Episode 17: Reshaping Marketing in the Era of AI and Data Privacy

Christi Olson, one of the world’s foremost experts on Search Engine Marketing and current Interim VP of Digital Marketing at Azul, discusses the rapidly evolving landscape of digital marketing in the age of AI and privacy regulations. Drawing from her extensive experience at Microsoft, where she led Global Paid Search managing over $500M/year in spend and served as Head of Evangelism for Bing, Christi explores the transformative impact of AI on marketing practices. She covers the importance of adaptability in modern marketing teams and the delicate balance between leveraging AI for efficiency and maintaining human creativity in content creation.

The conversation covers critical topics such as navigating privacy regulations, the shift towards first-party data strategies, and the changing dynamics of digital advertising. Christi offers valuable insights on media mix modeling, the challenges of automated advertising platforms, and the crucial role of data strategy in driving marketing success. Throughout the episode, she provides practical advice for marketers looking to thrive in an increasingly automated ecosystem while maintaining strategic control and fostering innovation within their organizations.

Listen & Subscribe:

From Microsoft to Azul: Christi’s Journey in Digital Marketing

Aaron: Tell me a bit about Azul. 

Christi: Azul is the only company that’s 100% Java-focused. We’ve been around for a little over 20 years now, focused on Java software. We are the biggest competitor to Oracle for any company using Java applications. We are one of the choices if you don’t want to use a contract with Oracle. You can come to us, and we’re on the Open JDK. We have support, and we build out our own set of features to help companies get better, faster, everything. 

Aaron: Tell me a little bit about your history, your trajectory, and how you came to be at Azul. 

Christi: I really got my start in the digital space when I got hired at Microsoft, and I owe a lot to Microsoft. When I talk to a lot of different people about their careers and ambitions, everybody’s like, “Oh, you have to go to an Ivy League school.” I went to the University of Idaho. I’m a proud Vandal. It’s interesting because people think, “Oh, you have to have this very high education.” It’s like, no, you know what you can be in your career? You can be focused. 

I studied Marketing at the University of Idaho, international business. I knew I wanted to go into something within marketing, and I fell into digital, sort of seemingly randomly. I started in sales in a completely unrelated industry. Through my sales, I met somebody who had an agency in their kitchen, and they needed to sell the product, but they couldn’t figure out how to sell search. They had me on my side, just helping them sell their business. I thought, “This is sort of fun. I want to do it on my own.” 

From there, I managed to get an interview with Microsoft. I had maybe six or seven months of search under my belt and got into Microsoft, into a brand new team managing paid search. I really just had a lot of grit, because if I didn’t know it, I wanted to learn it. I wanted to learn what to do, how to do it, how to track it. Throughout my career, I’ve had that mentality of, you know, I might not know how to do it now, but I can learn. I can be taught how to do something new and different. 

So I’ve been pushing myself to learn a lot of different things about digital marketing, advertising, how things operate and work, and really going deep to understand analytics, tracking, and how that impacts what we do from a digital perspective. I went from running a small team at Microsoft to running an internal agency at Microsoft, to working at the launch of Bing and running search on Google for Bing to promote Bing—weirdest thing ever—to working in Windows. 

After about six or seven years at Microsoft, I left and said, “You know what, I want to get experience elsewhere.” While I really appreciate Microsoft and the amazingness of the culture at Microsoft and the ability to manage the most insane budgets I’ve ever managed in my life—like, the amount of money we could spend was fabulous, and it taught you how to scale—I wanted to take it back and go into a niche. Could I go back to a smaller company and figure out how we then make it work in an industry niche? 

So I went over to Harry & David and managed SEM, SEO, and syndication programs. Again, taking a broad swath and narrowing it down into not just paid search, but then adding organic, adding in syndication, understanding the different programs, how they all tie together, and working with our social teams. 

I went over to Expedia, and I actually got out of search. What I was doing was called metasearch. It was figuring out, as TripAdvisor became a paid platform, how to take all the concepts of how you ran the programs pre-Google Ads, Microsoft Ads days and apply them to the travel space. We were building out the tools, the platforms, the practices to do what we did in paid search, but for hotels and flights, which was really fun to get to go back to the days when search was just getting off the ground and apply it to a different industry and space. 

From a career perspective, I thought I might want to start my own agency. So I went to a small agency here in Seattle, worked with a ton of different clients. Loved it, but decided, since I wanted to start a family, agency life was not for me. Agencies are fun. It was so exciting to get to work with so many different lines of business, so many different business types, to again round out not just B2B, B2C, but understanding how their businesses operate. 

Then I was very fortunate to get the opportunity to be the search evangelist for Bing. I had been waiting in the wings for Duane Forrester to leave his job for so many years. Duane, I love you, but thank you. They were able to open up the role, and so I got in as the search evangelist at Bing for Microsoft Advertising and Bing to just talk about search, the platform, take all the previous 10 years of experience of managing digital campaigns and search campaigns, and then talking to everyone from agency owners to businesses about how they should be thinking about search in their marketing mix. 

It was less of the marketing execution side, more that high-level strategy. How does it fit within businesses? How does it fit within programs? How do you think about not just Microsoft advertising, but Microsoft’s entire suite of products and services as part of your business and what you’re doing to accomplish whatever your goal is, whether it’s selling to a person or selling to another business and lead gen? 

I also got a chance to dive deep into topics that were really emerging at the time, like Voice Search, which still hasn’t quite played out as much as I was really hoping it would. But AI, that fundamentally is the backbone of so many things we see today that people are taking for granted—they don’t even realize. It was sort of that burgeoning beginning of AI just hitting the mainstream and really understanding what artificial intelligence is and how it fits within our marketing mix, and how it fits within marketing tools and platforms. I got to explain that to businesses and also to marketers who may not have understood how AI fit into the search algorithms at the time. 

