started the company. That's why we do what we do. ⁓ And yeah, that's just a little bit about U of Digital.

Brett (21:43)

Yeah, I thought what you, that point you just made about ⁓ communication ⁓ asymmetry within the industry. Imagine when you start to talk to people outside the industry, and we deal with this all the time, when we're talking to brand advertisers, and I'm like, that's a whole nother level of translation. And oftentimes, there's a lost in translation aspect if you're too wonky.

Shiv (21:59)

100%.

Rio (22:03)

I talked to a brand marker who didn't know what a DSP was the other day. I mean, I asked him, he's like, what's that? And I guess at first I was like, you know, this guy must be, maybe he's not very good at his job, but he actually is good at his job. And I realized, wow, he just has not been involved in the media side at all. So.

Shiv (22:07)

yeah.

Brett (22:19)

Don't address me. I knew an executive of my last org that didn't when I said linear television it threw him for a he was completely confused what that was linear television right and you just take you assume that what we're saying makes sense to the rest of the world and it doesn't so you start with our own industry like get your your communication house in order right

Shiv (22:38)

Yeah.

Brett (22:39)

back to the topic, the theme on hand. So thanks for introducing yourself, Shiv, and you're doing some great work in the industry. It's awesome to see, and you're super passionate about it which is great. You love what you do, which is, you know, one of the reasons why a lot of us are in ad tech and MarTech and related because we've, you know, we're like, there's no better place to be in a way. ⁓ So, so Rio referenced this article about meta. was a Wall Street Journal article with Meghan Bobrowsky and Patrick Coffee where they talked about what really got me about the meta announcement was that it was so far beyond what they're currently doing within their ad platform.

Right. Which is basic ad variation. I mean, it's it's the as Rio said, it's end to end from ad creation to media buying to analytics and measurement, which is super disruptive because there's vendors that play in all of those places. ⁓ Do you think this is real? Do you think it's what do you think about the timeline? 2026 is overly ambitious. What are your thoughts on it?

Rio (23:35)

is realistic.

Shiv (23:40)

Well, I think ⁓ the MVP of it is kind of here today, right? You could argue, I know you guys have kind of said a couple of times like, it's far away. You could argue the MVP is here today and available today, right? Like you go into some of these tools like PMAX and stuff and they do a lot of this stuff already. ⁓ You could argue that the creative stuff is still a little bit being worked on, it's in progress. But the MVP of, hey, just put in your budget and your goals and the thing will do all the rest of the work.

It's not, I don't see why that's actually far away. A good version of it may take a bit more time. A version that provides marketers with the right level of transparency, creates trust, gives marketers the right tools to kind of have oversight. I think that may take more time, but I think what he is saying, like, you know, he's Mark Zuckerberg. All the CEOs of the big tech companies are great talkers and Mark, maybe Mark is not the best a talker and marketer, but he, you know, he's not bad.

Brett (24:36)

Hahaha!

Rio (24:38)

He likes to talk. He had a six hour podcast apparently the other day.

Shiv (24:42)

really? Was it with Joe Rogan?

Rio (24:44)

Yeah, didn't. Joe Rogan only has two or three hours. This is 2X that. Yeah, I didn't listen to it. But apparently it was six hours. think people listen to it on like double speed or something like that. But yeah, it's a lot.

Shiv (24:55)

Did he wear a chain? ⁓

Brett (24:55)

So you're

Rio (24:56)

Yeah ⁓

Brett (24:58)

saying Shiv ⁓ that it's here today, that they've got to go to market plan around this product launch basically, that it's not an MVP, that it's actually relatively close to being ready to roll. That's what it suggests to me.

Shiv (25:11)

I think he can sit there and incredibly say that. Again, it may not be perfect, it may not be amazing, it may not drive the best outcomes, there may be lots of questions about it, trust may be an issue, it may make mistakes, all of those things, but he can sit there and incredibly say, we're gonna launch this thing in 2026. I think that's very real. If you compartmentalize all of those things, planning, creative, audience discovery, execution, optimization, measurement, analytics,

Brett (25:14)

Yeah.

Shiv (25:39)

Today, you can go out and find solid AI tools that are maybe point solutions that solve each of those problems individually. So why can't Meta, with all of the money and the market cap and engineers that they have, stitch those types of point solutions together or build them in-house and do it by 2026? I believe it's totally viable.

Rio (25:58)

I mean, I agree it's going to happen. I think the question is the timeline and I think it's going to happen quicker than many people suspect, right? mean, there was that old, mean, there's a great quote from Bill Gates about how people always overestimate what will happen in a year and they underestimate what will happen in a decade in terms of change. I think there's going to be a little of that, right? But I've never seen anything improve and change so quickly as I have with AI. I mean, was kind of like, I don't say skeptic, but I had my doubts a couple of years ago because it was just not that great, but...

I'm using it all the time now. It's like the, it's incredible. Yeah. And like we're old guys, right? So it's happening so fast and they're getting better and like these tools are like, it's, and I imagine the future is going to be this like, like synthesis of like robotics, what's happening, and AI, like it's going to be like, it's going to change a lot of stuff really quickly, which is I think pretty cool. you know, I guess counterpoint to that is, or impact of that. I had this, listen to this other podcast and you're talking about change over time and look back at.

Brett (26:28)

Yeah, it's like look at our own behavior. How much that's changed.

Rio (26:55)

when computers came out, people were predicting apparently that sonography as a profession was going to get absolutely obliterated by computers. And you know what, it did happen, right? But what they didn't predict was what got created, all these graphic design jobs, among many, many, many other jobs, right? That there are tens of thousands or millions of people doing today making great livings, right? So it's easy to predict what jobs will get wiped out. It's hard to predict what jobs will get created. I think that's probably true here. But looking at, I guess the risk here for these agencies and holdcos is,

Yeah, sure, things will be created. Maybe they'll be the ones doing that, but it takes time to do that and stuff's happening so quickly. So I guess looking at the question at hand here Shiv as AI begins to take these things off, whether it's in six months or a year, it's going to happen, right? Things like bid management, budget pacing, creative rotation, creative generation, right? These are things agencies bill for today. So which agency roles, in your opinion, are most vulnerable to contraction? What new roles? I mean, I know it's hard to look ahead. Could you potentially see emerging?

