Being a marketing agency used to be pretty dramatic. Coming up with an incredible, can’t-miss pitch for Heinz. Working at the eleventh hour to keep the Lucky Strike account. Showing up drunk to a presentation for Life Cereal.
Oh, hang on. We’re just having flashbacks to Mad Men. What a show, huh?
But how would modern-day Don Draper feel in a world where AI marketing tools are helping businesses create whole marketing campaigns—something that required a marketing agency not too long ago? How would he adapt? Could he?
Listen to our hosts, Pete Housely and James Thomson, find out the answer by talking to esteemed guest and marketing maven, Andrew Lionis, SVP of Strategy and Client Solutions at Datawyze, and learning how marketing agencies are using AI themselves to win in this new world.
In this episode, you’ll get answers to big questions like:
What are the benefits to agencies embracing AI?
How you can increase the volume and variety of your creative output using AI
How you can improve your prompts
If managing your data privacy is even possible in this new AI age
Listen to the episode to find out about how (human) marketing agencies can fare against marketing robots… or maybe team up and create an even more powerful force. You can also read the transcript right below.
Episode 3: Agencies of Change
[00:00:00] Pete Housley: Hey marketers. Welcome to Unprompted, a podcast about AI marketing. And, as you know, I’m Pete Housley, CMO of Unbounce. Unbounce is an AI-powered landing page builder with smart features driving superior conversion rates. We like to say our business is conversion rate optimization, and we’re gonna explore a lot about CRO in this series today.
Once again, my co-host is James Thompson. James is our senior creative director and leads our in-house full-service agency. Prior to Unbounce, James has worked for a number of agencies in both the UK and Canada, and brings a ton of knowledge and creativity to this role.
So welcome back James, and what’s on your AI mind this week?
[00:01:10] James Thomson: Thanks, Pete. It’s a pleasure to be back here again on Unprompted. So this week I’ve been thinking a lot about different use cases for AI, especially as how they relate to the creative industry and creative teams. One of the things I’ve been thinking about quite a bit is this difference between AI tool usage for drafting versus crafting.
So, you know, when I talk about drafting, I mean using an AI-based tool to help mock things up, to get placeholders, to get things into a conceptual place. Also you can use a lot of these AI-based tools for crafting as well. So for finalizing artwork, you know, for finished pieces, which are going live on campaigns as well. It’s interesting when you look at the different use cases for each to think about where the line is between the two. When does drafting become crafting when you’re working with some of these tools?
I was chatting with one of our designers the other day. She mentioned she’s creating a landing page, and she used a combination of Smart Copy and Chat GPT to create placeholder copy for the meantime until the Copywriter is available to get in there and write the actual final copy. But the question I had was, You know, the copy which you put in there, it’s really good. So at what point does placeholder copy just become final copy in some of those instances?
[00:02:27] Pete Housley: Well James, we talked a lot about the inputs and the outputs here, and hopefully as we get more adept at these tools, take command of those inputs, then we really do feel like the output is our own best work and still relatively proprietary. And I know I’ve spoken at great length with Garrett, our Lead Content Specialist here at Unbounce, and he just polished a 6,000 word white paper on CRO, and used Chat GPT throughout the process.
But his final piece he was able to do in six and a half hours. He thinks that’s some of his best work and it would’ve taken him almost a week otherwise. And I think as I’m nudging our team to get used to these tools, they’re also discovering just how liberating it is. I mean, we could get so much further, faster using AI and still illustrate the thinking and the team in a pitch situation, so I think that’s a very good use case.
Alright, as part of every episode we do a brief recap on some of the most recent stories that we’ve seen in the news. I have a number of news aggregators that come to me every day on AI, and I spend 30 to 40 minutes a day, and it’s hard to keep up. But some of the things that have recently caught my attention are that I listened to the Google Live show, and they’ve just announced their generative AI. Ad tools, copy generation, content and layout, you know, cropping, et cetera. And for one of the biggest publishing networks in the world, the ability for marketers and advertisers to use. That tool is going to be amazing.
And of course, I have come to departments like yours for most of my adult life, especially in the digital era, having you make all my Google ad units and social ads. And so I think that’s pretty game changing.
But then I saw Google Live once again. They’ve just announced a new product. It’s called Product Studio, and for those of you that have suffered with your Google feeds on Google shopping ads, this is a product that allows you to configure through AI product imagery all the products for your product feed. And you can take a low res image and you can make it high res. You can take an image and remove the background, or you could do a campaign.
