← AdExchanger
AdExchanger
AI Isn’t Something To Fear – Or Rush Into
00:52:03
AI isn’t something a marketer “does.” It’s not a singular action or task, and there is no AI easy button. And so AI researcher and consultant Cecilia Dones tries to “get underneath” the reasons behind why marketers want to integrate AI...
Loading summary
Transcript57 lines
- [00:13]
Alison Schiff
Welcome to Ad Exchanger Talks, the podcast devoted to examining the issues and trends in advertising and marketing technology that matter most to you. This podcast is sponsored by Data Axle, where actionable data fuels connections. For over 50 years, they've helped businesses turn information into insights, igniting meaningful relationships between companies and the people they serve. With cutting edge data solutions, AI and technology, they empower brands to engage their audiences in smarter, more personal ways. Learn more@dataaxel.com I'm Alison Schiff and you are listening to Ad Exchanger Talks, the podcast for nerds by nerds and starring nerds. And I mean that lovingly. My guest this week is Cecilia Doness, Founder and Chief Data Officer of Three Standard Deviations, a consulting firm and advisory that helps organizations extract the most value from their data, but without forgetting about the human aspect and how this affects people. We'll talk about what some consider to be a thorny topic, which is data and AI ethics. But Ceci argues, and correctly, that ethical principles should be foundational for any marketer, brand, buyer, publisher, ad tech vendor, really, any anybody that wants to use AI to create solutions and drive business outcomes. But don't call up Ceci and tell her that you want advice on how to, quote, unquote, do AI. AI isn't something that you do. It's not a singular action or task. AI needs to be implemented strategically and with an understanding of why you're doing it and what your goals are. We'll talk about that, but first, a quick plug for our next conference, and I'd like to invite you to snag your ticket. Programmatic I o Innovate will take place May 19th through the 21st in sunny Las Vegas. We've been putting together a great lineup for you, including can't miss sessions on key areas where AI will catalyze change in marketing. Speaking of, Ceci will be speaking at programmatic IO Innovate as part of our new Innovators Lab on May 19th. Her session is called what is a CMO to do with AI? And she'll share lots of practical advice for marketing leaders who are ready to move beyond buzzwords and into meaningful AI integration. Visit our website to learn more and reserve your spot. Podcast listeners get 10% off as a thank you for listening. Use code POD10 to get your discount. Welcome to the podcast.
- [02:50]
Cecilia Doness
Thank you, thank you. I'm very excited to be here.
- [02:53]
Alison Schiff
So I'm going to ask my favorite question, which is what is one thing about you that not a lot of other people already know? And it can be nearly Anything at all. But it has to be something that I wouldn't be able to easily find out by looking at your LinkedIn or googling you.
- [03:11]
Cecilia Doness
Okay, so the good news is I have had enough trips around the sun that not everything is online. Something I haven't said in a long time, so hopefully everyone's forgotten at this point. I used to be a competitive ballroom dancer.
- [03:29]
Alison Schiff
Wow, fancy.
- [03:32]
Cecilia Doness
Yeah.
- [03:32]
Alison Schiff
Do you still dance?
- [03:34]
Cecilia Doness
I do. I. I dance Argentine tango now, which is a little bit different, but I still love partner dancing.
- [03:43]
Alison Schiff
In college, we had to take a physical activity, a gym style class, so I took ballroom dancing, which counted, and I was terrible. And I learned all of the male moves because they didn't have enough men and women in the class, so they had me learn the male moves so I can sort of do male style ballroom dancing. Sort of.
- [04:11]
Cecilia Doness
Oh, okay, okay. I actually, I. I do switch, so I do both side, so it's a good skill.
- [04:19]
Alison Schiff
And the first time you and I chatted in December, you said something very funny, that you've always been proudly a cord, never. Mainly because you were so broke coming out of university that you, you couldn't afford a landline or, or a cable bill. And I was just thinking, because I mentioned college and it, it occurred to me that you were mobile first and I can really relate to that because I've never had a TV of my own and I got my first cell phone, a flip phone, when I was a sophomore in college, and I think my first smartphone in 2012 or something like that. So Chord Nevers, unite. And you're definitely a better ballroom dancer than me, though.
- [05:01]
Cecilia Doness
You're very, very kind. Thank you.
