- 4 days ago
AI will have a tectonic impact across many marketing functions from media buying to creative and internet navigation. If the storefront is no longer a website but an AI interface, what does a brand actually still own? How should CMOs prepare for an agentic ad world where everything from discovery to execution is intermediated by agents.
Category
🤖
TechTranscript
00:27Good morning, good morning everybody.
00:33Thank you for that warm welcome to the stage.
00:36I'm Will Lee, the CEO of Adweek, and I'm joined here by Terry Kawaja, David Steinberg, and
00:41Kurthika Reddy, three people who have a very different but incredibly well-informed and
00:48frankly success-driven viewpoints, points of view on this subject, life after search.
00:56I think we should probably reframe it more as like life after Google search and life
01:00and really like what the agentic and AI marketing world's going to look like into the future.
01:08So actually, I'm going to turn it over to my friend Terry, who created the Loom Escape,
01:14now the AI Loom Escape, to kind of help frame this for us.
01:18And because Terry, you know, loves a stage, he's going to, we're going to have Terry
01:24stand up and give us a little bit of a framing.
01:27So again, by this point at VivaTech, right, it's pretty much acknowledged by everybody
01:33that AI is going to have a massive impact on advertising.
01:36Advertising is pretty much one of the best applications for AI, given the amount of spend,
01:43given the availability of data, and the amount of change you can have.
01:48You can convert the customer or not.
01:49So great, it's going to have a massive impact.
01:52But where?
01:53We believe it impacts all aspects of advertising.
01:57And a way to think about this, you remember Maslow's Hierarchy of Needs,
02:01that sort of pyramid that showed the different layers, and the further up you got, the more
02:06important, well, think about the base layer, right, workflow, media buying, audience, data,
02:14measurement, the sort of the meat and potatoes, the nuts and bolts of advertising.
02:19Clearly, AI is going to have a big impact there.
02:22It already is, as companies are building agents so that they have interfaces to do many of these
02:29tasks.
02:30The observation I would make with respect to that layer is that workflow is largely going
02:37to be an equalizer.
02:38It's going to be a commoditizer.
02:39Everyone will adapt this new workflow.
02:42And I would submit that decisioning is actually the differentiating capability there.
02:49The ability to impact the effectiveness of advertising.
02:54And what fuels that?
02:56There are two things.
02:57One is your model.
02:59Two is your data.
03:01So to the extent, and you'll be hearing a lot about how data impacts decisioning, but
03:06watch for decisioning as the differentiator there.
03:09Next layer up would be sort of creative and content.
03:13And yes, there's going to be cheaper production.
03:16We're already seeing award-winning movies being created with AI.
03:19So that's great.
03:21But I would also submit that it's the data around creative that helps you determining
03:26the effectiveness of the ad that is actually going to be the true differentiator.
03:32And then third, and the subject of this panel, is internet navigation.
03:36So we're now further up the Maslow hierarchy of importance.
03:40And that, of course, is this shift from, you know, search, largely Google at this point,
03:46to the LLMs, and deriving advertising in what will be a new format.
03:54And think about it.
03:56There's absolutely no question that the LLMs are going to utilize advertising because of
04:02the high intent signal.
04:03In a Google context, if someone entered the search query, Cleveland divorce lawyer, you
04:11knew three, Google knew three things about that person.
04:14Listen, they live in Cleveland, they're getting a divorce, and they need a lawyer, right?
04:17So it was a high intent signal.
04:19It allowed them to build a quarter billion dollar, trillion dollar business.
04:26Take that to a chat environment, and the user is going to start with 25-word initial prompt
04:33and follow it up with multiple pumps.
04:34You're going to have a much clearer signal as to what the intent of the user is, and therefore
04:40be able to monetize it.
04:41Now, OpenAI, as one example, has announced that they want to hit $100 billion in five
04:48years' time.
04:49Very ambitious goal.
04:50So clearly, they view advertising as a key way to monetize their LLMs.
04:55The question is, will they go to the last mile, or will there be a vibrant ecosystem of
05:02companies that can help fuel that?