The example that the product lead used was, oh, you’ve got a skincare line and it’s summer, and you want to have peaches behind your image. It will do that in an instant, and you can do that and configure it all within the Google Merchant.
So I think that’s pretty amazing. So as I keep watching these toolsets, all I can think about was all of the pain and suffering I’ve done over the years to get through all of these, which were so manual historically, and so many pieces. So I’m excited to hear a little bit about that. We know that businesses have historically gone to agencies for the type of digital design and analytics and some of the things that marketers would generally use. And so now agencies are faced with some of these toolsets being in clients’ hands and are they changing their business models and changing their delivery to ride upstream with this? And so with that in mind, James, I’m gonna start with you. How do you think creative agencies are coping with and embracing AI?
[00:06:34] James Thomson: Obviously we are looking at such a broad industry. There’s so many agencies out there. I think there’s gonna be different levels of adoption across the agency world. I think there’s always gonna be those early adopters who jump on the new technology and they’re looking for those opportunities to do what they already do, but do it better or get better results or do things more efficiently for their clients.
And then I think there’s going to be a little bit of, you know, retaliation against something which is new as well. And sometimes, when something’s unfamiliar and untested, and it’s still fresh, you can almost go back and rely on some of those traditional methods. So I think we’re seeing a little bit of both with the agencies. I’m more interested in kind of discussing a little bit more of those ones, which have been embracing some of these technologies already and are starting to get better results for their clients, augment their processes.
[00:07:24] Pete Housley: Interesting. Well, with that in mind and embracing the AI tools, James, in our first episode, we had your team develop a whole campaign using only the AI tools, and that was a ridiculous notion, but they pulled it off quite elegantly in the campaign that we ran, actually got some decent results
[00:07:40] James Thomson: Elephantly or elegantly?
[00:07:42] Pete Housley: Well, it was. We did create a trampoline for elephants and that was the task. So a little bit of both then. But they briefly mentioned some of the tools they used, whether it be Midjourney or DALL-E. And so I just kind of wondered, can you explain to our listening audience what exactly these tools are capable of doing and maybe some of the use cases? Just demystify that for us a little bit.
[00:08:07] James Thomson: Yeah, so Dall-E and Mid Journey are the two which are used quite commonly. They’re both text-to-image generation tools, which means when you’re in the interface of either Darley or Mid Journey, you have to put text inputs quite often called prompts into the tool. And those prompts are gonna define what the output is, what those images are, which it’s gonna come back to you with.
So, a lot of what’s gonna get you some really great results there is often being quite specific with what you’re looking for. And the more inputs you are gonna put in there in terms of those text prompts, the more likely you’re gonna get something which is attuned to what you’re looking for.
So DALL-E and Midjourney, they’re both very similar in terms of output, but there’s a few little subtle differences in there as well. So, for example, DALL-E offers the ability to do what they call in-painting and out-painting. So in-painting is when you upload an image and you’re able to recreate parts of it through AI-generated imagery for specific parts of that image.
So say you wanted to edit an image of a circus, but you wanted more elephants in there, and our elephants are gonna be the topic of conversation today, maybe even throughout the entire podcast. But say you wanted a couple of elephants in that circus shot. You could paint those in through DALL-E. Say you wanted to zoom out on that image and recreate the audience for that circus, which isn’t in the original image. You can use what they call out-painting as well.
So that’s where AI is using everything it knows from its machine learning algorithms, which it’s been trained on, to fill in the gaps and paint the outside of that image to expand it to be larger than it originally was when you put it in there.
So that’s one of the benefits of Dall-E. Midjourney does have benefits as well. One of them is that you can upscale images to twice the size that you could with DALL-E. So if you want a little bit more of a high-resolution output, then Midjourney might be a good solution for you.
But these two products are, um, developing really fast, I think. One thing which is gonna change, especially over this coming year or two, as a lot of companies do come to rely on some of these AI-based image generation tools, it’s gonna put a lot of pressure on creatives and photographers to justify their craft and work harder to justify some of the ROI of that.
I personally think there is a lot of value in photography, custom photography, in that it does allow that customization to really kind of get in there and own a lot of those details. But I think that the case is gonna have to be made a lot harder to warrant that investment. When you are, you’re looking at the other option being, you know, AI-generated images, which you can get for free.
