Automation is here to stay…
But how is it impacting the field of digital marketing?
Frederick Vallaeys, co-founding CEO of Optmyzr, the leading and award-winning PPC management platform, is here to share his insight.
- Why automation is good for advertising
- The risks associated with PPC automation
- How to maximize your ROI on advertising
- Does ChatGPT impact PPC?
- Automation layering
- And more
Mentioned in this episode:
Voiceover: This is Performance Delivered, Insider Secrets for Digital Marketing Success with Steffen Horst and Dave Antil.
Steffen Horst: Welcome to the Performance Delivered Insider Secrets for Digital Marketing Success podcast, where we talk with marketing and agency executives and learn how they build successful businesses and their personal brand. I’m your host, Steffen Horst.
The topic for today’s episode is how automation is appending the field of digital marketing. Here to speak with me is Frederick Vallaeys, who is the co-founding CEO of Optmyzr, the leading and award-winning PPC management platform. Frederick is a Silicon Valley entrepreneur, author and leading influencer in pay-per-click search marketing.
As one of Google’s first 500 employees, he helped pioneer PPC marketing as the company’s first AdWords evangelist. He is a sought-after industry thought leader who contributes to leading marketing publications and conferences, and is called upon by journalists, writers and podcasters for his industry insights and vision. Frederick, welcome to the show.
Frederick Vallaeys: Thanks for having me on. It’s great to be here.
Steffen: Frederick, before we start talking about automation, and digital marketing, tell our listeners a little bit more about yourself. How did you end up at Google and being one of the first 500 people at Google and what led you to founding Optmyzr?
Frederick: Yeah, wow. So I graduated Stanford in 2000. And during my days at Stanford, I had started dabbling a little bit in PPC. So go to was available. So you could for the first time ever, by keywords and just pay for the clicks you got. So it was very cheap. And as a college student, I was like, that’s kind of an interesting way to do ads. And maybe I can sell a few things. So I started selling video cassettes.
And then coming out of college, I took a digital consulting job basically with Sapient. And that lasted about two years. Because as some of you may remember, in 2000 the .com bubble finally burst, and I lost my job. And so I sat down, and I wrote down a short list of companies that I thought were exciting. And there was this one company that a few people had heard about, but not that many yet. And it was called Google.
And it seemed to have some momentum. So I put my name in a hat. And a lot of Stanford people went there. So kind of had some connections into it. And they actually offered me a job in March of 2002. And they wanted me to review ads. And I was like, hell no. I’m not going to like, I’m a Stanford engineering degree, like, I’m not going to review ads for you, like, offer me something better. That was a big financial mistake, because the stock option was a little bit cheaper back then.
But come August of 2002, they came back and they were like, well, we still want you to review ads. But we also need someone who speaks Dutch. And because you were born in Belgium, and you speak Dutch, we might have use for that. I was like, okay, well, I’m really excited about this Google thing. So sure. I’ll come in to review ads for you. And that’s how I started at Google. And then that lasted 10 years. It was fun.
I mean, you know, Sergey and Larry, they used to do the personal introductions of everyone who got hired. And so Sergey was the one introducing me. And he had a bottle of Heineken in his hand. And he was like, oh look, there’s Fred from Belgium, from the Netherlands, just like this Heineken beer. And I was like, ah, Sergey, come on, like Belgium, Netherlands, two different countries. It’s not the same. Yeah, but it was good. It was good time.
So we had lots of fun stories from back in the early day. And then after 10 years, I decided I wanted to strike out on my own and solve some of the problems that Google wasn’t really solving as they were becoming a bigger company. So by the time I left, it was about 60,000, full-time employees. So that was tremendous growth from the 500 when I joined.
And I dabbled for about a year, having an agency and kind of like figuring out what the real pain points were, and then started writing some AdWords scripts at the time. And then I met my co-founders. And we decided to build a software company to make the lives of digital advertisers easier.
Steffen: Interesting. Fast forward to now. What is the state of automation in PPC, in particular?
Frederick: Yeah, I mean, I’ve been doing Optmyzr for 10 years now. And it’s certainly everything’s come a long way. Right? So when we started Optmyzr, there were a lot of bid management companies out there, because Google, for one wasn’t really necessarily automating even very basic things like bid management. And so over the years, more and more things got automated. Bidding, keyword targeting finding similar audiences.
And I think for advertisers, it became this question of, like, am I still going to have a job like, I’m a PPC marketer, like, am I in the wrong field? Should I be studying something else? Like, am I wasting my time if Google’s going to automate everything? So we continue on that path, right. And I’ve written two books now on that topic. And it’s really interesting, because I still fundamentally believe that humans are an essential part of this mix.