Even yesterday, I sat down and was explaining vector mapping, keyword vector mapping to our leadership team, whose eyes were glazing over. I was trying to explain like, “No, this is why these terms are relevant, even though they’re not relevant. It’s a vector. It looks like this, but if you look at the side, they’re in a line, but they’re not. They’re similar, not the same.” 

From there, I was an evangelist for several years. As COVID hit, I decided to leave being an evangelist. We had some personal stuff with my daughter and her health condition. I decided I didn’t feel like I could really fulfill the evangelist role to the ability that I would want to feel successful in the role. My boss was amazing, like, “No, we want you to stay.” I said, “I just don’t feel like I could do this role justice anymore because I need to take a step back from being as public-facing and spend more time with family, with my daughter’s condition.” 

So I went into the global media team and stepped into even larger budgets. I think I’m trying to remember what I’m allowed to publicly say. I can say we did over 500 million in paid advertising that was running through my team of six people to support all of Microsoft’s lines of businesses. So a good-sized budget, just a little bit. We had a lot of fun because the team had existed for a while, but we were really building out processes and practices. What do you do across Microsoft? Because we supported any business, any line of business at Microsoft globally that would like to do paid advertising. 

When people think, “Oh, your team must have been huge,” I’m like, “No, not really.” We had some people dedicated to very specific lines of business. For example, when you think of the B2B side, like Azure, we had one person who was pretty much solely dedicated to Azure because of how big that line of business is and how much revenue potential it had for Microsoft. They specifically were dedicated to the B2B side for Azure. We had another person who was solely dedicated to Microsoft Store. So we divided our team by audience and who we were targeting, and everybody went really deep. 

After this last year, I made the decision that I was ready to leave Microsoft. Again, I had a lot of fun running media and running paid search, but I wanted to dip my toes into marketing as a whole. Similar to the agency side, where you get to deal with so many different clients and perspectives, I wanted to do something that was more than just paid search—search, social, display, syndication, email marketing, lifecycle marketing. How do you do everything, end-to-end? 

So now I’m the Interim VP leading everything on the digital side at Azul. I have very strong teams that partner with me for partner marketing, for field marketing (so events and industry), for Product Marketing. We have a person dedicated to content, so our AR, our content team that we all come together to really figure out how we go deep and broad with a tiny team. 

Yeah, it’s been a fun journey to where I’m at today. Starting where I had very little experience when I began in search, but just had the desire to learn. This last year, I didn’t know almost anything about Account-Based Marketing, and I am now definitely the expert of the company on everything you need to know about how to use platforms, intent, the data, the systems. How do we log in? How do we create dashboards to empower and enable sales? How do we create dashboards to empower and enable us within marketing, or our lifecycle team, the social team, to know what to do and when? So it’s been really fun to go, I’d say, 16 inches deep and a mile wide, not just an inch deep and a mile wide across marketing, which is very similar to what you do at an agency. 

The Role of AI in Scaling Marketing Efforts

Aaron: That’s exactly right. Well, this incredible background has given you, I suspect, a unique and richer perspective on AI than most people have who just started paying attention when ChatGPT was launched. So as we think about AI, which is in the zeitgeist, and we think about the implications of AI for marketing, how are you using AI to scale, and where are you seeing AI really deliver from a marketing perspective? 

Christi: Whether people know it or not, as you just alluded to, AI has been embedded in so many of the platforms from an advertising perspective since the early 2010s. It’s just gotten so much smarter through machine learning, deep neural networking, everything. And now we’re starting to see the fruits of it. 

From an advertising perspective, AI’s been there forever. It’s partially what I talked about with Microsoft advertising in terms of the targeting capabilities and like smart bidding and the optimization side that’s automatically running on the back end. They see so much of the performance data on the conversions, and they have access when you have tagging on your site to information, to make decisions better than I can manually. I love that 20,000 data points are way better than me being like, “Ah, increase the bid 5% here.” It’s already embedded in the platform. 

So then on the marketing side, as a marketing practitioner, people are now taking a step back and saying, “Oh man, I do so many things manually, and I spend a lot of time doing all these manual tactics.” So from the automation and scale perspective, that’s really where the opportunity with ChatGPT came into play. People started saying, “Oh, look, I can see all these people sharing scripts to make my life easier to do different things.” 

I’m going to say ChatGPT—there’s a ton of other tools out there. ChatGPT is just probably one of the most well-known. What I’d say is it poses both an opportunity and a bit of a threat or issue. I spend a lot of time with our legal department going back and forth about what we can use it for, what we can’t use it for, and what are the implications of using ChatGPT and taking data information. Since you’re training a model, the model now has your information in it, potentially, depending on what license you have. Some of it you can block from having your information go into the model, but are you sharing information that might be proprietary or that shouldn’t be made public, to use to write ads, content, whatever you’re doing with it? 

So there are implications on AI that have not, and for a very long time will not be, resolved. Looking at how legal works on things, it’ll probably be another 10 years—I’m probably wrong, but maybe 10 years—before it really hits and we know what is going to happen and what you can do with AI or what the restrictions are. 

Aaron: The conundrum may be that if it takes that long, by the time they’ve settled the questions of today, 

Christi: it’s a whole new set of questions that they didn’t even think about. 

Aaron: Because We focus on healthcare and medtech, we have access to a lot of data that cannot be disclosed to anyone, so it’s all just locally resident here and there, never goes outside of our network, and never is used to train any other model. 