- [05:04]
Alison Schiff
So, a little bit about you for, for our listeners. You're a marketer. You were a marketer and you're a statistician by training, so a master in Statistics from Columbia. And then there were some digital analytics roles at Mindshare Way 2007, 2008 and the Zenith Media around 2014. And then you spent time on the brand side. You were Assistant VP of Advanced analytics and Data Science at L'Oreal and then head of Data Sciences. I thought that was funny because it's plural. I usually only see data science, but head of Data Sciences at Moet Hennessy and you've also been an adjunct at NYU and Columbia, which both your alma maters, teaching on a whole host of topics like marketing technology, ad tech, data strategy, analytics methods, and CRM data ethics. All of this good stuff. And that just brings me to three standard deviations, which is an Advisory that you founded in 2018, you put your quant skills to work and you help marketers with marketing research, analytics optimization, cross channel strategy development, digital marketing transformation, and then more recently with AI and advanced analytics strategy. And if you go to the 3 Standard Deviations website, the first thing you see in big letters is I help people and organizations navigate the discomfort of ambiguity and uncertainty with AI and technology. So my question is, I do have a question. What does it mean to navigate the discomfort of AI and technology? What are people and organizations struggling with right now?
- [06:45]
Cecilia Doness
I think there is always a challenge between what is and what you want it to be. And so I think part of the challenge in the ecosystem is there's a lot of excitement, there's a lot of hype, there's a lot of dynamic movement happening around AI. And so we can't stop talking about AI. We have to keep talking about AI. Why aren't we all AI ing things? AI is a verb. And I think the challenge there is we start inflating our expectations about what the technology and the machines can do. But the reality is, well, we're actually earlier in our data journey. We're actually earlier in our analytical journey. Actually. Our commitment to our consumers and what our consumers expect of us doesn't necessarily require that we AI everything yet. Maybe we just need to deliver good, consistent product. Maybe we just need to reinforce our brand equity. Maybe we need to just reinforce our brand DNA. And this is where I think sometimes, especially in marketing land, we sometimes trip over ourselves. We're trying to do our best for our consumers, we're trying to surprise and delight all the time. And so we get caught up in some of this excitement of AI when the basics, brand management basics, how to build a brand, how to do, how to get the right message to the right consumer at the right time in the right context. Those basics are fundamentals and if we continue to focus on those, layering on the technology can be a great amplifier, but it shouldn't be a distractor. And I think for many organizations at this time, it's a distraction as opposed to an amplification of what you're already trying to do in marketing.
- [08:32]
Alison Schiff
And there is such a tendency to overcomplicate in this industry. And as an aside for our listeners, listeners, I'm sure everyone knows, but I mean, three standard deviations, it refers to the idea in statistics that nearly all data, the vast majority, like 99.7% of data, falls within a range where a value is three times the standard deviation away from the mean I looked that up. So only a tiny amount of data, 0.3% falls outside of this range and that those are the most extreme outliers. And that really was a struggle for me because I am not a quant. I have to use a calculator twice in restaurants, once to calculate the tip and then assess second time to add the tip to the total.
- [09:13]
Cecilia Doness
But actually, here's a fun fact about me. I'm actually not that good at mental arithmetic.
- [09:20]
Alison Schiff
Ah.
- [09:21]
Cecilia Doness
I would also need a calculator, however, derivations and proofs and all the fancy maths. Yes, I can do that quite well. But knowing how much to tip is always a challenge.
- [09:32]
Alison Schiff
I like that fancy math, but I have a feeling I know why you called your advisory three standard deviations. I feel like it's a reference to the fact that, yes, applying data science to marketing can be complicated. Maybe it involves fancy math, but it's not necessarily as complicated or as like special snowflake, I think, as some people think it is.
- [09:58]
Cecilia Doness
Yes, I think there is a lot to learn from other industries and other disciplines that sometimes only speaking to marketing folks or only speaking to marketing technology folks can be limiting when it comes to creating solutions to kind of novel experiences in that sector. And so I think one of the values that I bring when I do have consulting clients or even just regular conversations with others in the ecosystem is that I do my very best to actually bring in perspectives that are from other sectors or other functions. So what does a CFO's office think? What does a CIO's office think? In marketing land, CPG is a very, very large constituency. I always try to think about, okay, what would healthcare think? What would finance think? What would manufacturing think? What are these other sectors thinking about when we think about marketing and speaking to consumers and using technology to mediate that communication? And so I think when you bring diversity of experience and diversity of an openness to thought, I think that opens us up to additional opportunities to find creative solutions to challenges that we're facing.