05:05I would assume the latter, because these are companies that are nascent, they don't have
05:13all the capabilities, there are multiple verticals, and I think what we're going to see is a vibrant
05:20ecosystem.
05:20Sorry to interrupt your soliloquy here, but what I would say is they also don't have any
05:25deterministic data unless somebody's logged in.
05:28So when you look at the evolution from search, and as usual, you're totally right, a year
05:34ago, 97% of all queries on Google were referred off of Google to a brand and e-commerce platform
05:42or to a publisher.
05:44Today, 67% of all queries on Google are being answered on Google.
05:50That is changing the game for how marketing is operating.
05:54And at Zeta, as you know, not only are we a large GEO partner with OpenAI and other models,
06:03we're also the first platform that they've announced a partnership with where we're serving
06:08ads into OpenAI.
06:12They have incredible signals, to your exact point.
06:16But they don't have the deterministic data to really target.
06:21So by working with Zeta, and I'm sure a few other companies, I don't think we'll be the
06:27only person who does this for them.
06:29At some point, they'll bring in others.
06:31The ability to take a deterministic data set and use that to target inside of their incremental
06:38signals, which to your point, totally right.
06:41You're getting a chat.
06:42You're not just getting two or three keywords.
06:44I think you're going to see a game change in marketing.
06:47Well, and the only way they can apply is on-prem.
06:51So when you're on the LLM, they can use that signal.
06:54They can optimize there.
06:55But the vast majority of all interfaces with consumers is going to be off the LLM.
07:01Yeah.
07:02And they're going to need the data.
07:03And we call that audience extension, which we're already doing.
07:07Perfect.
07:08And will I actually get a sense of the audience as well?
07:11Like how many folks here represent brands here?
07:14Maybe a quick show of hands.
07:16Yeah.
07:16Oh, good.
07:18Maybe agency, the agency world.
07:20Okay.
07:22And media industry, tech industry, entrepreneurs.
07:27Okay.
07:27Creators and influencers.
07:29Okay.
07:30All right.
07:30Beautiful.
07:31For the brands, how many of you are already on your way off in the whole generative engine
07:36optimization, AEO, you have a dashboard, you know what's going on.
07:40Okay.
07:41All right.
07:41I think about half.
07:42Amazing.
07:43Because that is the reality of today, right?
07:45I mean, having seen multiple technology and consumer transitions, consumers are going very
07:53rapidly from Google search to finding everything from skincare advice to which enterprise solution
07:58you want to buy on chat, GPT, perplexity, and the like.
08:01So, if your brand is not present there, you're essentially forgotten by the modern consumer.
08:07Our platform, we have the privilege of working with the Fortune 50 to startups across the globe,
08:12across sectors.
08:13And what we find the best marketers do is, one, establish their geo scorecard very quickly,
08:20but make sure that they're not thinking of these as just vanity metrics, but looking
08:25beyond those metrics to see what's going on, because these metrics are actually a window
08:31into everything that's happening across your marketing mix, and it's a return to the brilliant
08:35basics.
08:36What are you doing on your content strategy?
08:38What are you doing in terms of making sure the content on your site is crawlable and findable
08:43by these LLMs?
08:45What about your media strategy, partnership strategy, pricing, positioning?
08:48That's the window that these metrics provide, and the best marketers are looking deeper into
08:55those metrics and figuring out where they stand across different parts of the marketing
09:01funnel and optimizing for it.
09:02All right.
09:03So, the thing that I'm very curious about, and this will be very germane to the brands,
09:07also to the agencies, is given what Optimize GEO does, and also your company as well, David,
09:17in this world, where does the relationship with the customer truly begin and end?
09:23Like, this is a question that I think is very much in the ether, and I'm curious, Kurthika,
09:27you first, and then David, like, how you see it.
09:31I strongly believe that that relationship rests squarely on the brands and the different platforms
09:38and the different partners they bring together to help maintain that control, because these
09:42LLMs are simply, and the agents, when you are doing agent e-commerce, are simply finding
09:47information that you, the brand, have put out.
09:50Either directly, it's your classic owned, earned, paid media, and how is it that you're
09:55getting on top of it to understand what's being presented and influence it in a way that
10:02matches reality?