And how does this impact the agencies and holdcos?

Shiv (27:56)

Yeah, I mean, I'm going to probably just say some really obvious stuff to start, and then let's get into the nooks and crannies and the second and third degree stuff, because that's the interesting stuff, right? So the obvious stuff is any rote work that requires little human creativity ⁓ is going to be automated, right? So the most basic example of that is campaign trafficking, right? Like trafficking things. Any job that requires a human to go into a UI and push buttons as their primary task is going to get AI-ified very, very quickly. And so you can quickly extrapolate that to tons and tons of different roles across the agency. Now, where I think there is more future-proofness is planning. I think planning and thinking about strategy, those roles are going to... And this is how...

Brett (28:52)

Yeah.

Shiv (28:54)

things always work, right? If there's a lot of things in the mix and certain things become obsolete, the other things that are still remaining become more valuable and you throw more people at it and you pay them more, right? And so like, I think strategy, planning, I think even creative, right? Yes, we're talking about automating creative and AI will do a lot in terms of creative, but humans will still be extremely valuable when it comes to creative and the strategy behind creative and messaging, right?

Brett (29:06)

Yep.

Shiv (29:23)

So I think all those types of roles get more valuable. You throw more bodies at that. You pay them more as the rote kind of ⁓ push a bunch of buttons in a UI stuff goes by the wayside. That's the obvious stuff.

Brett (29:35)

And I think it's so how do you how do you ensure that what about all the people that are employed that are doing the road stuff? I think that's the crux of the problem here.

Rio (29:42)

Because that's a lot of the work today. And I'll give an example too. So like I recently read it like, so I remember Pfizer announced, I think Plow was going to be in housing. I mean, I think they announced a couple of weeks ago they're in housing all of their social, right? Which is a lot. I think they're spending about a billion dollars a year total in Pfizer for all paid media, is a lot, know, Pharma spend a lot. And I remember about a year ago they were doing this RFPs for support with their in-housing initiative. You know, they hired him and he's...in-house and I imagine, I haven't talked to about it, but I imagine a lot of that, the drive to do this was because they could do it now with AI. Why didn't they do it five years ago, right? I'm sure AI is helping them with that process, right? I again haven't talked to him, but I'd be shocked if that weren't the case. So if I remember there was over 200 people at OMG supporting that account, I think it's 225 or something, that's a huge number of people.

I don't know how many will be impacted because of this, it's got to be some. So that's one example, Shiv I love your thoughts on that.

Shiv (30:40)

Yeah, I mean, that's a great example. ⁓ I was talking to somebody yesterday who was a CMO at a huge ad tech company. ⁓ And basically that person told me two years ago, not even like six months ago, they basically were like, everyone's got to cut headcount by 20%. We don't care how the work is going to get reallocated. This is happening. We're going to just cut headcount first, and then we're going to figure out like how to solve the headcount loss problem.

Right? And so the cold hard truth, the emotionless truth is people are losing their jobs. People are going to continue to lose their jobs. Unemployment is going to go up, but this is how every, this is how every revolution in our history plays out. Right? Industrial Revolution, .com. it's always this happens, right? There is pain and there are short-term pain, but the hope is, right? First of all, the hope is that at least some employers have, have

Brett (31:25)

Industrial Revolution, the cycles.

Shiv (31:40)

a heart and figure out how to upskill their workforce and reallocate them to work that is useful and valuable. ⁓ The hope is that people take initiative to go out and retrofit their own skill sets. And the hope is, actually, I think this is very promising, because AI moves so fast, we just talked about this a bit ago, because it moves so fast, can we actually decrease the timeline of the pain?⁓ in a way right versus the industry industrial revolution or versus you know the the internet revolution whatever

Brett (32:13)

yeah. And you wonder if these problems become so large that can you do it at the company level or do you need organizations as large as governments? I'm thinking about when Margaret Thatcher sort of shut down coal mining in England, right? And it affected a lot of people's lives who had no skills to do anything other than dig out coal, right? So we're dealing with something like that that's gonna have such seismic shifts.

Shiv (32:37)

Totally.

Brett (32:37)

I mean, how do you think about this from an education perspective? Because to me, it's like you being a psychiatrist in today's world, right? You're always going to have a job. You're always going to have somebody to teach or train somebody on.

Shiv (32:51)

Well, yeah, I think 100%. And we are a training company. And so we're thinking a lot about how we help folks retrofit their skills around AI. We're doing a lot of AI training, as you can imagine. We're trying to give people tools, AI tools, to learn through AI. And so we're doing all those things. I think the thing, sorry, not to get existential for a second, but I think what we're talking about here is part of human evolution. This is micro moment in human evolution. In the grand scheme of the thousands and thousands of years of human existence, this is one moment that is pushing humans forward. anytime there's growth, if you're a teenager and you're growing and you're having a growth spurt, you have some pain, but then you grow. And so this is human evolution. And it's painful in the short term.

But if you think about what I just said about jobs in the agency world, we are going to get rid of jobs that require people to click buttons and do things that don't require creativity and brain power. And we are going to reallocate people's brain power towards things that only humans can do, at least today. And that is good. That is positive. That increases productivity. And this is exciting. know there's obviously a bad part about all this. It's the pain.

It's the part about people losing jobs, but this is generally a good thing. It's a moment of evolution and we wanna enable that. We wanna empower that. As U of Digital we wanna play a small part in helping people learn. I will tell you like a few months ago, we were like, just like every company, we were like, okay, what's our AI strategy? And we're like, okay. Yeah, right. Everyone's doing it. Like what's the AI strategy? There's so many different things you can do with AI. It's exciting.

