Technology
Services
Resources
Automation is a game-changer for advertisers. As Co-Founder, President, and Chief Product Officer of Fluency, Eric Mayhew has made it his mission to build automation technologies that improve how people buy digital media.
In this episode, Eric joins host Noor Naseer to share how automation and AI can reduce low-value manual labor, minimize friction in workflows, and help organizations gain a competitive edge.
Noor Naseer: Automation: It's an oft-mentioned word in the advertising world that can mean a lot of different things. Understanding its potential is something more ad pros are determined to figure out especially in the last year since the popularization of AI. Eric Mayhew is the Co-Founder, President and Chief Product Officer of Fluency, an organization focused on automation solutions for digital media. He joins us to share wisdom from his 20-year plus career in digital marketing, ad tech and software engineering to highlight how automation can reduce low value manual labor, minimize friction and workflows and just help us in organizations not get bogged down by some of the needlessly laborious aspects of media buying. Let's get into this episode on automation and AI with Eric right now.
Eric, I just want to thank you for the time you've made so we can talk about another iteration of artificial intelligence and where it intersects with automation and these are things that you have a lot of passion around. So, I'm looking forward to this conversation.
—
Eric Mayhew: Absolutely. Thank you for having me today.
NN: So, I'm going to take us a year plus back. Whenever the new renaissance of artificial intelligence really started, maybe you could say it was the end of 2022, early ‘23. I want to get everybody on the same page. How did we really start embracing artificial intelligence? If you can take us back there so we can kick off this conversation on where it intersects with automation.
EM: It is absolutely amazing that it's only been about a year where this has been this pervasive. I think open AI came out in November 2022 where it was at Chat GPT3 and pretty exciting to see what's happened since then. That natural language the fact that you can interact with it in such a natural way is really profound. So, I think that's why we're really here today to talk about AI in its use case especially in advertising where we're a vehicle and a voice piece to communicate between our clients and their clients. So we're trying to talk on a human level between the two. But I don't think that's when it really got started.
AI's kind of been around for a while I've been in some panels where even I was exposed to things that I didn't realize how much longer it's been in in our society. It's just been kind of guarded from the general population and practical use cases. But as it's become more prevalent and pervasive, because of those new innovations there's a scramble to find the great use case and get the efficacy from that that we think that and the promise that's there.
So, if we take a year ago - I spent a lot of time in the Meta environment, spent a lot of time in Google and Microsoft and in the programmatic space. Take those first couple that I talked about with Google and and Meta and they've been in the space of trying to do AI type of content management and prediction capability for quite a while.
Google went to Smart Bidding a couple years ago. So Smart Bidding was kind of Google's first space into prediction of all these attributes about users and their proclivity to take an action. One of the bidding strategies that they have is maximize conversions or maximize conversion value. Those two things are all in the propensity for someone to actually submit a lead and it doesn't really tie as close as it once did to keywords.
Now if we even rewind farther back, Google's broad match just as a concept, that is an AI type function. Here's a phrase somebody typed in, that phrase isn't exactly what you're bidding on as a keyword. But that keyword is for the same intention. So there's always been this interpretation. There's been this growing path through the ecosystem to get us to the spot where we can say that AI takes a simple concept a simple implementation and expands it out into something very tailored and curated and ready for the general public.
So, I think we've seen this journey over the last year. And in the last year with Chat GPT we've seen that we can actually automate something that's been elusive for a long time. And that elusivity has been around the creative side, talking to customers in a natural language in the way that they'd want to be spoken to. When you're going to generate creativity and you want it to resonate with a person it cannot look robotic. And the new LLMs and the new Large Language Models make that feel supernatural.
NN: You jumped into also very specifically the advertising applications and considerations that have taken place across this last year so going further into that and just very much asking that question thinking more broadly about programmatic advertising and how we've seen AI show up in those spaces. Do you have any additional commentary on what that connectivity has looked like and how people who are media buyers or planners have already had a lot of just general exposure?