But it’s how do humans use these tools and new automations from Google? And so, I’ve always said, you know, you must do automation, like there is no reason that a human should be calculating the bid to set when that’s just a formula, right? It’s like your conversion rate or your profitability goals. You put that in a spreadsheet. But do you want to do it a million times for every keyword in your account? No, of course not.
Just go and let the machine do that. But what you as a human can be focused on is that more strategy and strategic advice. Be real marketers, right? Like, we’re so used to being spreadsheet geeks in digital marketing. And we have to kind of go back now to let’s be marketers, let’s figure out what message connects with another human, and how we sell stuff that way.
Steffen: I mean, that basically, however, means that we’re looking for different people, right. Back in the day, so I mean, I’ve been in paid search since 2004, when I joined a small agency back in Germany, and the people that would do paid search management and managing the campaigns that was really focused on understanding a platform, understanding what had to be done to identify keywords, write ad copies and things like that.
We now talk about people kind of moving almost out of that area of responsibility and more into the strategic part. From my perspective, that probably means we need more educated people, because strategy usually is a little bit of a different discipline. Would you would you agree, Frederick?
Frederick: Yeah, exactly. It’s, I mean, it’s a whole different ballgame, right. And so in the first book that I wrote, I basically created a human skills to three roles that we’re pretty familiar with as humans. So one of them is the PPC doctor. One of them is the PPC pilot, and one of them is the PPC teacher. And so I think you as a marketer, you can choose which of these fields you go into.
And if you go big and become the teacher, you know, maybe you learn how machine learning works. How AI works, so that you can manipulate those systems to do what you need them to do for you. And if you’re more on the client management side, and you’re really like figuring out well, what’s this, what’s this advertiser trying to achieve? And what are the problems they face within the company?
Maybe you’re more like the PPC doctor, who has to have that conversation and figure out like, what are we trying to get better from? And then what are some of the options in terms of the medication that might work well for this patient. And of course, the medication in PPC, that’ll be your, you know, 15 different types of automated bid management you can choose.
But it’s putting those pieces together. And if you’re more like, hey, I want to go back, and like the old school PPC, maybe you’re more of the PPC pilot, where you set your monitor to systems and you’re more of a safeguard, in case the machine makes a crazy decision.
Steffen: Interesting. Now, a second ago, you talked about, there’s kind of a, or there was, is a scare by people. So whether that’s the person that mentioned campaigns, or even advertising agencies, that Google or any other media company, you know, whether that’s Bing, etc, would make them obsolete. Why is automation good for advertisers? Why is their thinking potentially not right?
Frederick: I mean, so I was talking to another PPC expert the other day, and he said it really well, I think. He said, digital marketers will not be put out of a job by chat GPT. But they will be put out of a job by other marketers who use chat GPT. And so it’s sort of that collaboration of human plus machine. And we see that all the way back to the days of chess when Garry Kasparov, he was beaten by a computer that played chess really well.
But one funny story that a lot of people don’t realize is the engineers at IBM, the ones who were making Deep Blue. They actually were days away from the match against Garry Kasparov. And the machine was not able to win every chess game. And they sat down and they said, well, we’re trying to solve the wrong problem. We’re not trying to build the best chess-playing computer.
We’re trying to build a computer that can beat Garry Kasparov. And so they just fed it information about how he played. And all of a sudden, the problem was so much more limited that they were able to beat Garry Kasparov. And so it became this thing about oh my god, the computers are so good. They could win at any game of chess. That was, in fact, not the case. And then Garry Kasparov, he turned around, he was like, well, if you give me a very basic computer, that can help me with some prediction.
And you give that to me, as a really good chess player, I can beat anything in the world. And that’s the whole point, right? So if we are good at digital marketing, we know the strategies and now we’re aided by these by these other systems, we can really advance quite a bit.
Steffen: I totally agree with that. It’s the automation plus the human being. So the human brain is a much greater output than just the automation or just a software solution itself, basically. There’s always an advantage to having the human there. And I think also you need to have a human there, because certain things at least until now, software couldn’t do.
Steffen: So what are the risks associated with PPC automation?
Frederick: Yeah, I mean, so aside from the things that the automation can’t really do, there’s also the fact that it’s, it needs to learn. I mean, most of these systems today are based on machine learning. And that implies that learning needs to happen for the machine to become proficient at a task. And oftentimes, it’s frustrating, especially if you have small to midsize advertisers. They don’t really want to spend $10,000 for the machine to figure out a bunch of different keywords and a bunch of different audiences.