10:03You know, by the way, we have had, you know, a consumer package, good company, CMO, come
10:08to us and say, you know, I went to the LLMs and said, is my product healthy?
10:12And the LLMs said, no, it is not healthy, and these are alternatives.
10:16Can you fix it?
10:18Okay?
10:18To which we said, if it's really not healthy, honestly, we can't fix it.
10:23Right.
10:23But however, if you also have some health product lines, and let's make sure that we
10:28can make sure that those products are surfacing more, in a more relevant manner.
10:33So that is what we can do.
10:35And that's an example of how brands can really own what it is that the LLMs are surfacing by
10:40digging deep into the data.
10:42And I would say it depends on the vertical, right?
10:46So if a product is being sold through an intermediary, whether it's a store or not, it's a different
10:52touchpoint with the consumer than if an enterprise is selling directly to that consumer.
10:58And what we have found as we look at it, because I don't know any other company other than Zeta
11:05that is plugged into OpenAI, partners with Meta, partners with Google, the open web, connected
11:12TV, messaging, so on and so forth.
11:14The different touchpoints are key to the different relationship that the brand has
11:22with the consumer.
11:23If you're trying to convince a consumer that your product is superior when they're going
11:27to go into a retail store to buy it, that's a different communication point and a different
11:32attribution capability, which we have through relationships with Visa, Master, and American
11:37Express.
11:38But you then have, if you're going direct, let's say you're an insurance company, a financial
11:43services company, an issuer of credit cards, you're going to have to get directly to that
11:48enterprise.
11:49If you're going through an intermediary, I actually think GEO is more important.
11:55Because when you're looking at building that brand and building that relationship, GEO is
12:01really important.
12:02And she's so right.
12:04You cannot take an unhealthy product and make it healthy inside of GEO.
12:08But you can point out the attributes of that product that are the most positive.
12:14It's much like, I sort of talk about it much like it was sort of the very early days of
12:19search engine optimization, right?
12:22Before you understood sort of link building and how the spiders worked.
12:26You know, GEO is different and it's more complex and you have to get access to the APIs
12:31to do it, which of course we've done and others have done.
12:34But when you're going direct, what will often happen is, and back to a point Terry made,
12:40they might start their journey to buy a car or to get a new credit card or to get an
12:48insurance policy
12:49inside of a large language model.
12:51They're not going to buy it through there.
12:53So you need to be able to extend that marketing outside of there.
12:58And that's something we're already doing.
13:01I mean, in many ways, I think this notion that AI is disintermediating the relationship between
13:08the brand and the consumer is a bit of a false premise in the sense that brands have been
13:15distanced from their consumer.
13:17There's been an intermediary at almost every evolution of technology.
13:21So retailers, you know, where they were on the shelf was a disintermediating, you know,
13:27component.
13:28And then search and in social, there's always been that intermediary player.
13:32Now we've got a new player called an LLM and maybe the complication has gone up, but brands
13:40are going to have to learn the new capabilities to be relevant and get recommendations and ultimately
13:46deliver.
13:47I think what the consumers, when they configure their agents to sort of give them recommendations
13:54in terms of their preferences, they're going to want to instill the trust.
13:58Because at the end of the day, what do they have with consumers?
14:02They've got trust, which no intermediary is going to be able to alter.
14:06So, okay.
14:07So I don't disagree with the notion of the sort of constant disintermediation, but the difference
14:13here is that to your point about Cleveland divorce attorneys or, you know, toothpaste or
14:23winter coats, right?
14:24The choice is being made for you in some respects.
14:28It is being presented to you as the choice because the LLM knows a lot about you and so
14:34can suggest a choice.
14:36With Google, you would have a variety of choices to make.
14:40Well, I just want to say, the LLM actually doesn't know a lot about the person yet.
14:46Yet.
14:46And they're working on that, whereas Google knows a lot about everybody.
14:52And obviously, we're big partners with OpenAI.
14:56We work with Claude.
14:57We work with Gemini.
14:58I don't see these guys disintermediating Google search business anytime soon.
15:04It's sort of like when Netflix decided they were going to go into the advertising business.
15:09They're still figuring it out, right?
15:11But I think you're right, though.