EM: You're right – it has had a much larger societal impact than just in the advertising space. Over the last bunch of years, we'll say even rewind seven or eight years ago in programmatic buying, real-time bidders have always been in the machine learning space trying to predict proclivity to interact. So, if you want to try to maximize your click-through rates (programmatic buying) that's usually done through some sort of machine learning algorithm that has been accumulating signals from users over time. Pretty amazing space.
We've had conversations around Google Smart Bidding. We're seeing it now in Google's responsive search ads and their responsive display ads where they go ahead and restitch assets together in compelling ways, and then measure the response for different user segments. That's all been done through large models through machine learning and through what they're doing with their algorithms.
Meta is in this space all the time. Lookalikes as an example are definitely in that space. So lookalike audiences which have been around for a while that concept is pretty pervasive in Meta and they're doing a lot of that work over there.
What I think happened in the last year however is that this is all in kind of the last mile buying that we've seen. It's bid management, it's proclivity to buy like these low-level signal type things. What we're seeing over the last year is that you can put it more in your workflow and the workflow can be very labor intensive. We collectively as an advertising community definitely find that as digital marketing gets more and more sophisticated, the levers to pull become more and more complex. They're more sophisticated. They interact together a lot. I'm going to guess a thousand different settings in any given campaign. And if that's the case if there are all those thousand different settings, there's a reason that there are ways you can set it in position A versus B (just as a concept why people have set those up that way). And it's not always obvious what the combination of settings are to reach your goal in the most productive way.
So, people in digital agencies, digital advertising tend to spend an awful lot of time figuring out what those pure combinations are. How do we link these audience settings together with my bid strategy? How do I get my creativity optimized properly for the space? And configuring those settings has been complex and labor intense. And I think AI is coming into play here in a great spot. It's helping people make those decisions.
Even with all that there's still a lot of labor in just the creative content and content generation to communicate a special. You've got a four-bedroom apartment and there's a move in special and all those pieces to make that sound natural and bring that to a user that's been challenging over time as well. And, again, very labor intense when you look at the wide landscape of the number of accounts that agencies typically manage that takes a lot of time and energy to just manage all that creative generation. AI has done an amazing job here simplifying that entire process, helping you understand the settings that make the most sense and helping you effectively communicate with your client base with compelling assets, compelling content and creativity.
NN: I think probably most people picked up on that when you're saying, by settings you mean different parameters or tactics that you can leverage, different data integrations that you can make selections around. To what extent do you think that there has been a shift in since the birth of (and I'm calling it the birth of generative AI) but I mean the accessible form that you were talking about at the top of this podcast has there been any notable shift? Because so much of the focus has been on generative AI. Have we seen any differences or distinctions whether it's AI specific or automation specific that you've perceived in the last 12 months?
EM: You're talking about out in the space. Am I seeing any signals that are coming back that say that the way we operate is different?
NN: The way that we operate with advertising or programmatic advertising in mind.
EM: I think the answer is absolutely that there are very simple ways to do these things. You can actually have fully AI generated campaigns through settings through messaging and that can exist now. I think there's some challenges with that but it does exist. Which means if we are trying to compare agency A to agency B. The labor savings that you can gain from AI and automation is astounding if you can really lean into it now. The first thing I want to say is I'm very cautious with how we do that. That wasn't a statement to say widen your margins immediately adopt AI and I think automation is solid and we'll talk about why automation is very cool in a little bit I'm sure. But AI can do so many things but it's a little bit non-deterministic. If you can tolerate a little bit of error - if you can tolerate a tone of voice that may or may not match with your client. I would never use it for a large brand that has an established voice. But if you don't have that case and you're using a lot of just data to power your information and trying to get a message out there, you can do an awful lot with it. That just changes the way we execute. So that we can offer more complete solutions at a lower labor point which means that margins can stay where they are. Businesses have to run. At the same time the end-advertiser gets a better experience, more long-tail on the advertising side and it's not paying for it on the other side, so it's prohibitive on the cost side.