[00:10:46] Pete Housley: Couldn’t agree more. The world is changing so fast here. Okay, let’s shift gears. Now we’re gonna go to the leader of a very data-focused agency, but of course, capable of doing content development and page design and all those good things.
So the gentleman I’m about to introduce to you today, I’m really excited about. One of the most savvy marketers that I’ve met. In the last decade, and coincidentally enough, when I met Andrew, he was at Adobe and almost in the role of like a Sales Engineer because he was so adept at data and marketing stacks and thought like a client so he could literally sew all those pieces together.
So Andrew Leonis is a partner and SVP of Client Services and Strategy at Datawyze. Andrew is a customer-focused business leader with deep skill sets in marketing, ad tech, AI and machine learning digital applications, but then through the line to social mobile technologies, et cetera.
Datawyze is what I call a new age marketing intelligence agency, representing that application of data and marketing research. Into business process and decision making.
So I’m very passionate about some of these new and emerging businesses, which I think are just so timely for what’s going on in this environment. We, as marketers, our return on ad spend and our cost per lead and our cost per customer, I mean these metrics that we’re being held accountable to are just so important. So with that in mind, Andrew, you’re a bit of a unicorn, as I’ve just described, you can virtually do almost any task in a marketing stack or an agency stack. So why don’t you just tell us a little bit about yourself, maybe some career highlights, but I’d love to hear about some of your most memorable moments of innovation.
[00:13:19] Andrew Lionis: Thanks, Peter. Yeah, so I’ve been afforded the opportunity to work both agency and client side, brand side. That is, and throughout my career, I actually started out in the agency world. So a lot of the items that James pointed to as it relates to being able to deliver volume and variety, whether it’s creative or from a digital production standpoint, the speed in which you could deliver that was limited to how many resources you had, what was the volume or variety of the experience of the individuals working on the work from agency.
I did have an opportunity to work brand side. It gave me a different lens to our business that afforded me the opportunity to work across the entire MarTech stack just by way of being able to work through the brand’s touchpoints as it relates to how it engages with the consumer, what type of value does it deliver back to the consumer, but also exposure to business intelligence, right?
When you’re working brand side, you’re working with folks from the BI team, the analytics team, trade marketing, sales, et cetera. One of my highlights when I was working for one of the world’s largest holding companies in the spirit and wine space was actually onboarding MarTech that was at the time, very, I’d say in its infancy.
So this had to do with delivering media at scale. You know, the right place, the right time on the right site for that audience that you want to get in front of. Leveraging first and third-party data, of course. So when I onboarded a the time, it was Two Mogul. That was something that was first in market in Canada. What it did was it exposed to me the opportunity, and that’s when I really got excited about technology and the way things were changing because I saw the way that we could move and the speed that we could move in.
So from there, I actually started working for Two Mogul and then by way of exit and acquisition, I got to work at Adobe when they acquired Two Mogul. And through that experience, I had some experience working at Wisdom AI, so I learned a lot about natural language processing, early adoption. Of course, working with some of the biggest brands in Canada that were, you know, fearless, some marketers that were like, “Yep, this is the future. Let’s start working with NLP.” As you know, a first step in the AI arena.
And now today, look at where we are. I’m a partner at Datawyze and we are leveraging AI in our day-to-day. So, exciting times, and I think throughout my career I’ve been afforded the opportunity to work on both sides of the fence, and I leverage that for our clients today. And I’m excited to be here and talk about my experience then now, and what I see for the future.
[00:16:12] Pete Housley: That’s amazing, Andrew. Incredible story, incredible skill sets, and I wanna hear a little more about Datawyze. Tell us generally what problem is Datawyze solving in the market?
[00:16:27] Andrew Lionis: So the biggest problem that we tackle is tying every effort back to revenue. And that’s where I think we found success, At Two Mogul and Adobe, and even at Wisdom. And I’ve carried that over to what we’re doing here at Datawyze.
And the challenge that we find today with legacy systems or legacy processes are that in which they are slow. There’s not a lot of variety, and the volume in which you can move around does not support the kind of agile way of working that you would need to deliver and move at the speed of which consumers are moving to drive incremental revenue for your organization. So Datawyze really tackles that in a sense of we’re using and leveraging tools that allow us to have that velocity that we need to identify an opportunity that might only be available for the next week, and what a grand opportunity that would be if you could drive an incremental a hundred, 200%, 300% in sales by being able to move that quickly.