Christi: Exactly. For me, about a year ago, we had a marketing offsite, and I sat down with the team. They gave me an hour and a half to go through what are a handful of different AI tools that I think would be important for our marketing team to be aware of and to consider using in their day-to-day to scale. Because, as I said, for my digital team, I have a social media manager and lifecycle marketing manager. So all of our email programs and social are handled by one person. I’m trying to explain to the organization that you understand that’s normally two jobs, two very distinct jobs, right? 

So if she is trying to write social posts to support every announcement, blog, product campaign, that is a lot. I have empowered her. We got her a license. For us, we haven’t said that you have to use a company license. You cannot use a personal license. So we have said, if you use it, here are the guardrails around what you can use and how you can use it. Here’s when you go to legal to review specific elements depending on what you’re trying to do. 

For social, we use it to help scale our social copy. We put in some ideas and say, “Here’s what we’re trying to do. Give us our social tweets.” We still have to vet it because so many times, it’s not exactly as accurate as you would hope. And there are some things where what it comes back with, we’re like, “Ooh, interesting.” 

Balancing Automation and Human Creativity in Content Creation

Aaron: Yes, it’s curious in that way. It’s like, you know, 80% scale and efficiency and great, and then 10-20% real editing. 

Christi: Yeah, real good editing. So you still have to be a really strong editor. It can help scale so that she can be our lifecycle marketing manager as well as manage our social so that she isn’t bogged down trying to hand-write social posts all the time. But you still have to have that critical eye for reviewing it to make sure it’s accurate and it’s correctly representing your product, and it doesn’t just randomly pull in things that are incorrect. 

There are so many examples of AI-generated content gone wrong. I remember the exact paste example that came out last month where the AI made recommendations that paste could be eaten. So you do have to have that critical eye. 

We met as a team and said, “Here’s how you can use it to scale writing, to edit your writing.” We have people all the time that say, take your post—if you’re going to do a blog, you’re going to do something—take it, run it through ChatGPT. We’ve worked with a bunch of different scripts. There are so many great scripts out there. I don’t know if your team has scripts you’ve published on your blog or website, but I know many different companies have published scripts to edit tone and whatnot. 

We use it to edit tone, make sure that it’s relevant and it might resonate with your audience. So we have scripts to tell us, “Who do you think this is written for?” And it comes back with a bunch of information. We say, “Oh yeah, we wrote this for a PhD. We’re really targeting an engineering manager. So this is way too technical and tactical of a post.” Let’s re-edit and write it. So we use scripts to tell us, are we tone-deaf? Are we using the right tone, right message? Is what we thought we were coming across actually coming across? We still haven’t figured out how to scale it as much as I would like us to, and there’s a lot of time spent writing the scripts, teaching the team. How do you write a script? How do you edit the script? How do you get it to do what you think you’re going to have it do? Versus just if you’re too broad and generic, you don’t scale because you spend more time on the editing back end. 

Aaron: Are you building custom GPTs and sharing them among your team members? 

Christi: We’re almost there, but not quite, because, again, we all are a tiny team at Azul, so we haven’t had time or resources to fully implement that. But I have a library of scripts that we’ve shared with team members. Some of my scripts are saying, “Okay, we know we have these assets that we’re going to be promoting through different channels. Let’s take certain aspects of it, use it to help us edit the landing page copy. Now let’s take it and have it write our display copy, our social copy, edit our paid search text ads.” Then we come back and re-look at it to make sure everything is on tone and brand. 

But I can, essentially, from the point we get the landing page copy done, have all the supporting media ready to be reviewed by our product marketing team within 30 to 45 minutes, which is pretty nice. Yeah, scale. So, we’re taking the different tools and platforms and figuring out how we scale with them. 

I think the area that’ll be interesting to see where it goes, because I’ve seen a lot of different people doing this, is now there are so many AI-generated images. So when you’re running images on your website, on banners, on social media, all the different channels, what is real versus what is AI-generated? What did we have a creative agency create, that they actually created? Or did they go to Midjourney, DALL-E, fill in the blank? 

And we spend a lot of time with legal like, can we legally use them? What can you do or can’t you do? I think in the image side, for images generated, there’s still going to be a lot to come, because a lot of the models are trained off of individuals’ proprietary art styles, right? And that’ll be interesting to see how that comes out. Because you can actually say, I for fun with our legal team, like, “Create a whole series of ads based on Disney, like Disney-style drawings,” and it came back and it looked like Disney characters. We weren’t going to use them, but it looked like Disney princesses. And they were concerned, like, Disney will come after you, because it is 100% their visual style. 

Scaling is a matter of understanding what are the different tools you can use. How can you use them? And then understanding what is the basis of the script, the platform itself, to know how you can set your team up for success to use it. Because there are so many different platforms out there, I think I covered 45 platforms in an hour and a half. I’m like, here’s all the ones we could have. Like, if we wanted to do something similar to audio, we can’t always afford to hire a voice-over actor, right? So there are platforms to let you choose who it sounds like. You can write scripts. You can run it. 

We’re doing a test campaign. I’m trying to think if we’re actually going to use it or not. We have a campaign launching in two weeks where we actually did everything using AI, and we worked with our creative agency and let them use AI. We have videos, scripts, audio components, visual components, all of it. All AI-generated on the back end. It’s the first time we’re doing that. Everything before, we would have our creative agency go and take photos because we didn’t want stock photos. We’d have them go take the photos. So now we’re helping them understand like this is what we have to do. Here’s our legal team’s advice on what we can and can’t do. Stay within these constraints. 

Aaron: As I’ve periodically dipped my toe into the water, I have yet to reach a point where I look at the result and think that’s usable. I continue to think, well, that’s better than what I saw the last time, and it’s really interesting, and it’s novel that I could generate this in three minutes, or something like that. But I haven’t yet gotten to the point that I felt like, particularly around motion graphics or video, the result was good enough to actually use. So I’m excited and intrigued that you’ve gotten there. 