Brett (34:32)

I'm doing that with my team right now.

Shiv (34:42)

but it's overwhelming. It's like, we could build a chat bot. We could build a personal tutor. We could like have, you know, a multimodal AI thing that, ⁓ you know, teaches you in this way and creates like different personas and coaches. Like we do all these things. And then like a few weeks ago, I was like, you know what guys, our AI strategy, first of all, we need to do two things. We need to think about how to introduce AI into everything we do incrementally to solve incremental problems.

Brett (35:13)

That, yes. You can't boil the ocean with this because it's so capable of doing so many things. You'll never get started if you try to start with a sprint.

Rio (35:19)

Yeah, it's not one thing. It's many things, yeah.

Shiv (35:24)

It's a, you get paralysis, right? I think that's one problem. The other problem with trying to go zero to 60 on this stuff is that it's changing so fast. So you could come up with the coolest, biggest idea today for like what AI could do. And then tomorrow some new model comes out or some like new capability comes out and your big idea that goes from zero to 60 is totally stupid and useless. Right. And so like, that's another reason why I think.

Brett (35:26)

Yeah.

Shiv (35:53)

in order to build AI and bring it to market, you should focus on solving a very specific problem. You should do it incrementally. ⁓ Yes, it's important to be strategic, but I almost think it's as important right now to be tactical about solving problems using AI. And so that's how we're thinking about it right now. We're doing two things. One is incrementally solving problems using small AI kind of tips, tweaks and tricks ⁓ that we can give to learners. And then the other thing we're trying to do is just how do we help people learn about AI? That's the most important thing I think we can do as U of Digital is learn about AI, become AI forward, figure out how to get yourself comfortable in the tools, ⁓ kind of transformation, right? Everyone wants to go or take their teams through that kind of like transformation right now, like get people using it, get people's behaviors to change. That's the most important thing we could be doing right now. And so that's what we're focused on.

Rio (36:46)

It's interesting you bring that up. like looking at the agency model as we talked about earlier is going to be under some stress, right? Because clients are going to say, agencies, you've been telling us about all these great AI announcements, tools you're building, show us the money, show us how you're going to save us 10, 15%. And I've seen clients do it already, tell agencies we're expecting you to come in under cost for next year. I think this puts a lot of pressure on the model, but look, agencies will not go away. forget who said it, but they are kind of like the cockroaches, right? They will evolve, they'll adapt.

It could be nuclear war, agencies will still be here, right? they'll just be different, Yeah, what can they do? I I've heard like, I've heard for Publicists, for example, is already going into like value-based contracts and going away from FTE. So like, does this impact the FTE model? How does this impact? And what are some like tips, I guess, like for like, that you would give to them and the people working there in order to like future-proof themselves?

Brett (37:19)

do they justify their fees is a key theme here.

Shiv (37:37)

Yeah, think 100%, again, this is like fairly obvious, right? FTE model aligns incentives to hours of time spent ⁓ and it's not aligned to outcomes. Yeah, it's billable hours. And like, let's be honest, AI or not, that is a flawed incentive model. And that's why there's always been some level of friction between agencies and end clients because agencies are incentivized to bill for hours and not produce...

Brett (37:49)

Yep, billable hours. Yep.

Shiv (38:06)

great content or outcomes necessarily, right? And so, you I don't think AI changes the need for them to change the incentive structure. I think AI accelerates the...

Brett (38:19)

The need, yeah, accelerates that need to change this incentive structure to align to this new model.

Rio (38:24)

Good point.

Shiv (38:26)

Yeah, so it already had to happen. AI now is just like a gun to their head. Sorry, that's terrible, terrible. But you get what I'm saying? Yeah, right? And it's something that the marketers, the brands can point to and say, okay, well in the past you kind of dragged me along because this is how work had to be done, but it doesn't have to be done this way anymore. So we either put up or shut up, right? And so, yeah.

Brett (38:28)

Yep.

Rio (38:32)

That's good analogy. It's not wrong, right?

Brett (38:34)

Yeah. What it? Yeah. And you're hitting on a super important theme, Which is AI might be a red herring in this whole conversation, right? And I think a good analogy, it's like blaming the guy that kicked over the lantern or the lantern for causing the great Chicago fire of 1870, 71 to the cow. The cow kicked the lantern over, right? It's not the lantern's fault. It was bad construction. It was overcrowding. was drought, like an unprecedented drought, high winds. The fire department got overwhelmed and the city burned to the ground, right? ⁓ Right.

Rio (39:03)

The cow did it,

Brett (39:17)

So I do think to a degree AI is a red herring. And I think there's a couple of key themes that have been driving the change in this sort of agency market or construct, I'd argue. know, one, a lot of advertising is moving in-house. I think what I've read is about 80 % of brands have moved at least a certain percentage of their, think of the entire kind of end-to-end work stream in-house, right? So they want that.

Rio (39:44)

Yeah, I don't see many of them in housing actual media buying. I Pfizer with social maybe being an exception, but you're right, Brett. I mean, that is a trend. It's been going on for over a decade of in-housing certain activities that once were really the purview of agencies,

Brett (39:47)

Yeah. Yeah, and they want a bit more control, right? And programmatic media, what do you need to pay? You know, we talked about this with Dr. Fu, right? Why do you have to pay this ⁓ extra amount per CPM to do something that you can do in-house very easily from an auction perspective? know, short-termism, I mean, everybody's heard this theme, CFOs, CMOs, under a ton of pressure, quarterly results. It's greater than I think at any time, probably in modern capitalism, right? And then,

Rio (40:10)

auction.

Brett (40:23)

The third one that we have all heard a lot of, CMOs are fighting for their jobs, right? 40 % of Fortune 500 companies don't have a CMO at the executive table, which is, that was something I read recently, which I thought was striking. I don't think it's surprising. But I mean, how do you think about these kind of overarching themes really driving the shift, the paradigm shift between the agency marketer relationship?