I think there's a lot that's gone on in there. You can definitely feel that the entire shift inside the Google Meta space Microsoft is leaning in extremely hard with OpenAI in that Google has an AdWords version that's fully AI driven. It gives you recommendations and walks you through the entire process. That's very difficult to use if you are trying to put together a bespoke campaign with deterministic outcomes. You're still going to get involved and you're going to make those choices but it's definitely coming and it's definitely getting better all the time.
NN: There's been a lot of excitement around the potential of AI across this last year and we've established that. And I think that the automation component is something that I know you have a lot of energy around the words automation for whatever reason in the advertising world don't always resonate with everyone. It's not everyone's first priority. They have manpower. They have bandwidth. They have systems that are already in place. How do you think the word automation is being received today in a way differently than it was in the last couple of years?
EM: Well, I think we're in a spot where it's almost a necessity there are agencies that are doing it and doing extremely well and those agencies are able to offer a deeper service at a lower price at maybe the same or lower price point than where we've been before. So those services are now very feature complete. I think when you think about automation. First, we'll kind of define a little bit of the difference between AI and automation. AI is a natural thinker it will come up with something for you. It will feel creative whether it is or isn't it will feel creative it may feel new and it's a little non-deterministic (you don't know exactly what you're going to get out of it when you start).
Automation is effectively a rule system that is very deterministic. You should know based on whatever you put in the outcomes will be deterministic you will know what you'll get. So, for example you can do a lot of things if you had a data set of specials taking those specials and converting them into ad content it's pretty straightforward. We know what we would do as people. We can model it and if we model it we can automate it and I think that's a really important piece. So, if you are an agency and you can think about the decision trees that you go through as a person and your best practices because you're you're probably going to tell your clients exactly what your best practices are. “We do XYZ in this scenario and that's what makes us an amazing agency”. We do these things and if you can say those types of words why you make choices when you see certain data sets, then you can automate it. You can say, “When the weather changes and I'm representing an HVAC client I prioritize heating systems. When the weather gets cold.” I've actually heard people go the other direction because heating systems sell themselves when weather gets cold. So, you might want to say that our secret sauce is to reduce bids because you don't have to be as aggressive.
So if you can articulate who you are and what your strategy is in scenarios then from that point forward you can go ahead and just set up your rule systems and let them run. I think it's had a negative connotation because we have two hemispheres really in advertising today and it's new. With digital advertising whenever anybody thinks advertising we think creative we think creative minds creative writers we think Mad Men a little bit. We think you know just splashing ideas on a wall and there is a place for that. But the digital advertiser today has a very technical job and that's new. So now you've got this one person that is expected to be both creative and technical.
So, what I see us doing with automation over the next several years is really taking that technical piece and simplifying it by putting it through the rule systems. Allowing creativity to flourish, that side of the house actually continues as is. But reduce the requirement of low-level settings knowledge on the digital advertiser, allowing them to be more on the creative side how they communicate with clients. I think AI has a great narrative on that side of the house specifically because that can be very time consuming also and as inspiration and creativity, super cool spot. For AI, I can go and ask it a question. I can get an idea, it could be my finished product, it could be just an idea spark and I love that about that side on the creative side of the house. But when it comes to the execution side that's a great spot for automation. The execution is going to take and process all that data, look at the scenarios out in the marketplace right now, bring in signals like the interest prime rate and the weather and the products that are in inventory that your account has and make decisions like you would make as a person and just stop requiring you to go and configure all these things over and over again.
We just had a conversation. It's actually really timely. Just before we got on we were talking to a friendly partner that talked about how they do Facebook. And the cycle over month over month with campaign renews in Facebook is a many person multiple day job. And it's not a creative job. It's a workflow job and those are the places where automation just a no-brainer in my opinion. Why would you put your people through that? Even if you had the manpower it's such a tedious and painful existence it's not fun. It's a burnout machine. It's prone to error. And not because people are significantly prone to error but because it's tedious, it's mindless, it's repetitive and I would love to use a computer personally anytime that that scenario comes up.