And yes, it’ll see the ones that don’t convert, and eventually stop showing your ads to them. But that’s $10,000 down the drain, right. So that’s kind of a frustration. So the big question then becomes, how are you the PPC teacher? Who can already tell the machine listen, this is an audience we’ve tried, don’t bother going there. Spend that money somewhere else. So that’s one risk, right?
And then the other risk is that the machine, the way it makes decisions is very much data-driven. So it’s going to make strange decisions. And one personal example, this happened to me last week, I was doing a video about Google’s automatically applied recommendations about redundant keywords. And so by accident, I left the automatically applied system on.
And then the next week, my marketing person talks to me, she’s like, well, did you know that the keyword optimizer your brand name, the one that has so many conversions, all of a sudden, it’s been paused. It’s been removed from your account? That is the one keyword that Google decided to remove because I thought it was redundant. It’s like, seriously, guys, like this is the keyword with the most conversions. And so then we’re like, okay, well, maybe the cost went down.
Maybe there’s some other reasons why Google did this. No, cost went up, CPC went up, cost per acquisition went up, the number of conversions went down, and the machine just made a bad decision. And that’s the risk, right? That’s why we still need to have humans overseeing these things, and putting some safeguards in place.
Steffen: Yeah, I totally, totally agree. And then there’s the other side on this. It’s like, you need to have a certain amount of actions before machine even can start making suggestions on optimizations, right? So there’s this statistically relevant amount that you need to have before you can say, hey, here’s the information. Now, tell me what I should do so to speak.
Frederick: Exactly. And so the machines have gotten amazingly better at that. And that’s one of the funny things is that in shopping campaigns, Google leads literally zero conversions from your data, before they can start automatically setting bids for you, because they figure the things that you sell are probably in some way similar to things other people have sold. And they can draw inferences from that and correlations and so but again, that’s their business, that’s everybody else’s business, that’s not your business, right?
And the more that you can teach the machine what you care about, and what a conversion is to you, that’s going to drive the best results. And that’s really the big shift then is, as humans, you know, how do you communicate the real value and the real goals to the machine. That’s your value ad these days. It’s not setting the specific bids. But it’s telling the machine where those bids come from, from a business perspective.
Steffen: And if a machine obviously, also doesn’t know what happens outside of what is in the platform, and this is where you as human being have to feed the machine with information, right? So what are competitors doing that is not necessarily visible, you know, within the search landscape, for example, that is read out by your software solution as an example. Now, Google and any other media company constantly make changes.
They make changes, you know, how their platform work, they add, remove targeting opportunities, etc. Now, last year, Google came out with Performance Max campaigns. Should advertisers use the super-automatic campaign type? Or can they get even better results with the old-school manual campaign?
Frederick: Performance Max is like this great example, right. And it’s kind of the reason that I wrote the second book, it’s called Unlevel the Playing Field. And the premise of that book was that Google is making all of these automations, like Performance Max, that make it really easy for an average or a mediocre advertiser to get pretty decent results, let’s say average results. And for many advertisers, that’s enough, right? If you don’t have to spend a lot of time doing these things, that’s great.
That’s better than before. But if you actually spend a lot of money on ads, getting a 10% more efficiency, like that can be a big deal. Or if you’re in a super competitive space, like yeah, you might want to work with an agency that knows what they’re doing, because that’s the thing that makes or breaks your business. And so yes, Performance Max, absolutely test it out, see what kind of things it comes up with.
But we also strongly still believe in running traditional shopping campaigns, in addition to Performance Max. And Google even recommends that you should still run this in supplement to your search campaigns, your content campaigns, etc. So it’s really supplemental. And it’s a very easy way to get into a couple of different channels within Google.
But, but it’s additive. It’s not a replacement. And then even if you run it, there’s still optimizations that you can do like adding negative keywords. Like having multiple Performance Max campaigns, where you have different targets or different goals based on the types of products you sell. So there’s many ways that you can optimize it.
Steffen: I think this is a great example of agencies or even marketers that go the extra mile, right. As you said, you could just use Performance Max campaign and you get okay results, you know. But if you want to squeeze out the last little bit of your budget, if you want to get that extra 5, 10% out of it, then if you, in parallel, have a manual approach, that can really get you something extra. And that’s, I think that’s also the value of a good agency, right. So a good agency that understands the platform, in this case, the Google platform that will know what to do in order to maximize your return on advertising spend.
Frederick: Exactly. And the other kind of cautionary word, I think, for advertisers. So they may try it out. And it may see really good results in the beginning. But see, here’s the thing, Google takes all of these existing visitors, people who come in through remarketing, people who come in, that you’ve already been exposed to through your SEO efforts and your social media efforts that Google may recapture that, and they may not really break it down.