- [11:23]
Alison Schiff
What would be an example of that? Something real.
- [11:27]
Cecilia Doness
So I do have a newsletter on LinkedIn. What is a CMO to do with AI? And the inception of it was actually people would ask me questions, what do we do? What do we do? What is this? Why are there so many acronyms? And so I started to just write down the answers and hoping that that could help people potentially trying to navigate the space. And again, everything's moving so quickly. So what was true last year is probably not true this year. And what I've been finding lately and most resonant and where I get the most interaction in terms of the articles that I'm writing and the newsletters I'm writing is when I bring in ideas from other sectors. So for example, I think this past week or so I was writing an article about different payment types. So principle based trading sounds very fancy, but it's a new way that the ecosystem is trying to adapt to changes in how everyone makes money essentially. And so it's a new way of thinking about it in this sector. However, this is actually pretty old hat in other sectors. So we can learn things from like healthcare, where they've figured out a way how to share the risk when it comes to different trades between the buyer and the seller in a different way. And those are formats and principles that we could apply in the advertising space to potentially create new ways of buying and selling that is more fair and more equitable in the ecosystem. In the finance space, which is heavily regulated, they are very good at around dealing with transparency and dealing with the challenges of opacity inside of their ecosystem. And so because of that, they're good at explaining things a lot and being as transparent as the regulatory requirements need. That's something that advertising, especially digital advertising, could learn from. So there are already solutions in place in other sectors for challenges we are facing now in digital advertising. And I not saying you can't copy paste homework, please don't. However, we can take the principles of those ideas and adapt it to the context of digital advertising. And so that's where I get very hopeful. Meaning, wow, things are moving so quickly. Wow, there are new problems left and right. Oh goodness. My friend just got laid off. Oh, this part of the business just closed. Oh, there's so much change. So much change. The good news is while we are experiencing this in ad land, other sectors have also experienced something similar. And so we can learn from them. So we're not alone in this. And I think another thing that I tend to bring up quite often is that technology, and we're labeling technology AI right now. But technology is a disruptor. And because it is a disruptor and it's at this point a fairly ubiquitous disruptor, all sectors are dealing with these changes. And so for those who feel lost and confused and maybe overwhelmed, we're all in this together. And so as long as I'll put my AIFX hat on, as long as we continue to treat each other with human dignity and treat each other with compassion, understanding that we're all figuring this out together, I think we'll find solutions that benefit everyone in the ecosystem. It just takes time.
- [15:08]
Alison Schiff
Well, when, when a marketer comes to you feeling confused and uncertain about how to navigate AI and technology, I know they sometimes ask you questions that make you roll your eyes, perhaps internally, you know, like, how do I do AI? Or how do I do agentic? But I mean, as if AI is a one and done thing that you do, like visiting Europe for two days and saying you did Europe. But when they come and ask you a question like that and you peel the layers, like what's actually on their mind.
- [15:44]
Cecilia Doness
Right.
- [15:44]
Alison Schiff
I almost imagine like it's a bit like therapy where someone comes in, they come in and they ostensibly want to ask you about one thing. Like they're, they want to do X, Y or Z. And then you realize through talking to them that you find out, oh, they actually want to achieve this thing. And so this technology is in service of that KPI or that goal. Or you might uncover other underlying challenges they have in their business that's preventing them from doing what they want or need to do. And they read an article that says they need to do AI. So now they're asking you how they do AI. So yeah, when a marketer comes to you and they're all confused like that, how do you help untangle these uncertain feelings?