NN: You've given so many examples that clarify that the industry has progressed to a place where a larger slice of advertising responsibility falls into some of those that, they're not low value but sometimes the feeling is that they're mindless activities or they're less creative or they naturally shouldn't be sucking up so much manpower time. Are you finding that most agencies clients that fall into doing this type of work are making the shift and embracing automation to the extent that it's available?
EM: It is a mental shift. The partners that I've gotten a chance to work with, when I get to work closely with them you can bring them through that process and it is a process. It's going to sound like a simple thing but if you're a software developer a “for” loop or an “if then” else statement is pretty natural it's common language. Automation works on similar concepts. I've got to loop through every one of these products. And if this scenario happens then I wanted to do this other thing and in natural speech that works but to set up automation can be daunting if you don't think like that.
Once you have the idea of how looping works and iteration and logical blocks like that you can do a ton with it. I don't think it's as foreign as it seems. When you first hear it and if you don't communicate it in a proper way it looks really technical. It doesn't have to stay that way though.
Expressing the concept once you get there, the savings are so wildly important. Businesses transform once they get there. If you can show them the win early. Like if you can give them a small small morsel of success, they can literally cross off labor intensive things like that thing we just talked about the Facebook campaign refresh. If you can show the value of that early in the process then the sky is the limit.
NN: The analysis of achieving efficiencies and cost savings is a conversation that's had with a certain buyer. And it may not be just somebody who strictly and exclusively does media buying. Of course the role of what it means to be a media buyer is going to vary depending upon the organization and what their responsibilities are. But I take it back to my own experiences working in a holding company being a media buyer. I got exposed to basically publishers and publishers that were doing outreach to us and I talked to them. They had some suite of sites or portfolio of sites that I could buy from or they're representing a single solution. I wasn't the one who was responsible for understanding the economic upside. So who are you talking to about this? Because some people may fall outside of their wheelhouse who's really going to be bought into or do you need to talk to who cares about this when you're reaching out to an agency?
EM: I mean, ultimately it's going to take executive sponsorship for us to take that leap with somebody. But I think before you even get to that point you've got to win the hearts and minds of the executor and the ad strategist. Long-term like overall we will have to also sell the economics. All software automation processes their process change, process change has cost, software has cost. Like all those things do end up having cost so you will have to get to the executive sponsor side. But we always start first in showing the person that does the job how their day will get easier. And once their day gets easier then the convince-up is a lot easier. It's basically saying you know they don't have to do that campaign refresh anymore. You can put them on more high value things. Have them talk with your clients more. Have them express your value. Let's have them grow budgets. Have them grow penetration and make your solution stickier. Let's mitigate some churn. They can be put on more high value business tasks which are really thinkers’ tasks anyhow. And get away from the tedium.
So, if we can show the value of the platform with people that do execution then talking to the executive sponsor that's going to actually fund this thing becomes really straightforward. Your team knows that they will not have to do this tedium anymore. The five person three-day job at the beginning of the month, every month has just gone away. When it comes to the economic side definitely we're talking more with the executive sponsor inside an organization.
NN: Are there markers of an organization that is going to be more open to embracing the future of automation compared to an organization that seems less prepared and therefore less inclined to actually receiving what automation can do for them?
EM: We definitely see that there are a lot of clients that if they are confident in what they offer as a service, if they can articulate why they are different than somebody else. You've got agencies always buying for the same clients and the same business and it's a competitive space. If you can articulate your value prop that means you have a set of values and rules systems that you operate by. That is different than somebody else. That's the first foundation that I always look for, someone that knows their business really well.