So the results look really good. But it’s really on the back of all the other work that you’ve done. So that’s one word of caution. And the second word of caution is all of a sudden, something external changes in the marketplace, in the industry. And maybe the system doesn’t pick up on it, maybe it doesn’t respond the right way.
And because you have so few controls, there’s really nothing you can do other than pray that somehow the system figures it out. And that’s what’s scary, right? So don’t put all your eggs in one basket of automation, because it’s great when it works. And when it stops working, and you have nope, no buttons to push. It’s like it’s the most horrible feeling you can have.
Steffen: Yeah, yeah, I totally agree. Now, we talked about it before we jumped on the podcast. Chat GPT is something that is in the news, kind of daily. I think yesterday, Microsoft announced a cooperation with open AI for the search product. Google obviously is having their own kind of endeavor with an AI to kind of bolster up their search offering. Is automating the last automated frontier in PPC, that of creative writing, is that impacting PPC? And if so, how?
Frederick: Yeah, absolutely. I’ll eat my own words there. I mean, I was one of the people who said, it’s going to be quite a while until the machine gets really good at writing ad text and doing everything for us. And then all of a sudden, very, very quickly, the GPT systems like Chat GPT, and a number of others and lambda from Google, got really good at writing ad text based on some inputs that you gave it.
And so now we run some experiments that we take the say, 10 existing headlines from a responsive search ad and we say, write me five more ads like this, or headlines like this. And boom, it gives you five more headlines. And it’s fantastic. They’re pretty spot on. So this is the type of thing where all of a sudden, we’re like, well, maybe we don’t even need to be that creative anymore. Now, I think we have to wait a little bit longer term and see how this plays out. Because Google in its early days, we had one product mad libs, okay.
So it was, you know the game Mad Libs. So they call it, we came up with something that was called ad libs. And we would automatically put like certain words into a template of an ad. So if you had a restaurant, we’d say your restaurant is in Los Altos. And it serves Thai food, and it has a four star rating. So we did this for all the restaurants. And all of a sudden, every ad that shows up on Google looks basically the same.
And evolution stopped happening. And evolution for Google is quality score. It’s putting some stuff out there that’s generated by humans. And sometimes it’s creative. Sometimes it’s a mistake. But you look at the click-through rate, and you can start to see what resonates with users of Google.
And that way, the system learns, okay, this is better, this is worse, and it starts to show a variety of good ads. If you take that variety out of it, then who knows? I mean, maybe we’re building an AI to feed the AI. And that is classic example too like, somebody just told me like, I’m writing my resumes using AI. But guess what, you write your resume with an AI and you give it to Google, well Google uses an AI to read your resume.
So now you got these two AI systems talking to each other to see like, who do we actually invite for an in-person interview? And it’s almost like PPC and SEO. Like, it’s headed that same way, right. It’s kind of an uncharted territory, we’re gonna have to see how this plays out.
Steffen: Yeah, I mean, that’s, that’s really interesting to think about it that way. But it kind of underpins still what you said earlier. I still think that human being is still required in the entire picture of PPC management. That role is just changing. And I think you’re right, when you said earlier, it’s probably changing more towards more strategic role than necessarily the kind of the hands on keyboard and doing every individual step that is required to create a campaign, to kind of analyze what the performance is then to create new ad copies, create new ad groups.
You name it all these tasks that in the past individuals had to do. That probably will disappear more and more. But that human input is still required and is probably really important. And that’s where the difference then will basically happen, from my perspective, whether you have a great campaign or whether you have an OK campaign.
Frederick: Exactly. And I think kind of what you’re getting into here is one point about experimentation. The agency or the advertiser who’s the best at experimenting is going to be the one that lives. Because they can, they’re the most nimble, they can most quickly figure out what works.
And in the last three years or so, like, we’ve all learned how quickly the world changes around us. And if we’re not ready to like, drop everything at a moment’s notice and change strategies completely, like we could be in dire financial straits. And so experimentation is like one of these key points, and then I think being ready to change ad scale. So if you come up with that new strategy, hey, GPT is a great tool to say take every ad text that I’ve written, and make it more friendly.
Or add this kind of messaging to it and boom, it spits it out in a matter of seconds. And then if you figure out how to connect that back into the ad system through tools like Optmyzr, then you’re going to be in that leading position against your competitors.
Steffen: Yeah, yeah. Do you have some examples of what advertisers can use Chat GPT for?