Shiv (40:48)

Yeah, mean, think in housing is interesting, right? Cause I feel like it had a moment five years ago and then, you know, like the agencies kind of wrangled it all back. And now maybe AI is, yeah, they always, you know, to your point, they are cockroach, they're Twinkie, whatever, whatever we're calling them. They they survive. And I think AI is maybe another catalyst and impetus for the in housing trend to kind of pick up again. I think another thing that's interesting is like,

Rio (40:58)

They fought back.

Shiv (41:15)

you know, maybe 10, 15 years ago, we were talking about how Google and Facebook are, and platforms in general want to eat up the agency model, right? And, you know, along comes trade desk and trade desk's like, no, we know that they're the cockroaches. We're going to prop them up, right? They're not going anywhere. We're going to prop them up. And that was an interesting kind of like counter to what the big tech platforms are doing. But I think there's an opportunity again, what's coming full circle to how we started this podcast.

What Zuckerberg said is, and there was tons of articles and headlines like, agency executive furious about Zuckerberg's statement. And the reason the agency executive is furious is because that statement is coming after the agencies and trying to eat the agencies alive. So is that on the table now? Again, history shows it's not and agencies will evolve and figure it out but maybe the time has come or maybe there will be some interesting stuff where it's like, well, now a bunch of marketers just go direct to vendor, direct to platform. They don't need the intermediary anymore. The intermediaries haven't evolved fast enough to keep up, you know?

Rio (42:23)

Well, I think them going direct platform, and we did talk about this last week, Brad. I I think that's something we're seeing as an increasing trend just for transparency of nothing else, right? I mean, I think that's something we're going to see more more of, and I do agree, Shiv. There was like this big in-housing push. The agencies fought back, right? They're good at that, right? They did a good job in retaining that. And I even remember I worked with a good friend of mine. She ran Global Media for Philips in the Netherlands. They in-housed everything, like lock stock and bar, including buying everything, built huge teams to do it. And you know what, after two or three years, they just decided juice wasn't worth the squeeze. It was too hard. couldn't, like traditional media buys, especially rest of the world, are just so difficult to do. Like, how do know who to give the money to, right? If you're running terrestrial radio or out of home or lot of the, or print, right? Which is a lot of the world. It's hard to do, especially globally. So they ended up spinning it back out.

Brett (43:13)

or CTV.

Rio (43:19)

and they put out a bid and you know, they're not an agent. And so they in-house a lot of the activities relating to it, but the actual buying itself is now back. So I don't know, it's like you can talk about like the in-housing and I think the general trend is for brands to do more things in-house. plus I think too for marketers is probably better, right? Like, do you really want to be a vendor manager? I mean, maybe that's part of your job, but you should also be doing strategy to your point Shiv. You should be doing like actually checking the homework of the agency. You should be hiring third parties to do audits and things like that. So I think it actually, the right balance, there's a better balance than maybe we've seen, but I agencies have a spot, right? I don't see them going away and I think what they would do would be important, but it'll change, right? So it's sort of like, how do you see that? Like, what are some things that you're seeing them do or do you, like, what would you recommend they do?

Shiv (44:04)

Well, so I think they're doing some of the right things, right? So like the case that you just outlined Rio, the reason Philips went back to the agency is because ⁓ complexity, right? Too much complexity to manage. ⁓ Probably like economies of scale are not their friend, right? Agencies still bring a lot of the economies of scale value prop. ⁓ And so, you know, certain things AI can fix and justify going, you know, brand direct, like the complexity component, right? ⁓ know, navigating the Loom escape, like that kind of stuff. think AI can address some of those things, but maybe economies of scale AI can address, right? Or, you know, increasingly we're seeing the agencies own tech. Like the brands are not going to do that. And I think that's smart, right? So it's like, okay, well, publicist has really made hay over the last few years because they've started owning the data layer, right? And I think every...

Rio (44:59)

What do think of that though? mean like about them owning tech. think historically they've been really bad at building and tech but is that mean, Brett, what do you think? I see this as changing. Do you think this is the fundamental like permanent change or?

Shiv (45:03)

They've been terrible at it. Yeah.

Brett (45:05)

Yeah. Yeah, I mean, think, I think, cause you're competing with a lot of the other players like the snowflakes, the data bricks, the, the, the, the real data players that know how to operate these sort of super highways and these data lake houses and such, right. Which have, they just have a core competency that I don't think, yeah, you can go out and acquire the talent, but I'm not sure, ⁓ that they have a good starting point. The agencies like in terms of core competency, I don't think that's where it lies necessarily.

Shiv (45:37)

Yeah, I think you'd have to tie human value and agnosticism to tech in order to make it work within the agency construct. I think ⁓ data and analytics, I think you can make an argument that agnosticism and human capital matter when it comes to some of that. ⁓ Because that's the core value prop of an agency, right? It's human capital, it's agnosticism that maybe you can't get directly from a vendor.

And so I think the measurement, analytics, creative, ⁓ and then I do think the AI thing is really interesting. Like owning the AI layer is fascinating for an agency. I think that's a big opportunity. Again, they haven't done it well traditionally. Maybe Publicist is changing that and rewriting the playbook about holding companies actually owning tech effectively, but they're all trying to do it. WPP is trying to do it, IPG is trying everyone's trying to do it that, we own the AI layer. I think in theory, the value prop makes sense, right? Because if I'm a buyer, I don't want my AI being supplied from Google necessarily or Meta, because they're also trying to sell me media, right? And it's the same idea with data and analytics and cloud and all that kind of stuff. ⁓

Brett (46:52)

Yeah.

Shiv (46:58)

And so, okay, well, the agency layer has value there then. Like, can you create a layer of AI across all of the stuff that I'm engaging in, all of my data, all my creative, all my media? I think that's valuable. And so I think that's a playbook, you know?