The next side is that they have an appetite for growth. If you're already doing it and you don't have an appetite for growth, probably going to be settled in where you're at. But if you have an appetite for growth and you don't want it to linearly scale headcount with growth that's another really good candidate for that because you can get buried in this. Just even finding talent even if the economics started work finding the talent can be very challenging when you have large growth goals.
And then finally progressive companies that want to get into the longer tale. I'll tell you a story from my background. I worked on an ad platform and then I left that company or I was about to leave that company and I was entertaining other places and I went to an audience company. I was going to sign with them and when I got to the end of that interview my little anxiety kicked in. I was having a hard time putting pen to paper and what was clicking with me was all this audience stuff is amazing. I believe in it. I was just at a place that was doing digital advertising and if they had brought those audiences to me to our organization I would have believed in them just as much there. And they still would never have seen the light a day just because we were underwater with the amount of labor it took to manage specials and to make ad copy updates, and to manage negative, keyword lists and keyword bids and etc. just all those settings that were just taking up all this time. Meanwhile running a very lean business that wasn't a gigantic margin business. Adding more on top of that advertiser's plate was very difficult.
I want to help agencies do more for their clients while not burying them. That is definitely my goal. And what I would see is bringing them something like more audiences, more complexity. And I just want to make sure that we can make that as simple as possible and help them through this journey. I think when we do that their end advertisers (like the end advertiser that's outside the agency or brand) will feel the benefit of more compelling and more complete advertising and yet not feel the pain of the expense.
NN: If I hear everything that you've said and it sounds like we try to automate some things but we're clearly not doing it with the sophistication that we need so we can stay competitive. What would I need to do at this moment from a prerequisite perspective to best embrace the type of automation that you're describing?
EM: Well automation is definitely a walk I don't think there's an all-in-one. You just made a statement that you know we're doing some things. I think anybody that's doing anything is off to a great start. You understand the benefit, you look for opportunities not to be repetitive and redundant. And now all we have to do is just more of those things as we go. Understand your business. Say you have a team of 35 strategists, understand where they spend their time. Where are those time syncs that pull them away? And evaluate each one of those things to say, “Is it value added?” Humans don't add a great multiplier effect to copy and paste. We have to do it because there isn't another thing that copies and pastes for us unless you get into automation.
So, if you're really taking settings from one account and moving it to another doing a small trial on a singular account and then want the worst thing that happens is it's successful. And then you have a thousand other accounts to go apply that setting to. We want to make sure that that's a very easy thing and no longer a deterrent to going and doing the right thing.
My last venture before this I was writing on a software and in the digital strategist pod and I like to work late. Every 7 o'clock on Friday I would look over and there were two people sitting at their desk with just a desk light on. And they were upset and frustrated and you could just see. And I'd find out what it is and someone had just asked them for a very complicated problem late in the day that needed to be done for Monday's solution. It's miserable. And I really wanted to make sure that that wasn't a requirement of the job. And I think as employers we are looking to help our digital strategist look for places that you can improve their life. I think when you improve their life it will reduce in some ways meaningful and tedious labor. So help them there and then it's great for the bottom line as well.
NN: I want to ask you a question about talent, in particular. You mentioned talent earlier on, what type of talent are we looking for, so we can best embrace automation in our agencies and advertising organizations?
EM: Well this is going to be one of the bigger challenges in the space like I said there's that two hemispheres of an advertiser today. There's the creative hemisphere and the technical hemisphere and that person is rarer than I wish they were. So we have to talk about where we can level people up and help them make them superhuman with systems. I think there's a way that you can help someone that's not like I'm not strong in creative writing. I'm strong on the technical side and not so strong on the creative side. What I see is the AI generative side really can help augment that type of person. So if you can find someone that's biased to the digital world and weaker on the creative side, AI is a really good inspiration tool. Now they still have to be careful to look for tone of voice and compliance and things like that but they at least will get some ideas from the AI side.