Christi: Yes, well, it’s not me. This would be our agency. Since I’m talking to another agency, I won’t put their name out there, but we have a creative agency we love. 

Aaron: If they’re great, then say their name. 

Christi: Yeah. So we work with an agency out of Texas called Good Folks. They have done an amazing job using all of the different graphical and motion graphic platforms to help them scale, because we have small budgets. I went from over a $500 million budget to, I think this year, my budget is well under $2 million. And that’s before it’s $500 million for just paid search. Now my budget’s under $2 million for everything – our media, our agencies, our tools, technology, everything, soup to nuts. 

So now I think what I’ve had to learn, not only with AI, is like, as you scale down, you also have to learn how to be really tight with your dollars. Because you do have to scale when you have finite budgets and very tight budgets. And so the agency, Good Folks, has figured out, okay, it doesn’t make sense to hire a voice actor, we can essentially use one of the voice GPTs or voice technologies to do this. Some of the motion graphics they’re using, I don’t know if it’s 100% all, some of it I think they’re using like Figma and different tools on the back end to create the HTML5 graphics and pull it together. But the basis of the graphics are all from, I’m pretty sure they’re using Midjourney at this point in time, maybe it’s something else. 

And I will say, the difference I’ve experienced – we were out of hours last month with our agency. We just capped out, and I didn’t have extra budget, so I got to be the display person and the creative person for three weeks. It took me eight hours to create an image that was usable for an ad. And I’m pretty sure he could have done it in 45 minutes because he knows his scripts. They know the scripts. They know how to write them. 

I could not get Google, ChatGPT, Microsoft, DALL-E, any of them to create a Whack-a-Mole looking video game. It could not figure out what Whack-a-Mole was. It kept on giving me real moles. Like, no, I want the game Whack-a-Mole. How else do I describe this thing? And it took me eight hours to finally get a very basic grid, like a checkers grid with ones and zeros popping out and a mallet coming down. But that was eight hours. That was not the best eight-hour use of my time. Luckily, I was watching Netflix in the evening. 

Aaron: That’s my experience with Midjourney in particular. But I feel like the issue with Midjourney, where I’m concerned, is that in order to use it effectively, I actually need much more artistic knowledge, technical artistic knowledge. Because, for example, if I want to create a photo-realistic image, I need to understand the lens to specify, and the focal length and the lighting and all sorts of things that are not innate to me. 

Christi: Well, and I’d say, if you’re on Midjourney, and somebody listening to this podcast is trying to figure out what to do, something you can do specific to Midjourney is look at the scripts other people are writing. The great thing about Midjourney is everything is public-facing, so don’t stay in the newbie channels, because if you’re in the newbie channels, you’re going to see really bad script writing, similar to what I do. But as you start to look at other people’s scripts, you’re like, “Oh, this is how they’re specifying to get these amazing results.” Because you can see crappy images and great images. And so I start to go through and I’ll copy out bits of other people’s scripts, depending on whether I’m trying to get a photo-realistic, stock-type image, or something more animated, cartoonish. Like, here are things they’re writing that make sense. 

So again, I’m not writing them myself. I’m actually copying what I see other people doing when I see images that I like coming through the platform. So it’s the reason I actually like Midjourney. You can see everything everyone is doing in those Discord channels. And so it can be helpful. Still, it can take you eight hours to get what you want. I use multiple platforms. 

Aaron: Well, I think you know, the key insight there is, it’s eight hours the first time. The second time would be 30 minutes if you needed to do something similar, which is the reason to work with somebody who actually does that stuff well. 

Christi: And I think it goes back to like, when people complain about different agencies and pricing models and fees, you’re like, “Yeah, well, I only want to pay you for 15 minutes.” Like, no, you’re paying – I might only take 15 minutes to do it, but it took me 10 years to learn how to do it in 15 minutes, and you’re paying me for the 10 years of experience and expertise to be able to do something at scale, quickly and efficiently. 

And I’d say, as we talk through AI right now, everyone should be looking at some aspect of AI and not burying your head in the sand to understand like, what are those trivial, menial, repetitive tasks that you do over and over and over? Can we look at them? Document them, write them down. Now let’s figure out, are there different AI tools and platforms that you can use to scale it so that you’re not doing that same thing over and over and over again? 

If we know there are certain things in our copy that have to get fixed every time that we write on our own, let’s write them down, and let’s write a script to edit and to hunt for them and find them and replace them over and over again, so that way it takes less pressure off of the manual editors and they hopefully find new issues. 

Building Adaptable Marketing Teams for the Future

Aaron: Yeah, exactly. Well, so that’s a good segue to your perspective, your vision for a marketing department of the future or an agency of the future. In two years, in three years, what do you think a marketing team looks like? 

Christi: It’s a bit crazy, because right now, with the economy, there are so many companies that are essentially not hiring, so everybody is having to do a lot more with fewer resources. So I think as I look even in the future, this has been the case since COVID that everything started to pull back a little bit, and I haven’t seen companies really expanding as much as they had pre-COVID era. 

So people have had to really – the people that are successful in roles are people that have the growth mindset, and are willing to learn, and are willing to step out of their comfort zone, to try different things, new things, and just fail, fail fast and keep on going. And it’s like the marketing department of the future – somebody that maybe you are an expert in a given topic, but you’re also willing to understand the peripheral topics and the peripheral roles around you. You don’t have to be like – you might be an email marketer, and you write email copy, and you can also help with the social copy, because if somebody steps out, you can step in and help with it. 