Rio (47:13)

You can see why they do that, right? Because let's say you're an agency, you want to build a large language model specifically. You're going to have your people building these rag docks. They're going to be uploading them and training this thing on your client's data, on the way you do business, right? So I think that, especially to your point before, Shiv, you can't just let an AI loose on your company and say, fix everything. No, you have to find specific things you want to do better. I want to do trafficking better. I want to do measurement better. I want to do this. then training on like

Brett (47:13)

Yeah.

Rio (47:40)

your data, your processes, the way you do things, and then supervise it and then work with it as it gets better over time. So I think these proprietary LLMs that agencies are beginning to deploy, I think, I can see why that makes sense, but I think the problem they've always had is just the hold-co model, right? Where, like I remember like I was selling to one of the big hold-cos. We were trying to sell them. They had this idea for this solution they wanted to build. And I remember like we gave them a proposal for like half a million.

I remember the CEO said, no way I'm paying. I have 600 AWS certified devs. I can snap my fingers. I said, really? you really get like 20 of them together to start next week and do like a series of sprints for the next 12 weeks to do this for you? He no, I can't. So here's the money. So I think that that's always stood in the way of what you're describing.

Shiv (48:27)

But if the model, but like going back to financial models, right? If the agencies are starting to figure out that their business model can't be tied to human capital and they need to function more like tech companies and their incentive structure and how they pay people and how they charge customers needs to be tied to outcomes. Well, then all of a sudden, maybe they're more equipped to run a tech company, right? Within their four walls. I don't know, I don't know.

Brett (48:28)

Yep. Well, that's

Rio (48:52)

Do they buy or build?

Brett (48:53)

one model or it's the consulting model. It's the Slalom Consulting where we came from, where you act as a systems integrator. You're moving up the value change in terms of the strategic advice that you're giving, the big ideas.

Rio (49:07)

or Omnicom acquired, Omnicom actually acquired us in SI, they acquired Crudera, right? Which is a very good, a great attack, right? So I think maybe you're gonna see more of that, Brett I mean, that's really good call out, right? And I think there is a bigger need for technology expertise and acumen to deploy at scale, right? And agencies haven't been good at it. They don't have big engineering teams, right? They have small ones scattered around, not necessarily a holdco. Maybe that's it. Maybe they either build or buy their own SIs in order to do this, right? As it become.

Shiv (49:14)

Yup. Yup.

Rio (49:35)

more tech, I don't know, it's a good call out.

Brett (49:36)

Yeah. Yeah, recommend partners, right, versus build yourself, right, is a mechanism and offer your value in terms of like, what is your campaign strategy? What is your creative strategy? What is your competitive strategy? Right. How are you going to win in market, you know, in your particular segment? Right. That type of stuff is the hard, that's the hard stuff that adds more value to the process. Right.

And is that where agencies should be playing versus the the lever pulling in the button pushing?

Shiv (50:06)

Yeah. I also think it's interesting. Like some people might argue against what we're saying and say like, well, the marketers will cut out the agencies, just go directly to the AI companies, right? Like the AI companies are the new frontier, right? Forget about Google and Facebook. Well, Google has Gemini, but like, why won't they just like, and we're already seeing this open AI is doing deals with agencies, like big deals with agencies. You know, Claude Anthropic is going to do deals with agencies, but you know, as ⁓ Eric Seufert says, everything becomes an ad network. We know open AI is going to become an ad company. We know Anthropic is going to become an ad company. And all of a sudden that direct partnership between AI company X or model company X, Y ⁓ and marketer becomes contentious and becomes biased. And you need an agency. You need the agency, right?

Rio (50:39)

It's inevitable.

Brett (50:57)

Yep, which back to your neutrality play, right? That sort of neutral provider that can offer advice that you can trust versus for their own interest. Yeah, yeah.

Shiv (51:04)

Well, they're supposed to be, it's called an agency. You're supposed to be an agent, right? You're supposed to be the agent of the client. And that, you know, just inherently means you should be agnostic, right? And anything that you can go out and buy that compromises that agnosticism, you should theoretically have an agent representing you for that tech or for that whatever it is, media, et cetera.

Rio (51:27)

Looking at the FTE model though, I mean, guess this begs the question of what replaces it, right? Does it mean that you just gradually scrapped altogether and you go more towards outcomes? Does this mean more kind of value-based work? Does this mean that they start billing? Do agencies start? I don't know, like does it become more project-based? don't know, it's, something has to replace it, right? I mean, like for sure it's gonna happen. We were talking about this the other day, when we interviewed Ben Wilde about...like SaaS, like AIs can impact SaaS because you can't charge for seats, modules, components, when are going to charge for services and outcomes? And that's clearly going to happen with AI over time. With AI, it breaks the FTE model, which I mean, I agree, like it had to go anyway. was a model that had bad incentives. It incented bad behavior, better stated. I think it had a lot of problems, but what replaces it?

Shiv (52:15)

I I don't know. I don't know. I think like the extreme opposite is the performance and outcomes based models. I think that creates lots of pain and completely messes up the entire business and is not viable in the near term. I think we've seen it more. There's probably some hybrid approach, right? Where you have like, you know, this part of this component of our business arrangement is FTE, but this other component is also tied to incentives and outcomes.

And you gotta be able to hit on both things. so I don't know the answer. And frankly, I don't think anybody knows the answer. I think the market will dictate it ⁓ over time. And I think the speed of AI will dictate it. And so I don't know. We'll see.

Rio (52:59)

Yeah, no, mean, you don't know the answers to everything, Shiv. Come on, man. No. Yeah, no.

Shiv (53:02)

I don't.

Brett (53:02)

Yeah, and we don't want the whole advertising world like shifting to direct response. That's what happens when you have outcomes based marketing, right? Everything shifts to kind of what programmatic promised. It's like, know, call to action performance marketing versus the serendipitous value of advertising creative where you're like, I didn't know I needed a Rolex watch, right? Right? I might not have been in the target segment, but there is something about advertising, the madman style of advertising that

Shiv (53:07)

Right?