- [16:28]
Cecilia Doness
So I tend to do the five whys to try to get underneath all of it. But another framework that you could use that I find also quite helpful. When we think about motivation theory, we're always trying to align incentives to motivations to behaviors. And so typically when they're saying, oh, well, I have to do AI, okay, help me understand. How are you incentivized to that? How does that help you in terms of your end of year review or your bonus or your opportunity, next opportunity inside of the organization or outside of the organization? Help me understand. Oh, well, maybe not, but for this meeting with somebody that I need to demonstrate something, I kind of need to do this. Okay, help me understand, what department are they in? What are their incentives, motivations, and what are you seeing as their behaviors when they interact with you? And then that's when you start to better understand. I know we like to say, oh gosh, we have our KPIs, we have our OKRs, and that's our guiding light. I think that tends to oversimplify the challenges people have working in ecosystem, working in networks. It's always a collective actor problem in the sense that there are power dynamics that may be implicit inside of the incentive structures, inside of an Organization, there may be information asymmetry by design inside of an organization that this person may or may not be aware of. And so those kind of underneath subcultural things that can prevent somebody from executing on a creative idea to use AI to surprise and delight a consumer if they're not fully aware. Those are the types of questions and inquiries that I kind of dig into because that is actually what's going to prevent them. The technology itself. Things are changing so quickly. Improvements are happening every day. Everyone is trying. There is a solution out there. However, the people side of things, the procedural side of things, the process side of things, the organizational culture side of things, that's the part that's a little bit more tricky to navigate for people, for marketers, and even outside of marketers, other functions. When they do come to me asking me about technology and how do we implement and do these changes and drive culture. It really is typically organizational psychology that tends to help a lot. I would also say sports psychology tends to help a lot. Also trying to coach and motivate and mentor. And then I would say sociology tends to help a lot. Meaning understanding that you don't work in a silo and that not everyone's KPIs and OKRs are exactly the same and they shouldn't be by design. Understanding those kind of nuances can be very helpful when you're actually trying to implement and keep it sustainable, a new technology like AI.
- [19:35]
Alison Schiff
But where do, where do ethics come into it? Data ethics and AI ethics. And how should a marketer approach ethics as part of their online advertising strategy? In a world where I think a lot of bad behavior has become almost normalized, or if not normalized, it's expected, like in ad tech, people expect bad behavior.
- [20:01]
Cecilia Doness
Yes. Not necessarily my argument, but yes, I can see the argument for there's an expectation of bad behavior. I think the challenge is there is a distinction between ethics and morals. So the first part is ethics is really okay. What do the regulators say? What does law say? How do we stay compliant? And it's kind of easier in a sense because all the rules are written down. And so if you choose to bend the rules, break the rules or follow the rules, that's a little bit more of a binary kind of problem. However, morals is down to the individual. And so what is okay for you may not be necessarily okay for me. And I think that's where people have to make individual choices around am I okay with this or am I okay with this? In the context that my organization is doing it and I used to Be a little bit more pointed when trying to make that distinction. But over time, I've softened a bit on the stance because I find that every time you come across a person who's had to make a very, very difficult decision when it comes to ethics and morals, there's always context that you don't know. Every single person is always fighting a battle in some way that you will never know about or never have the full context of. And so extending compassion for individuals that maybe we want to label may have acted in a bad way. Maybe it's not so bad. Maybe it's more that they were reacting or responding in a way that made sense in the context of the information they had at the time. So I tend to be a bit more compassionate about it. I think the challenge for organizations is that the regulatory environment is not going to catch up anytime soon. And so there has to be a proactive management of ethical risk and moral risk if you want to go down to the individual. And what that means is when you think about reputations, your brand has a reputation. How much are you willing to risk with a faux pas? How much are you willing to risk with some kind of salacious news? For some organizations, the equity is so strong, or they believe the equity is so strong, or the product itself has such high utility, they're resilient. Maybe they care less. For some organizations, maybe their equity isn't so strong, maybe the competitive environment is too much. And so maybe they are more sensitive to these kinds of risks. The challenge is there isn't a one size fits all. And I think that's the biggest challenge with AI ethics. We do not have a universal. This is the one rule that means everything is equitable, everything is fair. We don't have that joint value globally at all. And because we don't, that means the application of ethics becomes regional. And so that makes. That begs the question, when we think about the US versus the eu, for example, versus China, for example, what values are we reinforcing through our technologies in those regions? And so when I think about ethics, I'm always struggling with the tension of global versus local, because there's local culture, local values, local rules that may not be translatable globally.
- [23:46]
Alison Schiff
And there are also what you could think of almost as local rules, depending on the industry, rules that seem to apply to a certain kind of business operating in a certain kind of ecosystem. I'm thinking of, of ad tech. And I want to talk about a piece, that really interesting piece that you published in early march on LinkedIn. I'll read the headline now. Then we're going to take a break and really dive in. But your headline is AI Blind Spots, why Programmatic Ad Scandals Persist and How Marketing Leaders Can Reclaim Control. I know you got a lot of feedback that on LinkedIn, so stick with us and we'll talk about it. I'm Alison Schiff, Managing editor of Ad Exchanger. And with me I have Andy Frawley, the CEO of Data Axle. And I have a few questions for him about some burning issues. Hi Andy.