So Datawyze really focuses on leveraging tools that incorporate AI or machine learning, but also we supervise and optimize these tools. There’s a human touch to everything we do, but it allows us to move at the velocity that we need to drive value for our clients.
[00:17:56] Pete Housley: It’s interesting, Andrew. At Unbounce, we have 17,000 clients and we just completed a major research project to say what’s important to marketers today? Is it CRO? Is it return on ad spend? And to your point, revenue came back as the number one.
And I think right now, we’re in a bit of a recession and ad budgets are being cut and people are battening down the hatches, and so revenue creation and growth is on everyone’s minds. And of course, that’s why as marketers, we need to be sophisticated and use all the tools that are available to us, and I think that’s a good platform for the discussion on AI. Alright, let’s jump right into this then.
[00:18:54] Pete Housley: How has the use of AI impacted you and your agency’s approach to marketing?
[00:19:03] Andrew Lionis: The impact that it has is the value exchange between our agency and our clients. That value exchange has increased over 300% on what we’re able to deliver for our clients and these tools.
I’m gonna go back to three things IF there’s a takeaway for the listeners today: Variety, volume and velocity. So when we think of those things, we are limited as individuals from a variety perspective. So, by way of our experience, right? The variety in which the lens I look at a client’s problem is really limited to my experience with previous clients, whether it’s an industry related problem, whether it’s a product related problem, whether it’s a user related problem, right? So when we think about variety, we think of the breadth of variety that you get with AI tools. That’s number one.
The second is volume. How much volume have I experienced in my career versus having a large agency with 20 folks versus a small agency or a larger team within a larger agency, let’s say 20 folks of maybe a group or a publicist that has thousands of employees, versus a smaller agency that might have what I call a SWAT team of 12 to 15 folks, and that volume, again, is limited to how many hours we have it in the day.
And then when we think about velocity, we think of, you know, how fast can we move really? And how fast can we type, how fast can we read data? How fast can we extract the data, parse through it, identify opportunities to what you said, Peter, drive revenue for the clients.
So when we look at AI tools, We are making leaps and bounds from the perspective of variety, volume, and velocity. So when a client is paying, they can come to our agency, which is market fair, another agency which is market fair, and a third agency with which is market fair. But the agencies that embrace AI and can deliver more variety, more volume at a higher velocity, will give more value exchange to those clients.
And that’s what I see as the biggest benefit for agencies that embrace AI, specifically AI that either is a standalone like Chat GPT or, James, you brought up Adobe, we can bring up another one, Canva, if you want. Adobe’s competitor. They’ve got text-to-image embedded right in Canva. They actually have DID ai, Which embeds talking head video into your designs. So when you think about what you’re able to do with these tools, it’s just better for the clients. And I think, you know, the big takeaway for the listeners today is to think about how it affects your variety, volume, and velocity.
[00:21:53] James Thomson: Uh, you know, I think about the importance of audience insights and knowing your audience, obviously, and how that pertains to marketing also within creative as well. It’s super, super important to tell the right stories to the right people. I’m just wondering, in what ways do you leverage some of these AI tools in order to enhance some of the data analysis that you’re doing and to uncover some of those customer insights which are also important. Andrew, are you able to walk us through maybe some practical use cases where AI has enabled you to uncover some of those valuable marketing insights?
[00:22:24] Andrew Lionis: I liked what you said earlier about being an expert in prompting your AI tools, cuz I think it’s very important and it’s really the way that I wanna start the answer to this question. Really understanding data in, data out really understanding the prompts that you’re gonna give. The AI and the data that you insert into the AI will really give you what you’re looking for back.
So I’ll give you a use case. Let’s say you’ve got a product that, an international brand, they’re just starting to segment, you know, some ideas on different strategies, whether they’re working hand in hand with their agency or they’re just working internally to really kind of say, “Hey, we want to build a couple segments here.” And arguably, let’s just say it’s Australia, the UK, Canada, and the US now.
Traditionally, what you would have to do if you wanted to really kind of dig into the populace of these countries and look at demographic is you’re looking at the census data, right? You’re digging through census data. How many males and females in Australia, how many males and females in the UK, US, Canada, et cetera. And then what are the age cohorts? And let’s say you want to add a third or fourth data set. Well guess what you can do. You can sit there manually in the old world, and when I mean old world folks, if you’re listening here, I mean like six months ago before the Chat GPT launch, at the end of November. So the old world was not that long ago, right?