Rio (53:12)

Performance marketing, yeah.

Shiv (53:14)

Yep. 100%, 100%. Well, think what's, ⁓ sorry, sorry to cut you off Brett. I was just gonna say like, I love that. And I actually think like, when we think about the impact of AI on marketing and advertising, big picture, you know, we talk, there's been a lot of talk lately on like X and social media and in the industry, like we are in the outcomes era, right? I'm sure you guys have heard that. don't, I don't, I don't get, don't honestly, I don't get the outcomes era thing, but we are in the outcomes era, right? It's all about performance and I think it's very much a function of like, CTV, you can now.

Brett (53:32)

Yep.

Rio (53:48)

Yeah, we have.

Brett (53:49)

For the last 10 years, 15 years.

Rio (53:57)

I think we've both been on debates about that on Twitter and LinkedIn. Yeah, so yeah.

Shiv (53:59)

Yeah, yeah, yeah, yeah. Yes, yeah, yeah, yeah.

So it's like, okay, fine. We're in the outcomes era, whatever. I actually think AI pushes us more towards brand and storytelling and art and pushes up funnel, right? Because outcomes, to a certain extent, the ability to drive lower funnel outcomes is all about stuff that AI can automate where creativity is not that important anymore. It's about bidding.

Brett (54:24)

Yeah, it's more about versioning

Shiv (54:27)

Versioning bidding optimization. Yeah, like we're right now humans

Brett (54:27)

and it's not the big idea. Yeah, personalization.

Shiv (54:30)

can add a lot of value but AI theoretically can do it all way better So I think like getting the bottom of the funnel turning the bottom of the funnel into business outcomes will become more of a commodity right through AI and The upper funnel is where the human value will really shine, right? And so I think

Brett (54:49)

And it's, yeah, and it's also, so I've got a learning from this, you know, being at Neustar and working heavily in sort of the MMM and MTA space for a number of years. Our financial services clients were all almost entirely, in terms of total media investment, 70 to 80%, and this is the Bank of America, it's the JPMorgan's, invested in direct response through direct mail. There's a reason why it's a high cost per acquisition.

But if you're getting somebody as a lifetime customer, lifetime value is very high, it's worth it. And so they would invest because it was so imminently trackable.

To in you know, they could really tie the actual direct mail outputs to performance They would invest all their dollars in it and we actually did a bunch of advanced MMM plus MTA sort of models that really proved that if you switched your media mix a bit and shifted 30 40 percent of your spin to upper funnel halo effect advertising and it could be at the product level It doesn't have to be just at the brand level. It could be the new Sapphire card, right? ⁓ That it would actually have a direct measurable impact if you do the data science, right? At the bottom of the funnel. It would actually have greater ROI, greater contribution. But they were so slow to change because they were addicted to the acquisition play that had worked and they could prove, right? Nothing was anecdotal.

Rio (56:14)

What was that famous Nike case study, right? Where they just went all in on performance and then neglected brand. And the next thing you know, like no one was buying with performance, right? Because people didn't know what the products were. A whole generation of consumers had been trained to buy on cloud and all these newer upstart brands that had done a better job branding. It's interesting. But Brett, you said something before that I'd like to maybe go back to a little bit too. Like you mentioned AI being a red herring. I think there's something there. And Shiv, I'd your opinion on this.

I've done a lot of work with Amazon AWS over the years. I remember someone there recently told me that they have a strict, like they basically no hires. Like if you want to hire someone, someone has to quit and you need to replace them if they want to even renew that job, that job rec, right? So they're basically been told that we're going to do more higher margins, more revenue with the same or fewer people. And it's not just Amazon, it's company after company on brand and agency and hyperscaler side are all doing this, right?

So, and I've been joking, if you have a good job, hold on for dear life because like there's not a lot of jobs, but I don't think this is necessarily because of AI, maybe a little on the margins it is, but I think there's something bigger, I think companies, this is the mandate, we're gonna grow without hiring and we're gonna use AI maybe as an excuse. So I think there's something there, your thoughts on that.

Shiv (57:29)

Yeah, mean, 100 % agree. ⁓ AI just accelerates the inevitable. AI is just accelerating the inevitable, right? Which is, why are we paying people to do things that machines can do better? Whether it's an AI machine or not, machines theoretically can do these things better. humans are not actually, ⁓ I don't know how else to say this, like, humans are not actually exerting themselves. To like, try, like, you know, what's the, yeah, like what's the, you guys saw Silicon Valley, was that character that sat on the roof, you know, and like, just like collected paychecks. And it's like, that role is gone. And you know, that's an extreme example. But again, the people that sit at a desk and click buttons, no offense to them. They're not bad people. They mean well, but from a,

Brett (58:04)

The age of leisure.

Shiv (58:27)

corporate overlord productivity perspective, that is not something where like human exertion is adding value. And now AI from a red, like it's a red herring, but AI is just kind of proving that to be true faster and more capably and quicker. ⁓ And so, you know, I think it's actually a little flawed to call it a red herring. Like it's, it was inevitable, but AI was almost needed, right? As the catalyst to make it, to be a forcing function to make it happen, you know?

Rio (58:57)

Yeah, I

Brett (58:57)

These things should

Rio (58:57)

think that...

Brett (58:58)

be automated is what we're suggesting, right? A lot of these tasks should be automated, right? You should take away the manual work. You should ⁓ level up the value that each, it's more career fulfilling. It's more interesting. We're more intelligent ⁓ to be able to offer more value, but that takes a lot of training, a lot of education, right? And there's portions of our population in the US and around the globe.

Rio (59:00)

Yeah, they probably should.

Brett (59:25)

that just don't have that training. It's sort of the pain. It goes back to the Margaret Thatcher model, right? The coal miner, you can't just change their job in day one, maybe not even in their lifetime. So how do you think about that education gap?

Rio (59:39)

Yeah.