15:13And I wasn't trying to disagree with you.
15:14I think it's going to be on that curve as they get to know the individual and more about
15:20them.
15:21Well, David, they know a lot about you.
15:23Oh, we see their data every moment of every day.
15:25They actually have a small percentage of their users that are logged in.
15:29You know, even if you're not logged in, it actually keeps track of the conversations.
15:35And it's scary.
15:37Like, I ask for a seven-day nutrition plan.
15:39It knows exactly what my preferences are and suggests things that are tailored to me.
15:45And at the end of it, I don't even actually have to ask another question because all the
15:49LLM says, oh, do you want me to tell you more about X, Y, and Z?
15:52So all that I do is say, yes, yes, yes.
15:54And I'm ending up with a shopping cart.
15:56And I'm not suggesting they don't know stuff, but I would say it's 1% of what Google knows
16:01about you.
16:03And I see the data from both.
16:05So you might be special in that you're feeding in a lot of stuff that is then informing them.
16:11Most people are using OpenAI and Gemini today to write poems, write letters, respond to emails.
16:20Claude is doing more complex stuff.
16:22But from what we're seeing, and we operate in both ecosystems, nobody knows more about the consumer than Google at
16:31this point.
16:31And by the way, if they leverage that data, that's going to be – the agents are not going to
16:37make de novo decisions, right?
16:39They're going to make decisions on the basis of what they know about you in terms of your preference.
16:43There is a preference stack already built in, which would suggest that legacy brands, brands that the consumers have historically
16:52utilized, could be fine.
16:54I would have more of a concern for the direct-to-consumer, the new brands, trying to break in.
17:00Because how do you build that database of preferences that the agents are going to immediately adapt?
17:07So you said the word that I'm very curious about, which is trust.
17:13So all the data, all of the machinery behind the scenes, but Kurthika, maybe you can address this first, is
17:23what does brand trust mean in this?
17:26And how do you build it in a world when your trust is basically being embedded in a machine?
17:35Yeah.
17:36So I'll take it in two different ways.
17:37You know, one, you know, we work with a lot of companies in highly regulatory spaces like Heltec, like FinTech.
17:44So there, there's a lot of work happening about are your products being accurately represented in LLMs, right?
17:51Because they could be severe, like consumer issues and regulatory risks if you're not.
17:57So there, there are tools like ours and a couple of others that help you find your accuracy score or
18:04your trust score.
18:05Where you give, this is my source of information.
18:07If people are asking about the pharmacological composition of X, what happens in chat GPT, version X, version Y, version
18:15Z?
18:16And then further, where did these LLMs find those sources of information, whether it was your own content, other content,
18:24and how is it that you need to fix them and have agents that can help you do that, right?
18:27So that's one part of it.
18:29But the other part of it, which I think what you're asking is just a basic trust.
18:33Like if there is a product and how do you get that brand loyalty, customer loyalty with the brand, right?
18:40And so there, again, you are able to dissect different parts of the equation.
18:46We work with prominent brands that could be number one in AI visibility.
18:51But for even the most prominent brand with the highest media spends, on an average, we're finding that there's about
18:5840% of the user queries that the brand has determined that is important.
19:02That they appear in where they have 0% visibility, right?
19:06So then it's about digging deeper into where those 0 visibility gaps are and how is it that you fix
19:12it.
19:12So, for example, we have found for a Fortune 50 baby care product where they were very high in visibility,
19:20but their consumer safety queries were not being addressed based on all of the information that they were putting out
19:26there.
19:27Or for a personal device product, there were a lot of questions on post-device care that was not being
19:33addressed.
19:34So how is it that you get that entire landscape and start filling those gaps where there is no visibility
19:40in these LLMs?
19:42Yeah.
19:42And when you think about brand trust, right, it goes across every methodology, right?
19:49GEO is going to be important, but it's still a de minimis percentage of how consumers interact with a brand
19:56today.
19:57It's an important one for trust because when they're reading it, they're going to see that trust and they're going
20:03to understand it.
20:05Effectively, brands build trust based on their products and services and how the consumer consumes them and feels about them.
20:15We can do anything we want from a marketing perspective to try to amplify that.