So you can sit there in the old world and do all this research and put it in a table, you know, start pivoting, start cutting and pasting, and kind of parse through that data, or you set up the prompts, which is what we do so that we can actually start building segmentation across multiple data sets within Chat GPT, and just say, here’s a prompt.
“Hey Chat GPT, can you give me the population cohorts that look like male and female age, maybe by city?” Then you can even go into other data sets, or if you’re looking for a data point that you want to include in that, either on the initial prompt or when you get the data back, feed it back into Chat GPT so that they can reorganize the data for you on a second or third pass back.
So that’s one great example where you can actually start with no data prompt. Systema like Chat GPT give you an audience segment that you might be interested in so you can see are there more males and females in this age cohort in which country, and how does that influence what James said earlier, which is how we’re gonna develop the creative or the messaging.
[00:25:09] Pete Housley: So Andrew, it blew me away. You had told me this story a couple of months ago about creating table structures within Chat GPT, and so I’ve been dabbling in that ever since. And now I see we can even bolt our marketing data onto conversational AI. So if I have the hooks into Google Analytics, for example, that I could then just conversationally get all my table structures. So now as a marketer, I don’t necessarily need to do advanced pivot tables and relational tables and all that kind of stuff, but I can shape my data and my board presentations, my segmentation insights, using my logic as opposed to my syntax. And I think that is absolutely game changing.
[00:25:57] Andrew Lionis: Absolutely, Peter and, and the thing is, a lot of these tools are going to have that silver bullet soon where you press a button, you prompt it, and it gives you the data, whether it’s Microsoft Office, you mentioned Adobe, all of the big tech companies, Salesforce, et cetera, et cetera, they’re all on it. So it’s a good way just to add that velocity that you need to drive more value for your clients or for your own job internally on the brand side.
[00:26:24] Pete Housley: What is the expectation of yourself and your team in terms of using the like, are there mandates, is there workflow? Like, I just kind of wanna understand your business model, your ethos, your workflow around ai.
[00:26:38] Andrew Lionis: So we’ve adopted an AI-first approach to our marketing strategy for all of our clients. We start with AI, we don’t end with AI. I think one of the things that is still important is supervision of AI. So what we always do as part of our agency work is we start with AI. Let’s see what it gives us back. We typically have, and this is something that is part of our ethos and the way that we work with our processes, we always send it back. So if you’ve ever watched Hell’s Kitchen, Gordon Ramsey, he sends it back. It doesn’t matter how good it is. So we do the same thing. We send it back to the AI. Whether it’s a send-back for a prompt or if we’re whiteboarding, somebody brings up something that they’re working on that may be started with AI that they supervised and optimized with. Returning prompts to, let’s say, a chat GPT, for instance, but we try to send it back and that is our ethos. I think that agencies that embrace it and work it into their processes will find a lot of success, whether it is more efficiencies or delivering more value for the clients.
[00:27:51] Pete Housley: You know, it’s interesting. Years ago when I was at Procter & Gamble, we would go to how to evaluate a storyboard and how to give feedback to creative teams. And of course the ethos was always like, please understand that this creative person has just responded to the brief and they have put forth what they think is their very best work, and this is their craft. So if you say you don’t like it or could you see something different, it becomes a little personal. So we have to talk about what is the ad communicating and what am I taking away and is it on strategy?
But generally speaking, we always wanted to figure out a way to give feedback to creative teams, to inspire them to do the other iterations and not necessarily hurt their feelings. And I think what’s interesting is you can’t really hurt AI’s feelings as you do iterate. Try something else. Try it with humor. But I do think it’s interesting that you can literally have AI do it over and over and over again.
[00:28:45] Andrew Lionis: That is a great example of variety and volume, right? So again, your resources with your creative team, they’re limited to the variety in which they could produce or the volume in which they could produce it. And you’re absolutely right. The AI tools do not take things personal. In fact, you can send it back for unlimited revisions. It comes back really quick. You get a variety of volume of iterations in a quick manner. And I think that’s really important so that you remove emotion and you give everybody an opportunity to really be the driver behind the result. And we talked about bad data and bad data out, or good data and good data out, or the prompter. It actually is empowering to the individual to build it into the process so that they are the ones that are driving the vehicle, whether it’s Chat GPT, Canva, or Adobe Firefly.