So I think it’s the idea of not just being an expert in one area, but understanding how everything comes together and ties together. Because you may be an email manager, but email only can get sent once you have people in a database. How do you get people in the database? What do you need to know about the people coming into your database so that as you’re doing the email sends, you’re getting the right content information to get the clicks, get the engagement, everything you want on the back end? 

You also have to know everything on what’s coming down the pipe in terms of restrictions and compliance around GDPR, all the different policies that will impact what you’re doing in your day-to-day job. So the marketing department of the future is people that are inquisitive, ready to learn, people that want to know more than just one thing. 

Aaron: Yeah, I agree, energized by change and very curious. And I think people who are very effective at orchestration, as opposed to just being a technician in a single discipline. 

Christi: It’s interesting. I have conversations with my team on a pretty regular basis like, “Well, what is this? You’re doing these things really well. What is the bigger picture? Take the step back. Think about strategy. Think about the universe of what we are trying to accomplish.” It doesn’t matter about all the things you do. What is the impact of the different projects and things you’re working on? Because if there is no impact, you’re just doing projects for the sake of projects. You’re not delivering value. It’s like, think about the value behind it. Because we can all spend 100 hours a week doing things that won’t actually drive business value. 

So there are things that drive business value. Sometimes we have to do things because you have a CEO or a client or a customer who says, “No, I need you to make the button red.” Okay, fine. But then you also have to look at the bigger picture of like, okay, so now we’re making the button red. How does this impact conversion? What is conversion optimization? Oh, I manage the website. I probably should also be thinking about conversion rate optimization. How does the structure all tie to – it goes back to curiosity and the willingness to maybe go outside what is your normal day-to-day to look at how all the things I’m doing are impacting the funnel, the business, etc. How do they all tie together? And let’s come together to make sure the strategy is cohesive. So I’m not a person with my blinders on like a little horse doing a role, not looking left or right, and just hoping what I do will get us to the goal we want. 

Aaron: I think that’s exactly right. I think you mentioned strategy. I think that the marketers of the future are deeply strategic, expert strategists, again, not technicians. I think you mentioned as well, if you’re doing something that can be documented and is repeatable, then that documentation can be used for automation, and there isn’t value in your doing that work anymore. It’s in orchestration and strategic guidance that you’re actually delivering value. So I think that’s exactly right. 

Christi: Yeah. And I think on that same side, like we’re having, I had a conversation with somebody yesterday, like, “Okay, so you’re our social manager again. You’re taking all the time to write these posts. This is great, but it takes you forever to publish them in whatever the toolset is. We can’t automate necessarily the publishing side. So let’s look at what do we do to come alongside you? Is it a vendor? Is it a contractor? Is it an intern? Is there somebody that has absolutely no digital experience, and I’m fine with them having no digital experience, but they’re willing to learn?” 

Sure, that we can start and again, everybody starts somewhere, whether you’re in college, high school, post-grad, like somebody that wants to learn, that’s willing to step in, that they’re willing to do something that maybe isn’t the most glorious job in the world. But you start by learning how to schedule all the posts. Then you start by understanding the scheduling of what works and doesn’t work in terms of timing. Then from the scheduling side, maybe you start to understand, like, what types of media work on the different channels, images versus not, like you gain your knowledge so that you can go from that tactical level to being more strategic, to driving more value over time. 

And so, like we’re having these conversations, like we don’t have to hire people all the time that are deep subject matter experts. There’s value in bringing people on that have a growth mindset and are willing to learn that we can train and teach, that have growth potential. Now, when you hire somebody on that’s senior but doesn’t have the growth mindset and also just has their blinders on for a role, then that’s how do you then look at marketers of the future and the roles of like, making sure you have the right fit within a job, a right fit within the role. 

Because unfortunately, I have, I don’t think I’ve ever been in a job where the job description and what I actually did matched up about six to eight months after I was hired for a job. Like, everything shifts and changes. So if you’d be like, “But no, I was hired to do this job,” like the marketer of the future has to be agile. You have to be able to grow, change and have that curiosity, because the job description never, in my experience, matches what I’m actually doing on a day-to-day basis. And being able to do the tactical but also having that strategic view and strategic lens will make you successful. 

Navigating Privacy Regulations and Data Strategy in Marketing

Aaron: I absolutely agree. You also mentioned compliance and privacy regulations. So we focus on healthcare and medical device clients, where privacy regulations are tremendously impactful for us, particularly with the expansion of the definition of HIPAA coverage to third-party cookies, which rendered pretty much all third-party cookies HIPAA violations a couple of years ago. We have been dealing with it in a context where a lot of the data that otherwise is available to marketers, to agencies, isn’t available for our clients. And we’ve been, like everyone else, sort of beginning to think about and reckon with Google’s impending supposed deprecation of support for third-party cookies, which they now are not going to do. 

 So given the toing and froing, given the frothy waters around privacy regulations, the data that will or won’t be available to you as a marketer. How do you think about compliance and privacy regulations in the future, and how does that impact how you think about data strategy? 

Christi: It is consistently on our mind, especially writing email now. Like on paid search, it was more so like, “Oh my gosh, the cookies are going away. How are we going to do all this remarketing, retargeting, and the audiences?” It was there, but now taking a step back and going to the 10,000-foot view like, “Oh, how does this impact what we’re doing in Europe? How does it impact how we message, how we do the outreach to different companies?” 

Oh my gosh, not just Europe. Australia has its own policy. China has its own policy. Japan has its own policy. In the United States, we are going down the path of every state could have its own separate policy. Yeah, 19 states so far, but you could have 50 separate policies, right? 