- [24:55]
Andy Frawley
Hello Alison. Thank you for having me.
- [24:57]
Alison Schiff
So it's no secret that many businesses struggle with fragmented data spread across multiple platforms. This is something Ad Exchanger covers all the time and this leads to inconsistencies and inefficiencies. So how can organizations break down these silos and create a unified data foundation for better marketing and better business performance?
- [25:17]
Andy Frawley
It's a great question, Alison. This is one of the burning questions that the data actual we see within our customer base. And while a lot of brands have made progress on linking legacy data together from their operational systems, what's happened over the last really five years is we're seeing this massive new set of data that's being generated which is the exhaust of all the digital advertising platforms. And so linking all that data together with third party data with of the first party data really requires, requires an identity spine. Historically, brands have relied on third party identity graphs to do that work. The trend we're seeing is that brands would like to own that identity spine and so help build that identity out with third party data, with first party data and have a spine that links the known to the unknown. Obviously in a highly compliant fashion.
- [26:12]
Alison Schiff
Yes, always in a highly compliant fashion. That's very important. Well, marketers are shifting from vanity metrics to outcome driven strategies. How can brands use cross channel analytics and AI to ensure real business impact in a changing digital landscape?
- [26:33]
Andy Frawley
We certainly recommend to our clients to focus on business outcomes, whether that's new customers, more customers, more profitable customers, customers that stay longer and, and really move away from the vanity metrics of opens and clicks. So there's two important concepts when we, when we think about this. One is incrementality. You know, is the marketing effort creating incremental outcomes I.e. sales and causality. Is it the, the marketing treatment that's actually causing the, the consumer to buy or act in a certain way and the, the crush. Analytic tools need to embrace both of those concepts. You know, the complication or challenge is somewhat. Back to the first question. First you have to be able to link a lot of data together. And second, there'll be, you know, places where you have sparse data, where you don't have complete data sets. And so we're also seeing people using Gen AI to help generate those customer journeys and have a, you know, analytically based view of what the media exposure is across multiple marketing treatments, down with.
- [27:38]
Alison Schiff
Data silos and down with vanity metrics.
- [27:41]
Andy Frawley
Absolutely.
- [27:42]
Alison Schiff
Thanks for the insights, Andy. All right, we're back and we're going to talk about that piece that you published. So there's a disclaimer that you put on it and I want to read part of it because I think it's quite interesting. So you wrote. I spent a lot of time thinking, writing and teaching about the ethical implications of AI. It may seem unrelated, but the persistent problem of ads appearing alongside harmful content in programmatic advertising should give anyone working with AI cause for concern. We cannot champion ethical AI development while allowing it to inadvertently prop up harmful content or perpetuate opaque practices. And those feel like self evident truths, right? Like it's time for the shenanigans to stop. Even though shenanigans is way too cute of a word, honestly, for some of what we've been seeing. So I'm going to ask you a question that you pose in your own piece. So with all the ad fraud and the wasted ad spend and the analytics report on child sexual abuse material getting monetized, and those are just a few examples, why does this kind of thing continue to happen despite widespread outrage and awareness? Because it does seem in part at least like a problem of incentives. But is there more going on there?
- [29:19]
Cecilia Doness
Incentives is the primary, I would argue. And then I would say it's the challenge of, I'll put it in maybe overly simplistic terms. There's just too much money to be made. And so the funny thing I find so interesting about the digital advertising space, and I am native to the space and so I still really love it. And as we alluded to, I am mobile first. So digital ads were always first for me anyway. I think the challenge is it's ballooning budgets into digital spaces, but there's not enough eyeballs, there's not enough devices, there's not enough attention to be sold. And so when you have a surplus of capital flowing into these ecosystems, what ends up happening? You start to put resources in places that maybe could be exploited in ways that maybe are unpalatable, but it is ways that people are making money. Whenever you have a situation where there's intermediaries, anytime you have a situation where there is a lack of transparency. And transparency is not binary. It's not one or zero. It's just some places are a little bit more opaque than others. You're going to have players that try to exploit the system to in this case make financial gain. And so as long as the money continues to flow that way and we do not create new incentive structures that reinforce pro social behaviors that we want in the ecosystem, this will continue to happen. So we can try doing sticks and with stics, that could mean regulation, fines, these kinds of reports that kind of call people out, which is un unpleasant but you know, facts are facts and therefore we have to move forward from there. Or we can try new carrots or maybe cake, maybe less so carrots, but something more positive. So for platforms that really do hold accountable, making sure that this kind of content doesn't flow through through four networks that kind of provide these guarantees and so they' sharing the risk burden with the brand. Ah, that's a new incentive. Or for brands that make public statements to say they will have zero tolerance on this and hold everyone accountable in the ecosystem, including themselves for this. That is a new way to create a new incentive in the ecosystem. So the, the article is really saying it's, this is where it gets complicated. I don't believe in good guys and bad guys. I believe that individuals are driven by their incentives. And so if we change the incentive structure, oh, that changed the motivations. Ah, that changes the behavior.