Shiv (59:41)

Well, I just have one sentence to say, right? We want to enable people to learn how to think instead of enabling people to learn how to do. And I think that's a very important distinction, right? Like we, as a business, we've always had this philosophy of like, we do not do hands-on keyboard training. We do not think it is valuable. We only train on how to think about technology, how to talk about technology, how to form opinions, how to think about the future, right?

And I think AI is just kind of accelerating our thesis about learning and enablement and development is don't go learn how to code something, right? Learn about why you are coding or what you should be coding or the strategy of building, you know, the architecture of something, you know, like that, that is a key nuance. Yeah. Go learn how to code. Right. Go learn how to code.

Brett (1:00:29)

Yeah, it was just a few years ago that we were telling the younger generation to go into computer engineering, go into computer engineering and now that is the number one replaced job ⁓ for college grads, which might come as a surprise

Shiv (1:00:39)

Yep. Yep.

Rio (1:00:42)

Yeah, but it's interesting though too. mean, I guess counterpoint to that, right? So, I mean, I don't disagree with anything that was said, but I guess my counterpoint would be some of this mentality in corporate America, especially in tech right now, but we're going to grow our revenue. We're not going to add headcount, right? I think some of this is the hangover from the massive rampant over-hiring during COVID. I I was like, I worked at an SI. I was hiring four people a week. We were losing two some weeks. I mean, it was insane. We were hiring people without ever meeting them. Like people...We caught so many people two or three jobs like every couple of weeks. It was insane, right? And like we grew 40, 50%. I think everyone even app.

Brett (1:01:18)

You grew from like 2,000 to like 10,000 employees, right? In like a four year period, something like that.

Rio (1:01:22)

It was crazy, right? But it wasn't just us. was like Accenture did that, Deloitte did that, Amazon did that, Google did that. It was just bananas hiring. So I think, and then this gradually, okay, like a lot of these people are sitting on a rooftop collecting a paycheck or working two or three jobs or don't know what they're doing. Like, I think there's a realization over time that we overdid it, right? And then it's just been this natural, every action there is equal opposite reaction, right? So think some what we're seeing is kind of understandable reaction to that over hiring and CEO is saying like, don't want to be the guy that does that again. I don't want to be the one that over hires and like drives the company and do a ditch, right? So some of it's that. So maybe it's an overreaction, but I think Shiv, to your point, AI is opening up better opportunities and avenues for companies to optimize things, to automate things, even if it's not quite there yet, it's getting there quickly. And I think companies, I think that's part of what's driving this mentality, let's call it, right? Of the way companies are just, the way they want to operate. It's fundamentally different from the way it was four or five years ago.

Shiv (1:02:26)

And I also think about like, you know, COVID, why, why did we hire so much during COVID? It's because there was growth. There was, first of all, there's a lot of change. Consumer behavior shifts were like, you know, all of a sudden changing. It was crazy. And there was a lot of growth during COVID, right? ⁓ people were consuming digitally a lot more because people were at home. And so all tech companies were like, we got to capitalize on this. We are not equipped to capitalize on this using machines.

Brett (1:02:36)

Consumer behavior shifts. Yeah.

Shiv (1:02:56)

So we need human firepower to accommodate the demand. And so we don't care if we over hire, like maybe this is just a runaway train. And obviously it wasn't, right? COVID ended and like things semi-returned to normal. And so I think like it's important to think about that aspect too. And now, like, first of all, we don't have that kind of growth anymore, right? It's like, why would we be hiring like that? And second of all, we have AI as firepower to handle demand increase and also we over hired during the pandemic. So it's like all these things are, it's a confluence of all these factors.

Brett (1:03:31)

So how do you thinking, so looking sort of next year, you know, with the meta announcement to bring it full circle, ⁓ do you foresee a kind of a ceiling of how far automation can take us? Do you see an end to some of this compression of head count?

Shiv (1:03:46)

I do, I do. I don't think it's indefinite. I think you at some point, we talked about evolution earlier and we talked about revolutions. And so we're going through one of those. There are short-term pain, people are going to lose jobs. We're going to retrofit the workforce. And then we're all going to be ready for a new AI-centric world. And there will be jobs that will be needed in that world. ⁓ And then five years from now, there will be a new revolution or 10 years from now, there will be a new revolution that we can't even conceive and we'll go through the whole thing again. But yes, there is a ceiling to the headcount compression. ⁓ we need AI, the pace of AI, when it starts slowing down, right? It will, it will, right? At some point it will, maybe it's three years away, maybe that's eight years, I don't know. But at some point it will stabilize a bit and we will live in a new world where we know how many people we need to do what kind of jobs. So yeah, I think there, I don't think it's next year. I think it's probably a three to five year time horizon, but.

Rio (1:04:47)

It probably takes a couple of years. And Shiv, if you agree with the hypothesis that a lot of this has just been driven by a reaction or overreaction to the hiring binge, that comes to an end. At some point, companies, okay, we're healthy, we're profitable. So think part of it's driven by, I think that will subside. I think that the new jobs created by AI, we're only starting to see those. We don't know what many of them will be, but that will happen. And I think at the same time, like the, you know,

Shiv (1:04:56)

Yeah. Exactly.

Rio (1:05:12)

I mean, these models have to evolve. I think that's natural, that's normal. things change, right? And we have to get comfortable with that. Agencies will be here, they're going to look different, they'll bill differently, but they'll still be here and they'll still be doing probably a lot of the same things, but they'll just be doing it different.

Shiv (1:05:27)

Yep. Yeah, and think it's just the simple, sometimes it's like, you know, the simple thing is actually the answer, right? Routine, rote work will become commoditized and automated and AI-ified. Creativity, strategy, things that require human exertion of the brain will be valuable, will become more valuable, right? There may not be more jobs, unfortunately, in the near term, but those people will get paid more, the ones that do it well will get paid more, et cetera, right? So like,

That's how, know, if you're somebody out there thinking about your career, think about it in that way. Think about thinking, think less about doing, think more about thinking. ⁓ That would be my advice.