So, from a data perspective, for me, it’s the understanding by country, by within country, like, at the US state level. What are the different rules, regulations, legislations that are enforceable today, but what also is coming down the pipe for the future? Because it’s not just what’s enforceable today, because you can focus on today, but really you need to be looking at what is coming. 

And then the lens I’ve been taking with our team, and we’re talking to our legal teams, is, let’s start with the most restrictive, and let’s base off the most restrictive, because if we can make it work for the most restrictive, then everything else is a cakewalk. You can always pull layers back to get more information. But how do you function off of the most restrictive policies that are out there? 

We do have different tools and platforms that we are using. The biggest one for us right now on web and email side is Ketch to make sure that we are in compliance, especially with GDPR and all the different rules and regulations that are coming through with how we are collecting data, how do we use data? How to make sure we can delete data if we need to delete data. 

But it’s really we have, like a giant Excel document grid of all the different policies we need to be aware of, and what are the restrictive elements, so that as we talk about data and what we’re capturing, how we use it in the future, we at least have a good view today of what we know and then are figuring out as we go. Okay, so this shifted. What does that mean? 

And so we meet on a pretty regular basis to talk through okay, the FTC announced that hashing data is no longer enough to be considered compliant with privacy policies, right? So hashing data just meaning taking an identifier and making it a generic alphanumeric value that still technically can be tied back, which is why they’re saying it can be tied back to a user, but it’s not enough to identify that specific user. So now we’re like, “Okay, what tools are we using that hashes data so that we still have internal systems that can tie it together?” That is now against the policy. Now it was announced last week, so no one is following this today, but it’s like, hey, this needs to be on our radar, because potentially in the next 12 months, we’re going to have to figure out what does this mean and how do we comply with hashing is not enough. 

So, I mean, it’s just keeping track of what is coming, what are those changes coming down the pipe, and then trying to figure out, how does that impact us today? And then what does this mean for the future? Like, we’ve been prepping for cookieless since, I don’t know when was that announced? 2019? 2020? 

Aaron: Yeah, it feels like forever. 2020. 

Christi: I think, like it feels like forever that we were told cookies were going away, so everybody had to figure it out on their own. Like, cookies are still here-ish, there’s a lot of ad blockers, a lot of things happening. And so like, I explain to our leadership team, internally, to Azul, over and over again, like, yes, people have the right to not be cookied. When that happens, we see roughly 20% of the users on our site are considered anonymous. We can’t track, we don’t know what they’re doing. We don’t know who they are, but it also means we have roughly 20% of the leads coming in are unattributable, right? So on our media side, it goes back to like attribution. We know that the numbers in Europe look way worse than they actually are, and it’s because 20 plus percent of people choose the “I do not accept.” 

Aaron: And in healthcare or medical device, it’s more like 70 to 80%. So you lose fidelity. And what’s interesting, cautionary to me, is that, although Google has not said what they’re going to do instead of deprecating support for third-party cookies, the indications are in the initial announcements sound a lot like cookie consent on a global scale. So if you go to cookie consent and you get the same sort of opt-in rates, there you lose massive fidelity almost immediately.  

But I think what’s impactful for me is that a third party is still in control of your destiny. And so as we’ve been in a context in which you can’t rely on third-party cookies, we have wanted to, and we’ve been working with clients to take control of their own destiny and just shift to a first-party data strategy and not rely on third-party cookies. Is that a part of your thinking as well? 

Christi: We are starting down that path, and luckily, I have an amazing director of marketing operations who gets to own that aspect of the lifecycle, and so I am trusting his strategy. Let’s go back to, like, scale, right? There are things I can’t do. I have to be abreast of it. Luckily, I’m like, “Hey, let’s meet to talk about our data on the back end.” Because, like I said, GA4, yeah, we know our GA4 is not as optimal as it should be, like we have it implemented, but the reporting, there are certain things that just were really wonky with how it rolled out. So we have some fidelity issues through Google Analytics. So we have other platforms that come in and use it. 

So even our first-party data, like, okay, how do we tie all of our data together? We have a lot of different tech platforms we are using for different things. So understand, like, what are the implications across our entire martech stack? People smarter than myself are working on that problem, and I am very happy for that. 

And we do have a lot of first-party data. Now it’s figuring out, okay with the first-party data, how do we leverage it? So if we want to use it for advertising purposes, we can, how do we marry our first-party data with second or third party for the platforms that allow and enable it, so that we can then shift messaging? Like I really, really want us to move toward some of the platforms like Mutiny or Followoz, which allow you to update the website in real-time if you have information about the person coming in. So we can make it look like our website is for the gaming industry, healthcare industry, like, edit it to make it look like we are more specialized in a given industry versus a general homepage. 

Aaron: Yeah. And for us, because we’re working in the healthcare context, we need to bring that data together in a HIPAA-compliant data warehouse. And so we do that in a HIPAA-compliant data warehouse that we build, and then you have to surface insights that are actionable and activate them in a compliant manner as well. So you are right that it all starts with data strategy and a martech strategy to bring all of that data together in a way that adds value. 

Christi: Yes,  and a way that you can use it on a regular basis. Because I think the challenge I’ve seen – not like I’ve seen it at Microsoft, I’m pretty sure Expedia had this as well – you end up having different teams acquire different tools and technologies for a specific use case. So as you start to look at your martech landscape, it looks an awful lot like the map of all the tools available in the martech landscape, right? It went from what, like 600 tools to, I think the last time I looked at the published version might have been like two years ago, and I think they had like 4,000 or 5,000 tools. And then if you just dug into the AI-based tools, I think in there, it only had like 15 at that point in time, and now we’re at hundreds to potentially thousands of AI tools. 

So as you start to think of your technology landscape, how are people using different tools, even like talking AI, the data you put in and use on ChatGPT, you have to have your LLMs in a compliant GPT. 