- [32:24]
Alison Schiff
Well, how do we trigger that though? Because you put it very well in the piece. You say no player unilaterally adopts rigorous due diligence as competitors benefit from cutting corners. Right. So everyone would have to make this change at the same time or, you know, good actors lose, nice guys finish last.
- [32:45]
Cecilia Doness
That is true. And I think this is where I have to look at leadership. In every case that I've ever joined an organization, a team, a project, I've always looked to see what would the leaders do, what can I learn from these leaders? And if there was a word that I would encourage all of us, myself included. And I think maybe this article is a little bit trying to be like that. We have to be a bit more courageous. And that means sitting with the uncomfortableness of what has happened, but not getting stuck in it and moving forward. So it means having courageous leaders willing to take a stand. And that may have unintended consequences, both good and bad in the ecosystem, but it requires first a courageous leader taking action. So this is why I think the work that the firm analytics had done was that's a courageous move. Not everyone would do that. I think also the work that the ANA has done in the report that I cited in my article from a few years back around AD spend being wasted, that also is courageous work. I think the individuals that do this reporting work, this research work to help hold the ecosystem accountable for what's happening and create light where there is darkness. I think that is part of the courageous leadership we need. And then once you know, you can't unknow it. And so once you know that your brand has been exposed in a way that is risky, what are you going to do with it? And that's where every individual leader has to make a decision.
- [34:35]
Alison Schiff
It does seem like, and not to put the burden on marketers, but they really could make a big change, catalyze a big change by voting with their wallets. Because there is a lot of anger and disgust and how could this be happening? And my agency didn't tell me and I'm so mad about this. But then spending habits don't necessarily change as you would imagine they would after really bad revelations.
- [35:02]
Cecilia Doness
This is where it gets a little bit more complicated because I also recognize that for some organizations, not all. For some organizations, regardless of your function, your budget size also indicates the amount of influence and power you have inside of the organization. Sometimes it's team size, sometimes it's budget size, sometimes it's portfolio that you cover. And so by choosing to take a courageous position or a courageous action, that can also be, that can also cause unintended internal consequences for that individual. So I recognize it's a very difficult position to be in, so you could risk yourself internally while outwardly you're taking a courageous action. And so that's where again, I extend so much compassion and sympathy towards leaders who have these tough decisions to make. I also think consumers, they can totally vote with their wallets too. And last time I checked, there are more consumers. And so if consumers don't want to be assaulted with, visually assaulted with some of these advertisements in places that are out of context, or consumers don't want to support kind of some of these nefarious dealings, then they can also say, hey, I'm, I'm not going to be on that platform. Hey, I'm not going to look at this. Hey, I'm going to switch whatever tools I use. So I think there's opportunities to take action at every part of the ecosystem. What the article alludes to is just making that all happen in a coordinated way. It's a bit of A tough challenge, not impossible, definitely not intractable. But it is tough.
- [37:01]
Alison Schiff
And changing gears a little bit, but still talking about incentives, this is something you brought up with me when we talked at the end of last year that there's no incentive in the online advertising industry. And this is not. That's only true about online advertising. But there is no incentive in this industry to be efficient. And I mean not, not to be super cynical, but you alluded to this already. I mean the purpose of adtech is to keep the dollars flowing and maybe keep marketers thinking that their spend is having an impact regardless of whether or not it does. You still just want your budget next year to spend. Ada companies want that budget. Lather, rinse, repeat. And as long as everyone gets their cut and doesn't get too greedy, then the status qu can just chug along for a, for a very long time. Maybe not in perpetuity, but I feel like brands should be more annoyed, like very, very annoyed about the inefficiencies in the programmatic advertising ecosystem because it's their money and it's their money that's being wasted by inefficiency.