Frederick: Yeah, I’m still in the early phases of playing with that. But some fun things we’ve done so far is I had my support team, and they were all like, hey, can you translate some strings for us in Finnish in Portuguese. I tried Dutch because I speak Dutch. And like, it was perfect, the translation was fast, and it was perfect. So we can now start doing customer support in different languages, that was very difficult to do in the past.
It wasn’t difficult because we have people who speak those languages, but taking a string and translating it is gonna take you a few minutes. Now it’s a couple of seconds, and you help them in their language. We I also started experimenting with feed optimization. So we have tools that let you put in structured data, run it through a template and generate the ads based on what you have in inventory.
But oftentimes, companies will find that, you know, their landing pages, their data sets are not that clean. They might be using a color name like burgundy, or the product name includes the brand, and they can’t figure out how to split the brand into its own column. GPT is really great at that. You just say, here’s the URL of the website, what brand is this about, and boom, it comes back. Perfectly knows what brand it is. And then it comes back with a color burgundy. And then you can say, now give me a more common term for burgundy.
And it’ll say red. And you’re done, right. Now you take that structured data, you feed it through your other mechanisms. It hasn’t been that good at bidding and budgeting anything to do with numbers. It’s still struggling, but it is getting better. And so but that’s also the areas where we already have great technology, by and large from Google from Microsoft. So we don’t need quite as much help with that.
Steffen: Now before we come to the end of today’s podcast episode, you know, you talked about automation layering in your book. Tell us what that’s about.
Frederick: Yeah, and so you know, to kind of like, maybe put a bow on this whole thing. Like we said, automation is here to stay, you should use it to do the best that you can. But you have to be careful, right? It’s just like, if you hire another person on your team, they might be fantastic. But you kind of have to look at the work they do. Because sometimes they misunderstand something, maybe they took something you said the wrong way.
And then they go down this path. That’s not the right one. It’s the same with automation. And so you can either say listen, I’m going to review every single piece of work that the machine does. But you’re going to spend more time doing that, due to the scale than any work you’ve done in the past. That doesn’t make sense. But if you can come to, you know, look at a script, or an automated rule, or a tool like Optmyzr, and you can say this tool is going to enforce some rules and some limitations and some restrictions.
It’s kind of like PPC insurance. It’s your automation that you’ve defined what it should do that controls the automation from Google, which is much more opaque. You don’t know exactly what it’s doing. But your own system can now monitor it to make sure that you get the best outcome in the end. And that’s what automation layering is all about.
Steffen: I think you’re absolutely right. Being kind of the, or monitoring what what the automation does, from A to Z is a lot of work. But if you identify which parts basically have an impact on when they break, for example, you know, where do you see a problem the quickest, then you can focus on those things rather than on the entire picture of the multiple automation you have in place.
Frederick: Exactly. Yeah, monitoring systems in place first as a first layer. And then you have an investigative layer secondary which says okay, we found something that’s broken, tell me why it’s broken and then it can, it says okay, well, these are the five things you probably need to do to fix it. And then again, technology is going to help you fix that faster than in the past. So it’s just kind of a different way that you put the systems together to do the work, the strategic work that you’ve decided needs to be done.
Steffen: Perfect. Well, Frederick, thank you so much for joining me on the Performance Delivered podcast, and sharing your knowledge on automation within digital marketing. In this case we talked about PPC. If people want to find out more about you, about Optmyzr, how can they get in touch?
Frederick: Yeah, so Optmyzr is spelled a funny way. o p t m y z r.com. So that’s the tool that we have, we got a two-week free trial. So a great way for people to check it out. And then our team will talk to you and kind of show you what we can do. It’s a full-fledged free trial. So even if you just want to run an audit on your account and not pay for it in the end, that’s fine. I’d love for you to see what we can do.
And then personally, I’m at Silicon Vallaeys. So that’s a play on my last name and Silicon Valley. So at Silicon Vallaeys on Twitter, and I’m trying to be a little bit more active in other places as well. And of course, you can find me on LinkedIn, Frederick Vallaeys. You can connect with me there. But yeah, thanks for having me on the show. This has been fun talking to you.
Steffen: Great having you. And as always, we’ll leave that information in the show notes. Thanks, everyone for listening. If you liked the Performance Delivered podcast, please subscribe to us and leave us a review on iTunes, or your favorite podcast application. If you want to find out more about Symphonic Digital, you can visit us at symphonicdigital.com or follow us on Twitter at Symphonic HQ. Thanks again and see you next time.
Voiceover: Performance Delivered is sponsored by Symphonic Digital. Discover audience-focused and data-driven digital marketing solutions for small and medium businesses at symphonicdigital.com.