The Importance of First-Party Data and Media Mix Modeling

Aaron: Yeah, for various reasons, you lose fidelity in the data that’s available to you, and the attribution modeling that’s available to you is more suspect than it was in the past. Are you shifting to different forms of modeling, things like media mix modeling, to enable you to more deeply understand what’s happening and where to spend media in the future? 

Christi: We are doing a bit of that. So we’re using our model. So I think as we were coming in, we were talking, we have, we’ve started with last-click attribution. We knew we were losing fidelity. Like, sure. Last click was like, “Oh, your brand on paid search is the number one thing you should do.” Shocker, yeah. Who would have thought, yeah, there’s no value in display. 

 I’d also put virtually no value in our email programs, because email is what you do to get them as a lead qualifier. So it’s a lead qual, it’s in the middle of the funnel. It’s not ever at the beginning of the funnel. It’s almost never at the end of the funnel. So it’s a qualifier, and then something else happens, because they click through the email, then go to the website and they do an action on the website. So, like, email was completely undervalued. 

So we started looking at, what happens if we shift the model from last touch to first touch? And then we also did a W-based model and sort of a U-shape first and last model. So looking at, what are the different valuations we see across marketing channels, marketing touch points? Where do the different touch points fit? So understanding pathing, not just the attribution, but what does that general path look like that goes from the first touch or the first time somebody engages with us all the way down to that thing that drove them into a marketing qualified lead that put them into that sales funnel. 

And so we are starting to do media mix modeling based off of it. But even then, it’s a little bit challenging, because there are so many touch points that we see, like average companies probably say like 20 or 30, depending on what the person is doing. For us on the Java side, there are some areas where our touch points really are like five to six, and it’s somebody has identified they have an immediate problem. They are on one of our biggest competitors. They’re on Oracle. Oracle is reaching out, auditing them, and they realize they’re gonna have this massive bill due if they don’t migrate off of Oracle to something else in a very short time frame, or migrate off whatever they’re using to another platform. So, like those people, they are very incentivized. They do like, five or six things, and then they are like, we want to talk to somebody today. We will fill the hand-raiser form. We’re on drift chat. We are calling your offices. That is few and far between. 

Then we have ones that we see 90 to 100 touch points, and on the channels, on the 100 touch points, you’re looking at that mix of, okay, we have a good amount of awareness that happens, and we have a lot of awareness assets. And as you start to look at the mix between the channels and the funnel and assets at funnel stages, that’s where we’re starting to model like, how many touch points do we need? 

I think I was talking a little bit earlier on the Account Based Marketing side, like again, audiences. So it’s not just the touch points, it’s not just the modeling, but where we’re looking at our modeling right now, and an account level by touch point. So we can see that we had a deal that closed this last quarter that had like 120 contacts, 120 form fills on our website. 

Aaron: Oh my gosh. 

Christi: We talked to a lot of people at that company, consistently for the last six-plus months. And so you start to look at modeling, okay, so we have three products. One of the products we know has a very long purchase cycle to it that people might realize they have the problem. They want to know more about us, but it’s a very customized solution for our prime platform, and so it’s a very long buying cycle. 

So for that product, we had to figure out, like on the media mix, it’s like, okay, we know from the point we get that first lead and first touch point, it’s probably a year, maybe two years. So now we have to size the opportunity. Is that still, when we get that lead, is that company the right size for us to continue to pay for eight to 10 months of touch points to bring them down that journey? Or do we just trust that, if it’s a small to medium-sized business, our sales can do it themselves? And how do we then adjust our media mix based on company size? The right contacts in the database? 

So it’s not automated, it’s not scaled at this point, it is still very much manual, and trying to evaluate as we go. But we’re adjusting the mix between our products. We’re adjusting the mix between our channels on a quarterly basis based on what we saw the previous quarter, and how we see the funnel shifting, and the length of time from that initial lead to the lead that hit the sales process. And then when we see our sales, we hit a sales qualified lead all the way down to closed won, what that stage of the process looks like. 

So we view everything on a monthly basis with our board and our leadership team, and then we go back and on a quarterly basis, we will readjust as we go, maybe a little bit more in the mix in between, if we see leads trusting less on certain channels. But it is a consistent reevaluation to make sure we’re looking at the metrics and that we are investing in the right areas at the right time so that we can show impact. 

And it’s been interesting. We started doing this really in-depth in November of last year, and we had our best quarter to date with marketing this quarter. We just, we are in a weird quarter cycle. Our quarter ended Wednesday. We just began the new – we are now in Q3 officially yesterday. So we ended our quarter, and it was one of our best quarters. And it’s looking at, we have a lot fewer people filling out the forms across the board, but the responder rate, because we’ve started to look at the audience and how we’re doing and adjusting our mix, we’re hitting the right people by focusing on a buyer center and focusing on our target accounts, that we’re seeing the conversion rates and the marketing or the responder to marketing qualified lead, marketing qualified lead, sales qualified lead percentages and rates have doubled by really honing in on audiences and personas. 

The New Frontier of Digital Advertising

Aaron: We think about where search engine marketing started, where paid advertising started years ago, where we are right now, the way that we’ve shifted from manual bid management, lots of time in spreadsheets, everything really powered by an analyst and a manual process that then was inputted into a platform to a context now where a lot of it is automated, not all of it. Where do you think we’re headed? What does digital advertising look like three or four years from now,  

Christi: Just the paid search side, or everything? 

Aaron:  So I’m interested in paid search, but I’m actually interested in it all sort of contextually, because I think it’s all kind of headed in the same direction. 