- [38:09]
Cecilia Doness
Yes, I would be very, if I had my magic wand and I was a CMO and I was very worried about my line item being spent in advertising and that's media, creative content, all the things I would be very curious to see what's my working to non working ratios. And that's what I would be encouraging all of my teams to monitor and make sure that as much of it that is working is facing a consumer, a real consumer. However, I do recognize this is an ecosystem, there are intermediaries. The buy side sell, sell side is very complicated because everything is digital and so you're never going to have a hundred percent working to zero percent non working. So there's always going to be a little bit of overhead. I guess where I would be very curious to challenge that idea is I know we want to say efficiency because with technology doesn't that just make everything more efficient? Therefore we can scale. Isn't that a wonderful thing? So we can keep growing bigger and bigger. What I love about marketing and why I was drawn to marketing as opposed to some of the other functions. So my undergraduate I went to stern. I could have easily become an I banker. I chose very much not to very very quickly. I went straight into marketing very fast. Part of the reason why was there again surprise and delight. There's something about knowing your consumer and being able to deliver on their needs and wants and relating to them and having a connection with them that I found so fascinating about the function that got me very excited about the space. I would argue maybe we talk less about efficiency and maybe we talk more about innovation because I think that's a bit more positive leaning. And I think that's where we take the best of what technology can produce and bring and focus the energy around innovation, which would create new experiences, incremental value for consumers, new ways to maintain and grow that connection with consumers, as opposed to how can I trim the fat? I think trimming the fat will get us into a race to the bottom. Yes, it's going to be beneficial to anybody in the ecosystem.
- [40:46]
Alison Schiff
We've been racing to the bottom for a while. It's time to slow a roll and. Yeah, and innovate. But we do have to make sure that our innovations are actually innovative and serve us. So another thing that you brought up with me was a very interesting conversation we had back in December is that we're seeing human behavior adapt itself to the constraints of machines as opposed to the other way around, which would be machines being there to serve us, which is what I think most people believe they're there to do. But we are complying with the needs of machines rather than creating machines that better comply with our needs. And it happens so quickly. And I think you gave the example of a slack message at 10pm like you feel compelled to respond because machines are not set to our circadian rhythms. But that wasn't always the case. And not only that, we can be reached at any time, not just by our friends and family, but by media messages. There's not really a time when you're not reachable, like even in the middle of the night. And I was just thinking to back in the 1920s, like the late 20s and the early 1930s, when TV was first invented, like it did not operate on a 24 hour schedule. And there were lots of reasons for that, technical limitations, maintenance needs, lack of content, economic considerations and whatever. But that was also a time when TVs went to sleep and people went to sleep and you couldn't be reached with messaging trying to get you to do something or buy something or whatever in the middle of the night. So I don't know, that was a little bit of a ramble, but it makes me think a little bit of AI and technology. AI isn't new, even though some people talk about it like it's new, but it is newly becoming part of every conversation, part of the culture. It's how people seek information. So how do we make sure that our AI systems and our conversational chatbots and our agentic whatevers, that they're not being developed in a way that complies with, you know, not our needs. I want them to be developed in a way that helps us do better instead of sort of conforming us to something else.
- [43:04]
Cecilia Doness
So I'm a big fan of the framework, human in the loop. So when it comes to development, deployment, any kind of long run type of situation, making sure there's always people involved. So I am not a fan of complete set it and forget it. Because you always have model drift, you always have data drift, the context may change and yes, agentic systems can adapt somewhat to those changes and that's what makes them fun and fantastic. However, at the end of the day, we are, until I am proven wrong, which could be in the next minute or so in, in Internet time, I don't see AGI. And because I don't, we still need people to ultimately be the arbiter of, okay, is this a good thing to do or not a good thing to do? I'll give you another example about the how machines are kind of changing the way we behave as opposed to adapting to us. Something I also find very funny about the AI space is, gosh, maybe it was a year ago at this point we got all very excited. Everyone needs to learn how to prompt engineer. That's the newest job, latest job, just prompt engineering. And I agree, yeah, we do need to be able to know how to interact with these new LLMs. Great, fantastic. But I was taking a step back and I was thinking, wait, the machines are so smart that we have to learn how to reconfigure how we type to them and communicate with them because they're so smart? Does make sense. It sounds like we're changing our behavior to adapt to the limitations of the machine. I say this because as the technologies continue to develop and there's a new behavior that we all have to adopt, I, I personally do this, I, I take a step back and say, gosh, is this behavior change actually beneficial to me or is it because I'm trying to make this machine work? And if that's the case, then I need to put my humble hat on and say, okay, the, it is still a machine, it's still a tool in my repertoire of things that I can use to create value in this world. And therefore I need to always keep that in mind. And then also as a personal thing, we all have to create our own boundaries around our technology utilization. Always being plugged in, probably not a good idea. Never being plugged in, also probably not a good idea. So what is the right balance for you in that moment? In that context? It's up to the individual. But being aware of it, I think is like the first step.