Rio (1:06:05)

Yeah, it's funny consulting. I would always joke with people because, you know, it's easier to get people to take notes to like synthesize this to build like 100 slide decks. But I would say to people like, what does it mean? And what are you talking about? So well, the clients are paying us. Yeah, clients are paying us like $250,000, $500,000 for a bunch of PowerPoint slides. we not so you can regurgitate with what you heard. They want to know what does this whole thing mean? Like what's your conclusion?

Brett (1:06:15)

The So What.

Shiv (1:06:28)

part. That's the hardest part. Like people read, but they don't then form an opinion because the forming an opinion is the hard part, right? People will make decks, but don't think about the strategy behind why they're making the slides they're making. So the people that aren't thinking about those things are the people, unfortunately, that are going to lose their jobs until they start exerting themselves to think and not just do.

Brett (1:06:49)

Yeah, thinking with the end in mind, right? So you always have a goal in mind. And then I think bridging strategy and execution to me is oftentimes the hardest thing. A lot of people have good ideas, but very few people are actual creators, builders that can, you know, whether it's a product whether it's a creative execution, whether you're a news publisher and you're investigative journalism, those are the builders and the creators. That's the stuff we need to value. Everything else you can automate, the versioning and the, ⁓ but it's the builders that that's where humans add value. And have you noticed how we've used the word human?

A lot more in our daily lives than I think I have in my entire life. I hear it in board meetings, in leadership meetings. People always talk about the humans, and I'm like, that's us, right? It just sounds very odd. And it's like the future is here now. ⁓ I've noticed that just in their diction, like how we talk to each other, right?

Shiv (1:07:39)

Hahaha! Well, I actually like, I'm a glass half full optimistic person and I'm very excited about this new world because I think in the last 10, 15 years, humans have kind of become robotic, right? We're addicted to social media, we're doom scrolling, like we don't understand human connection, we don't value it all that much. And I think like as AI comes in and automates like routine things and rote work,

It's going to force humans to have to think more and be more human and make human connections and value that. And I think that's very positive. again, it's like, think evolution is positive. This is a moment of evolution for humans. ⁓ And I think it's gonna lift us. I think it's gonna lift people.

Brett (1:08:40)

Yeah. And it's sometimes hard to see it when you're in it. Right?

Shiv (1:08:44)

Totally, yeah, well, people are just thinking about their jobs, which is fair. It's fair, know, the livelihood matters. ⁓ you know, but looking at it from a, you know, bird's eye view, and I'm not saying I sit in that place, like I'm a human too, and my livelihood matters too, but if I'm able to kind of separate myself from that for a second and try to think of it from a bird's eye perspective, it is positive, right? We just have to like kind of think of it that way, ⁓ change our mindset and like get on the train. Everyone, let's all get on the train, right?

And getting on the train is like empowering yourself, enabling yourself learning. One thing that like, one of my biggest ⁓ tips to people is like, you know, people always say a superpower is to always be learning, right? And that's like one of the best things you can be doing. And I think that is true for the AI era. Like you have to be in a learning mindset constantly. But I also think that it's going to be really important to be able to forget. And what I mean by that is like, Humans have inertia. We start doing things in a certain way. We feel like we're good at it. And then we just want to keep doing it that way. But because the pace of AI is so fast, is changing so fast, the way we do something one week, we have to be willing to forget it next week and do it a new way. And I think that requires like a mentality of I'm willing to change constantly, right? I'm willing to forget. I'm willing to evolve constantly. So I think that's like a, yeah.

Brett (1:10:05)

Shiv, this is like your mantra. This is like you're moving into Buddhist territory. Like you've got to let go, right? It's all about letting go in the moment. No, I tell myself that every day at work. It's, you know, as you move up in your career, it's all about letting go of things that you once held dear, right? And it might be something rote. It might be something individual contributor. Yeah.

Rio (1:10:09)

Yeah like inbox zero.

Shiv (1:10:28)

Yeah.

Rio (1:10:29)

Like that was advice I got. Let go of I used to have inbox zero I was told let go of it. It's stress. You're spending too many hours a day messing around with emails. And I have like, I don't even want to tell you how many unread emails have my inbox. Yeah. Yeah.

Brett (1:10:38)

my God, like in the tens of thousands.

Shiv (1:10:41)

67,205 right now.

Rio (1:10:43)

there's no way you could ever go through them all. That would be your job and that's not a job that, that's not how you could best spend your time. So I think that's a good analogy, I like that.

Brett (1:10:52)

Yeah.

Shiv (1:10:53)

One last thing that I didn't even get to that I'll leave the listeners with, leave you guys with to sleep on is like, we talk about agency jobs, we talk about how marketing and advertising is gonna change. One thing we didn't even touch on is how the consumer experience will change and how that will change advertising and attention. We didn't even talk about like, okay, well in the future, people are not going to be using browsers and clicking around, right? They're not going be looking at display banners, right? They're going to be talking to their AI bot. How does that, the change of the consumer behavior and how they go about their day and how they use productivity tools from maybe a voice-first perspective, how is that going to change marketing and advertising and jobs and agency models? It's just going to turn it all on its head.

Brett (1:11:49)

I think that's a whole nother episode of a pod, totally.

Shiv (1:11:52)

like media and advertising and user attention and consumption is all gonna be completely different a few years from now. How does that affect what we do and our jobs and our livelihood?

Brett (1:12:01)

Yeah, and speaking of the end in mind, and we'll make this the end in mind here is that really maybe that is a follow-up podcast where it's like, start there. Start with that end delivery and consumer experience, right? And then move back and you'll probably come to a better solution,

That was terrific, Shiv.

Rio (1:12:18)

Thanks again, this is amazing.

Yeah, it was good.

Shiv (1:12:20)

Thanks guys! Yeah, that was a lot of fun.