Christi: I want to hear what you think. I don’t know. I think I’m still bitter with paid search. I think I’m still a little bit in the bitter boat. And it depends who you talk to, for those of us who have been doing paid search for a really long time, like the 2000s – and I’ll say early 2000s I mean, like, I am sad of some of the controls that have been taken away from us, and I’m still sad and bitter of the control taken away. I like, yeah, we should – I’ve gone through therapy. I understand I should drop it, but even now, like the other day, I think last month, Google said, “Okay, we’re gonna decide if we think this keyword doesn’t have enough volume, we’re just not gonna allow you to advertise on it doesn’t matter if you think it’s relevant to your business.” It’s not enough volume for us to really justify the long tail anymore. 

I’m like, “You got to be flipping kidding me,” because I am in a very niche business, right? Oh, like, you’re right. Maybe there’s only 100 people searching for this niche term on a monthly basis, but those 100 people are so qualified for my business, right? That if we can get them, if we can advertise to them, if we get them into our website, we have the solution for what they’re looking for. One of them would be enough to make my marketing numbers. Like, yeah, great, Google. I’m glad 100 is menial to you, but that 100 for me is super qualified, and I want to reach them. I want to be able to advertise to them. And the fact that you’re telling me I can’t – give me the control back. I want that control. 

I get that you’re doing this for the small business that doesn’t have the time to go and manage everything on their own. Please give us advertisers the control to target as we want to target, the batch types we want to target. Do the bidding. Yeah, you do the bidding on the back end. I am fine with that. I will give you data all day, but let me decide what’s right for my business. Soapbox done. 

Aaron: I agree with all of that. I don’t think that we get it back, but… 

Christi: We’re not getting it back. It just makes me sad, like you’re just like, please, please, please, please, give it back. I think we are going to be going down this path, especially looking at Google’s – was it Google Next? Or the Google event back in May timeframe where they talk about the future, everything coming is really us giving them a website and telling them the thing we want to happen. We give them here’s the site, and then here is the end objective, and eventually they will just be going down the path of like, “We’ll do it for you.” I am not happy with it. I do not like that. I think it’s great if you are a team of one, but I think there’s a lot of control that we lose with that. 

Search engines have been helping people by telling them, “No, you really meant to do this instead” for a pretty significant amount of time, where I’ve been in situations, agency side and in-house side, where the search engines kept on rewriting our title tags. So, like, “We think this is what you meant.” We’re like, “Actually, no.” I was in an instance where it put us into a HIPAA compliance issue by what it updated on a title tag. Like, “No, no, we actually can’t say that. I get that you think this is going to drive better click-through. We legally cannot say what you’re putting into that tag and it doesn’t match what’s on our website. We have not done anything to say this.” 

So I think the challenge for marketers, the challenge for businesses, is going to be training the algorithms of the platforms of what is allowable, what is not allowable. Who is your target audience? Because even now, when we go to search today, who search thinks is my target audience isn’t who is part of my buying center. They use my product, but they don’t have a say in the purchasing process. So we can target them all day. They can fill up forms from here until kingdom come, and they will never help us make – like they will never help a deal progress. 

So like the future of advertising is being able to understand how the platforms work, to train the platforms of who you are actually going after, who you are targeting, what you are trying to accomplish, and making sure it has those constraints in place to get the end objective you’re actually going after. 

Aaron: Yeah, I completely agree. I think we’re headed to a time during which we’re dealing with increased automation, whether you like it or not, and what you need to do to drive effective search results and digital advertising actually becomes an extension of your data strategy. You have to ensure that the data that is being supplied to the algorithms, to the search engines, is the data that drives the result that you want. 

Christi: Data is just so fascinating. So as you think about the agency, the future, the business of the future, like when I was at Microsoft, one of the challenges we had, we are one of Google’s largest advertisers, right? Spent a lot of money on Google. We could not get tagging on our site for Google Analytics, Google Tag Manager. Any of that for some odd – who would have thought Microsoft doesn’t want Google tags on their entire website? 

So like, how does Google then work if you’re not feeding it data? So we had a lot of issues with our paid campaigns because we weren’t giving it data, and it was purposeful, like we had to go to Satya Nadella and say, “We would like to feed offline – we want to use offline conversion import. We want to feed it data so it has information, so it can make decisions on the back end, because it’s already making them, but it’s hamstringing us.” By not giving it data, it’s like, we promise. We will create our own algorithm. We will feed it information, and we are feeding it data. We’re feeding it information that is not the actual information, but it’s directionally correct, so it knows what to do. And it says, “Oh, this action behavior person was the right type of thing to be serving an ad to,” because when we didn’t, it made it really difficult to use any of their automated AI on the back end. You had to revert back to the dumb version of ads. 

We won an award for a case study based off what happens when you give Google data. It’s amazing. The performance improvements that happened on our campaigns by giving data. It wasn’t accurate data, it wasn’t the actual data, but feeding data, offline conversion import as a signal. It’s like, data strategy’s the way of the future. It’s like, yeah, I get it. We can’t give them our actual lead gen information. We can’t tell them like, this is lead, this is this. We can give a signal. Yes, we can make an algorithm to give a signal. 

Aaron: It’s very similar in healthcare and medtech, you can give them a conversion indication. You don’t need to give them anything else. It’s just an event happened or it didn’t. It’s a one or a zero. And that can be enough if you’ve done it in the right way.  

Christi: You have to have a data strategy. You have to understand how the tools and platforms are working and using the data on the back end so that you can manipulate it in the way to get the outcome that you need. 

Aaron: I think that’s well said. That’s a nice summation. I really enjoyed talking with you. 

Christi: Well, thank you. I’ve enjoyed talking with you, too. 

Aaron: Thank you.