[00:29:39] James Thomson: It’s funny, we talk about the emotion and removing some of the emotion from that process. Speaking on behalf of the creative process, I think emotion still has belonging, but it’s just that it lives in a different place.
As you talked about earlier, Andrew, I think having the right inputs to put in there that it’s working with and then I think also in terms of the outputs as well as when we’re getting some of those results back from our AI tools. Is what does this make me feel? We need to be evaluating on a human level. Is this compelling? And I know we can test a lot of those sorts of creative and see how it performs with the numbers and the statistics and the quantifiable data. But in terms of the qualitative, I still think the human emotion and the human validation still has a belonging in terms of that process just in a slightly different place than where it has been previously.
[00:30:25] Andrew Lionis: I agree with you, James, and you know, one of the things that I think as a tactical takeaway for the audience is build your prompts, right? If it comes back and you feel that it’s really not speaking to your audience when you send it back, that supervised optimization that you just did, you need to take that back and put it into your initial prompt so that next time when you prompt the AI, you’ve incorporated that piece of feedback on the first try. So like anything, baking a cake. Let’s stick to the cooking example. It’s just gonna get better. A little bit less salt, a little bit more pepper. Let’s see how we add a little bit more sugar.
[00:31:05] Pete Housley: Hey, Andrew, I was really surprised— not surprised, delighted—to find out that Datawyze is actually leveraging some of Unbounce’s AI tools. So why don’t you talk a little bit about landing pages and Unbounce and AI? We’d love, uh, well, we’d love any customer, but we’d love to hear your take on that.
[00:31:26] Andrew Lionis: Yeah, well, as we know, whether it’s B2B or B2C, lead generation in revenue is super important. Just the speed in which you can optimize what you’re doing when your audience lands on your digital product, right? So if you’ve got this website, depending on what channel they come in, on what device they come in, the great thing about tools like Unbounce is it gives you an opportunity to do a lot of things that you would historically have to do manually that is built into the product itself.
So when we think about A/B testing, incorporating the AI tools that Unbounce has, so that the landing page itself will be optimized for the right audience, depending on the channel and device they come in from, is really where we see the value in using Unbounce.
So, for ourselves, we use Unbounce for all of our lead gen clients, so most of them are B2B, we have some B2C clients as well. We use Unbounce for a lot of clients that are launching new products, new services, even clients that are looking to do things like setting up webinars, clients that are offering value through some of the content that they want to deliver to their audiences, and it’s been a great tool for us. So far, so good.
I do recommend folks try it. We have not been disappointed and most of the things that we’re seeing with Unbounce is drive incremental value as it relates to conversion, specifically around 15% plus across the board, across different industries for different clients.
So we’re happy. I would consider us maybe power users. There’s still a lot that we can unlock as it relates to some of the features that are part of the solution, but we’re very happy with it and so are our clients.
[00:33:18] Pete Housley: Well, music to my ears. But right now the trend that I’m seeing, and I’m also seeing Google and Meta get on this trend, is multi-variants, multi-audiences.
So no longer are we A/B testing a single variable impact, but now what we encourage our clients tp “build 10 versions of your landing page,” both on Smart Traffic and to your point, depending on the device, the geo-location, you know, and a number of other factors, it serves up the right ad unit to the right audience, and it’s constantly optimizing. And now, with the tools, whether it be Smart Copy or chat GPT or our in-app products, you can create these multi-variants so easily.
But yes, out there marketers, if you are not AB testing and or using multi variance, you really are leaving dollars on the table. And whether it’s pre click or post-click, it’s just what good marketers do today.
[00:34:23] Andrew Lionis: Absolutely. And you know what, Peter, just one last thought on that. Always be optimizing. Always be testing. And if you can’t do it manually and you need the help of tools like Unbounced or others, I highly recommend it. So, uh, you’re absolutely right.
[00:34:38] Pete Housley: Always be testing. And Andrew, we lose sight of those things. We get busy in our work lives and we have our functional tasks and our project priorities. And if we’re not setting aside for testing new ideas and growth ideas, then we’re just failing to innovate. So I think that’s great advice, james.
[00:35:00] James Thomson: Yeah, completely. And being a data-driven agency such as yours, obviously getting a lot of value from on bounce products as we discussed is, is really great to hear. So obviously as I mentioned, being from like such a data driven agency, I think where. Any sort of customer marketing data is involved. I think there always tends to be a little bit of conversation around like privacy, security, and how that plays into things as well. I was wondering what your perspective on that side of things and how it relates to what you do as well, you know, are, are you worried about that side of things at all?