20:21But ultimately, if a consumer has a bad experience with a product, they're not going to trust it going forward.
20:28So back to her original point, which I thought was so good, you can't take a non-healthy product and
20:36make it seem healthy inside of GEO.
20:40It's the same thing with brand trust.
20:42What you're trying to do for brands is you're trying to amplify their message to the customers who do trust
20:49them and keep them on the forefront of their buying strategies and bring new customers in.
20:56It's very hard to win back customers you've had a bad relationship with.
21:00We try and we do some of that.
21:04But the real sort of marketing capabilities are the amplification of a good relationship or the creation of a new
21:14brand from a thought process perspective with the consumer.
21:18And then, of course, it all goes back to attribution.
21:20And most of the models we operate, we plug into the CRM systems for our clients, and they get smarter
21:27based on the purchase.
21:28And what we see is the greater the retention, the greater the percentage of wallet share a brand has, the
21:36better they interact with that customer in the real world, which then leads to a better experience in the digital
21:44world.
21:44Okay, so let's shift the focus a little bit to how this is going to affect teams.
21:51There are clearly a lot of managers, executives, practitioners here.
21:56And, Terry, I'll start with you.
21:59When you think about sort of the AI-first org chart, right, and when you think about the companies in
22:07the AI Loom Escape and then the effect that they have on the companies that are outside of it, what
22:11are the roles and the functions that you think will be irrevocably changed or potentially eliminated?
22:20I think that's important to know.
22:21Sure.
22:22I mean, there is this premise out there.
22:25I noticed this last year.
22:26And it was a thesis that your existing incumbent service providers, intermediaries, were finished.
22:36In this AI world, it was only going to be the AI-first companies that, with these new solutions, that
22:41were going to obviate the need for all of those legacy intermediaries.
22:46And Kierthika and I were together in November.
22:49We gave a big presentation.
22:52We said the exact opposite.
22:54It's not David versus Goliath where David wins.
22:56It's the smarter incumbent companies actually have the advantage in an AI world.
23:04Because, as we've seen, Claude code building is recursive.
23:10Claude helps you build more code.
23:13So, building the code is not the challenging part of it.
23:17What matters, what permeates, what sustains is having tons of interaction.
23:24It's why Publis' bot LiveRamp is because for all those data connections throughout the ecosystem.
23:30And the unique signals, the unique data sources David referenced earlier, those are the differentiating capabilities.
23:39So, Will, when you ask about, you know, what capabilities, I'm not sure they go away.
23:45But as I said in my sort of opener, workflow becomes kind of, not commoditized, but we adapt to this
23:52new workflow.
23:53And therefore, the most important thing is your algorithms and your unique data sources.
24:00Because you combine those two together and you've got a clear signal on which you can make decisions and determine
24:06the effectiveness of the advertising.
24:08Totally agree.
24:09You know, when you think about the internet, right?
24:12When the internet came down the pike, everybody said it was going to disintermediate JPMorgan Chase.
24:18It was going to put Walmart out of business.
24:20You had two types of companies coming out of the internet.
24:24You had the ones who adopted the internet and became juggernauts and the ones that did not and are mostly
24:29gone.
24:30When you look at the beginning of AI, I think you're going to see organizations that are very focused on
24:36workflow management and not the creation of intelligence are going to have a real challenge going forward.
24:44I think companies that create intelligence, meaning, to Terry's point exactly, they have proprietary data sets that they can utilize
24:54in these models that allow them to create actionable intelligence around either, whether it's business intelligence, workflow management, marketing, healthcare,
25:05evolution, so on and so forth.
25:08Proprietary data sets are going to be a massive differentiator, but I think you're going to see intelligence creation is
25:16going to be the single biggest differentiator between the winners and the losers.
25:21And I don't mean just organizationally.
25:23I mean as individuals.
25:25I mean as the people who are operating and, of course, organizations going forward.
25:31You know, I couldn't agree more.
25:32So there's, you know, I'll give you an example.
25:36We work with over 16 brands in a Fortune 50 company, and even within those brands, it's about the brand
25:42managers and how they're structured and how they approach the problem, right?