- [46:04]
Alison Schiff
I'd love to know what AI tools you use in your daily life and your work, what's been useful to you and how do you use it.
- [46:15]
Cecilia Doness
Okay, so I really am a fan of canva. I use PowerPoint for a very long time, but I do like Canva for graphics. That's, that's pretty common for me, I would say. Perplexity, I do appreciate. So I was kind of a fan, going on and off, kind of switching between all the, actually all the different platforms. But when R1 came out and Perplexity also had the capability as well and did their own wrap around it, that's when I got really excited. And the reason why I got excited was again, I'm AI ethics. I'm about explainable AI, understandable AI. So not just explainable, it has to be understandable to me. And so having AI tools that can kind of help you understand, oh, what was the logic of their process that is so helpful. You can see the errors and assumptions that they're making, but you can also see the errors in your logic as well. So I find that kind of explainable type of capability to be very, very helpful. And I use this example all the time. I use email. I don't click on spam emails. That's machine learning. So by definition it is technically AI.
- [47:45]
Alison Schiff
That is a good point. And who among us has not hit that button?
- [47:50]
Cecilia Doness
Exactly.
- [47:52]
Alison Schiff
So we're nearly out of time, but I wanted you to leave us with this. It's a two parter. I'd love an example of just the absolutely coolest, most interesting application of AI to marketing that you've seen recently that really impressed you. And then the flip side, what is the BSE est most ridiculous thing related to AI and marketing that you've heard or seen recently that made your eyes just roll up into the back of your head?
- [48:23]
Cecilia Doness
I'll give you a mixed bag answer. So I actually posted about this. Today's Tuesday. Okay, I posted about this today. The Studio Ghibli thing.
- [48:37]
Alison Schiff
Yep.
- [48:38]
Cecilia Doness
There are moral, ethical, privacy rights, you know, IP rights, all those things, questions around it. However, the capability to mimic a specific style so well and at scale may put it in a different context, maybe a different creative context. If a brand was able to take their core corpus of brand assets and Kind of share it in such a way that allowed people to be creative with it in, in the way of Studio Ghibli. I think that would be quite exciting because the machines can do that now and that would be a surprise in the light and everyone would be excited. That specific case is a little bit mixed bag because again, of the ethical issues associated with it on the other side, where it's a little bit of hashtag cringe. I was reading not your organization. I was reading a different industry publication and there was a firm that was excitedly, excitedly talking about how they were utilizing AI to generate copy for SEO. And so they were saying, okay, this is so fantastic because we no longer need a director to really kind of manage all of this. And it's so much savings. With 30% savings this, the brand can see themselves climbing up in the rankings and we save on time of staff and we move faster and all these efficiency plays. And they were describing and how the improvements of the Google algorithm was responding to the machine producing all of this SEO content copy. And I was thinking, gosh, this is a machine advertising to a machine. So we're clapping because we're having machines advertise to machines and we've completely removed people from the process. And so I thought it was a very odd moment and I'm going to put it out there into the universe. The textbook that I have yet to find, but I would love to read one day or either contribute to is AI Marketing for AI. So when machines are selling to machines, how do we do that?
- [51:05]
Alison Schiff
You might have to write it.
- [51:07]
Cecilia Doness
Okay, well, I'm very collaborative. I'm very open to co authors.
- [51:12]
Alison Schiff
Cecilia's DMs are open I guess. So if you want to contribute a chapter, you should reach out. That's a wrap. Thanks for listening. And a special thanks to our sponsor, Data Axle. Helping businesses create deeper, data driven connections with customers. Whether you're reaching new prospects or strengthening existing relationships, Data Axle delivers the insights and solutions to make every interaction more impactful. Explore more@dataaxle.com.