[00:35:33] Andrew Lionis: Yeah, James. I think that it comes back to making sure that you have a chain of custody for first-party data, right? So within your organization, whether you’re on the brand side or agency side, I think one of the things that we do on our side is if we are using third-party tools, we make sure there’s no PII.
Privacy is a concern across the board, especially when using these tools, and I think that what I do recommend to agencies and brands alike is just look at your policy as it relates to privacy and data, especially if you’re a brand that is sitting on folks’ first, last name, maybe their address, you know, where they live, what does their profile look like, as in, you know, how much did they spend with you, their basket size, et cetera, et cetera. So we always strip that out.
We focus on data sets or points of data that are ambiguous in the sense of it’ll still give us a lot of value without giving out PII.
But yes, it should be embedded in the process and you should embed it into the prompt process as well as make sure that you’re doing the right thing for your customers as it relates to their data. So definitely a concern. And I would say check with your organization’s policies and make sure you treat it like any other tool.
[00:36:54] Pete Housley: Yeah, I mean, I think the risk is that we upload data into, you know, AI and then it becomes public domain. And we’ve heard a few examples of companies that have even uploaded their code base because their developers were consulting chat GPT.
But I do think, Andrew, you make a great point. I mean, the world of data and the big tech giants right now are, you know, We’re deprecating cookies. GA four is no longer gonna give you the individual customer journey at the individual level. It’s gonna aggregate that, and we all have to understand how we’re gonna run attribution models, and you know how we’re still going to leverage data, but we might not get the individual user data. And certainly we need to make sure we’re anonymizing any data that we’re serving over to the AI, machine learning, giants.
[00:37:50] Andrew Lionis: Absolutely, Peter. And again, you know, I think that going back to what we typically see with legal and compliance, they’ll get involved, they’ll slow it down. I think you build a use case around protecting individual’s data and you’ll find a way to get approval to use these wonderful tools that we’re using today.
[00:38:10] Pete Housley: So Andrew, from the story that you’ve told us so far, it’s like all smooth sailing. What are the pitfalls or what problems have you faced with implementing or using AI?
[00:38:22] Andrew Lionis: Yeah, I think the biggest pitfall I would say is you make sure you have a combination of both waterfall process and agile process. I think one of the things that you’ll run into an issue is, if you’re continuing with something that becomes expired or stale, right?
And that’s where you run into an issue because your client launches a new product. Maybe seasonality has changed the way your customer’s engaging with your product or service, and also the environment around us, right? Depending on what season we’re in, depending on, you know, the state of the economy, for instance. So I think one of the pitfalls that you will find is picking up on a supervised optimization opportunity without kind of revisiting where you left off. And I think that’s super important.
So as much as we’d like to work in an agile way, I think inherently as individuals and humans, we’re always self-aware of where we’re at, right? We’re self-aware of where we’re at in the project, where we’re at with our clients. Maybe you know what’s going on in the economy with consumers, whereas the AI tool is not conscious, right?
So it is actually, you know, you could prompt something that’s stale or expired from months prior, and it’s not gonna know that. You know, the Denver Nuggets have made it to the NBA finals, so that that would be my biggest piece of feedback from a tactical standpoint.
I think the other is, going back to what we talked about earlier, James, with the Gordon Ramsey example, is you don’t want to get lazy, right? It’s like you wouldn’t put a blindfold on, even if you’re in a Tesla today. Not to say that, you know, the self-driving car isn’t safe. You know, some people that are more adventurous will fall asleep in their Tesla while it gets them from A to B, but I think we still have to, even if we’re gonna be passengers in this journey, just be aware of what’s going on and making sure that wherever we’re picking ourselves up, that the tool itself is brought up to speed.
[00:40:22] James Thomson: Yeah, it’s a great point. Actually reminds me of a quote from the ethics side, actually from Jason Allen. And the quote is in response to some of the questions around the ethics of AI and some of these tools. And Jason says the ethics isn’t with the technology, it’s in the people. So for me, that resonates with what you’re saying around, you can’t get lazy, you can’t leave it, just saying forget it.