25:46And also, what are the silos as you think about org structures in terms of who talks to who and
25:56who has the influence over whom in making the decisions of the recommendations coming from what you need to do
26:03to increase your visibility in AI searches?
26:06And for those who say, oh, I am limited, I have five other organizations to influence, and I can't make
26:11those change.
26:12I can only pick, you know, these small things.
26:14We actually see a very big difference within even the same company in terms of the approaches and in terms
26:20of how organizationally they're structured.
26:22So, actually, I have a flashback back to the days of the times when we, you know, we used to
26:27have a head of the internet or head of mobile or head of social to, you know, the head of
26:32AI or head of AI search.
26:33And now it being something that is everybody's responsibility, really.
26:39And I remember in the meta days, I was the first employee for Facebook in India.
26:43They're managing directors who saw the country go digital and mobile.
26:46And we used to say, like, social can't be the sprinkling on your fries.
26:50It has to be the main ingredient that you make the fries.
26:53So, that's how you have to be thinking about the information you're getting from AI search and AI is how
26:58does it permeate across your different functions and move the bar.
27:01So, true.
27:02But do you think that, you know, again, we hired SEO specialists.
27:08We hired optimizers.
27:10We had, you know, and so two questions, actually.
27:13One, how many companies do you advise that have that function today?
27:17And are you advising them to create that function in the next six months to 12 months?
27:23Yeah.
27:24So, we are finding that it's either the head of SEO who takes the mandate, the head of growth, the
27:30head of innovation.
27:31So, it's actually very disparate in terms of who takes the mandate of parsing the intelligence coming from AI search.
27:39And the ones that are successful are those that have the ability to influence the other function versus it just
27:45sitting in one function.
27:46To her point, the higher in an organization the ownership of AI adoption is, the more likely it is to
27:56be adopted.
27:57And I would tell you that, you know, we started programming AI as native to our application layer in 2017.
28:06Everybody thought we were totally insane at that time.
28:10And if I hadn't been the chief AI officer from 2017 to 2021, there is no scenario that we would
28:19have been able to be there.
28:20We work with 51% of the Fortune 100, 25% of the 500, and, you know, I think we
28:26have 600-plus global enterprise clients that spend about $110 to $120 billion a year on marketing in this calendar
28:36year.
28:36And from our vantage point, when the CMO is taking ownership, when the CEO of the organization is taking ownership,
28:46the adoption rates skyrocket.
28:49When they take a junior person that's the sixth person on the deal team and say, you're the AI person,
28:56you don't see the adoption happening.
29:00I mean, look, there's a reason why CMOs have an average tenure of like 27 months or some ridiculous thing
29:06like that, 18.
29:08It's getting shorter as we go on.
29:10They have a challenging job because of all of these technology advances are accelerating.
29:15At the end of the day, you know, they have to bring their organization into this new realm because it's
29:22basically existential table stakes for them to survive.
29:26And Terry makes an interesting point because for many years, the CMO, oddly enough, has been reporting to the CFO,
29:34which didn't make a lot of sense to me.
29:37You're seeing organizations now that are hyper-focused on growth making that CMO report directly to the CEO and making
29:47it their job to be a revenue generator, not a cost saver.
29:52And I think it would be interesting, Terry, to do a study of the success of organizations where the CMO
29:59reported to the CEO versus the CFO.
30:02From our vantage point, they're able to create and grow much faster.
30:07You should never define your job title by the expense that you put out.
30:12It should be chief growth officer.
30:14I'll give you the last word.
30:15Well, you know, I want to give a huge shout out to all of the creators and influencers in the
30:20room because your voice matters even more in this new world.
30:23And the LLMs are looking at you as a source of authority versus, you know, necessarily where the media spend
30:29is going.
30:30So kudos to you and kudos to all of the brands who take that approach.
30:34For the challenger brands, this is market share is won and lost at times of transition.
30:39It's a huge time of transition.
30:40It's your opportunity to leapfrog.
30:42And if you are the dominant brand, it's your opportunity to behave like the challenger brand.
30:47Awesome.
30:48Fantastic.
30:48Thank you very much.
30:49Thank you, everybody.
Comments