If you've got someone on the creative side but they’re a little bit weaker on the execution and the technical side, I personally feel like automation isn't a great enhancement for that user. They can put something into the machine in the beginning with very simple inputs. And then let it do all of the marshaling and all the construction of the complicated campaigns and A/B groups and A/B sets and audience management and all those other settings that are more on the technical side.
A technical marketer that has a creative flair is the ideal state and I think where you'll find is that everybody's got some bias. I am extremely biased on the technical side and not so much on the creative writing side. And then there are counterparts that I've met that are completely the other direction. And I believe that we are in a spot right now where you can level up either of those two sides appropriately and find your complete person with some augmentation.
NN: If I am thinking about the opportunities that AI can afford my organization, maybe I'm a media buyer, maybe even some of what you just described in the last answer doesn't represent me strictly. Me being whoever is responsible for media buying but maybe a larger team. Maybe there's three, four or five people that cover the work that you're describing both technical and creative and what do you have. Who would I want to go to inside of my organization to get buy-in for some of these things? And that almost feels like a repetitive question because you said earlier on I guess if I'm the hands-on keyboard person then I would naturally want to go to somebody at the executive level or somebody who's responsible for bigger buying decisions. But I'm just curious what types of questions or information do I want to put forth or ask about so that there is a greater potential that we can be doing smarter media buying in our organization's future?
EM: I've seen a few of these different cases and I'll kind of talk about what we've seen so far. As people want to be able to self-service more, you just described a scenario where there are four or five people involved in a singular process. And you know at each of those handoffs there's some overhead in the communication. I put together my part and then I had to write a document about it to send it off to the next person who had to read that document to do their next part etc. There's overhead in those transition spots. And if you were someone that really wanted to start to take singular ownership over more of the process. We've seen that as one of the use cases for trying to get into this transformative state. I'd like to do this on my own with these clients. I'd like to be able to go from end to end and I think we can save money because we'll have a shorter cycle.
If you can set your processes and automation up right, it can work for each of them. And now you've got the capacity instead of for one account. And I know we just use simple numbers there instead of five supporting ones. Now you have five supporting five and they're all whole. And the nice thing is the learnings can get shared at that point. So, if strategist A finds that a certain bidding strategy or placement position or audience works extremely well and they know why, they can say in these scenarios this works better that can get distributed and that can become best practices across all five of those users because it's an automation system. And they'll just immediately reap the benefit of that knowledge. That knowledge share becomes not just conversational not just in a document but actually in a living breathing piece of software that will help you propagate that through your entire business. It is a very cool spot to be there.
So, I've definitely seen people who want to expand the scope and have fewer interfaces with handoffs from team to team that work extremely.
NN: What we're going to see moving into the future and I'm sure many organizations have already seen this trending, is that smaller organizations will see people who have hands on keyboards building more of these use cases and rationale for why change and embracing a combination of AI and automation is important to do now. I think there's a lot of exciting things to come in the year ahead. So with that Eric thanks for the time and having this conversation with me.
EM: I had a great time today. And the last thought I kind of want to leave with is that it levels the field for a lot of people. When you just mentioned the smaller agency. I think the smaller agencies can get overshadowed by large agencies that cast a big shadow. Take the same concept with large brands casting a big shadow over small brands. If you can do excellent work with the same pedigree of work that you could dedicate to a large brand but you can now dedicate it to even smaller brands, what an amazing spot to be. It kind of levels the playing field. The product stands on its own and you can tear up your narrative there. I think it's a great leveling system here with automation and AI.
—
NN: Thanks to Eric Mayhew for his wisdom on automation and AI, there's so much change we're struggling with right now in the ad tech space. Most notably the deprecation of third party cookies, reconciling MFA sites, managing brand safety, CTV measurement and more. On top of all that, organizations need to level up on any existing traditional systems that remain to ensure that they're inching out the competition. Now is the right time for agencies and brands to do that deeper analysis of their existing text deck to see what automation can do for them. That's all for now. This is Adtech Unfiltered with Noor Naseer. Another episode coming out real soon.