You need to run those checks and balances over time and ensure it is still serving the purpose. You know, as we talk about a lot of the use cases for AI and how they’re helping to give agencies a leg up, especially your own, Andrew, as we’ve talked through, it’s interesting to also see how fast things are moving even within the last three months. Their technology’s developing so fast, and then what might even happen in the next three months, six months, even in the next twelve months. Just interested on your personal perspective: What opportunities do you see with AI in the near future for agencies and marketers alike?
[00:41:21] Andrew Lionis: You know, I’ve got no crystal ball. But if we look at the trend and where we’re headed, we’re gonna see a lot of efficiencies realized on both sides that’s gonna disrupt the make-up of organizations. I think that’s gonna be the biggest disruption as it relates to agencies and brands and enterprises in general as they adopt these tools and as individuals become the drivers of these tools.
So I think we just have to rethink the construct and make-up of our organizations and how that impacts how we deliver value as a service provider for brands, as an agency service provider, same thing as an individual.
If you’re in a role, I think that the best piece of advice I would give is, you know, if you’re a fast follower, I know the early adoption, as crazy as it sounded six months ago, that ship has passed. But be a fast follower. Take some time out of your day if your organization is not promoting you to use it out of your own time.
Think of ways that you can incorporate this into your day-to-day. And I think what we’re gonna find is a lot of folks just really enjoying the pace of their workload.
Because what ends up happening is you’re starting from a platform in which you can launch versus starting with building the platform in which you’re gonna launch off. And I think that gives a lot of opportunity for people to get back time into their day. To what Peter mentioned earlier, we’re stuck in the day-to-day. We don’t have time to take a step back. We don’t have time to think: “What is the strategic approach I need to take next month or next quarter with my individual contribution or my team’s contribution to the enterprise or as an agency to brands?”
So I think getting that time back will be a beautiful thing for everybody, whether you are at the entry level of your organization or team or you you’re leading strategy, top to top.
[00:43:24] Pete Housley: I’d like each of you to just, you know, reflect for a few seconds and maybe summarize kind of your one recommendation to our audience based on AI marketing and some of the topics that we’ve explored today.
[00:43:42] James Thomson: For me, you know, I’ve heard one narrative and it’s around a lot of these AI tools and how they are allowing companies to do some of the things. That they have previously relied on agencies for, you know, the tool will kind of do a lot of the work for them, and I’ve heard a little bit of fear there. And personally I kind of wanna address it through a quote, which actually our art director Ceci provided to me yesterday, which I thought was pretty funny, having worked in agencies the most of my career. And the quote was, “As long as clients continue to not know what they want, then agencies will be safe.”
So for me, I think agencies will always be those specialists, spirit guides that know what is gonna work best, know how to identify a problem, a strategy, and know how to look at data, how to evaluate it, and make some of those recommendations.
But I think the agencies which are going to survive and get ahead and be a little bit more successful in some of these changing climates are the ones who do jump on some of these AI-based technologies themselves and, as Andrew said earlier, test some of these tools.
Why not leave Chat GPT open as a tab and see how you use it within your existing process? I remember working at agencies and having the occasional little bit of downtime and wondering what I’m gonna put in my time sheet because we haven’t got work on this specific week. Those are the weeks, and those are the days where you need to be doing a little bit of AI-based work and do some, you know, experimentation with some of these tools.
See what works for you, and there’s gonna be stuff which doesn’t work as well. Take what you want, move forward with it, and you are already gonna be one step ahead there and getting better results for your customers. So get in there, try some of these tools, see what works for you and, yeah, you won’t regret it.
[00:45:36] Pete Housley: All right, gentlemen. That’s a wrap. This was Unprompted, a podcast about AI marketing and you at Unbounce. Our purpose is to help you grow smarter.
And gentlemen, I feel smarter after spending the last hour with you two today. And thank you so much for sharing your wisdom with us, that was wonderful. See you guys.
[00:46:03] Narrator: This podcast is brought to you by Unbounce. Most AI marketing tools are kind of the same. That’s because they’re built on the same generic machine learning models, and they get you generic results in your marketing. Unbounce is different. It’s trained on data from billions of conversions, which means it gives you content and recommendations proven to get you more leads, sales and signups.
If you’re a marketer or just someone doing marketing, you need Unbounce. You can build beautiful high converting landing pages for your ads and emails. Plus get AI copywriting and conversion optimization tools, all powered by more than a decade of marketing data. Get the most conversions with Unbounce. Learn more at unbounce.com/unprompted.