E100: Frederick Vallaeys

Richard Hill:
We made it, 100 weeks straight, 100 episodes, 100 amazing guests, and here we are with our 100th episode. I'm Richard Hill, the host of eCom@One and welcome to our 100th episode. In this episode, I speak with Frederick Vallaeys, CEO Optmyzr. Fred was one of Google's first employees and fast became AdWords first evangelist almost 20 years ago. Now running Optmyzr, a PPC SaaS with billions of dollars in ad spend under management, Fred literally helped create the AdWords platform and has over 20 years at the forefront of the PPC scene. I've watched Fred's talks over the years at the many conferences I've been to, and his tools, softwares, books, and frameworks have shaped thousands of agencies and hundreds of thousands of accounts and billions in ad spend.

Richard Hill:
In this episode, we talk eCommerce and ads. Should every eCommerce store be running ads? Fred's thoughts on the current AI landscape. We talk automation layering and why it's so important. Feeds and Fred's top three recommendations. Fred talks me through his ideal account structure and Fred's opinion on the future of the PPC landscape. All that remains is to say, thank you. Thank you to you, our listeners, for getting us to 100 episodes. Now, let's head over to this fantastic episode.

Richard Hill:
This episode is brought to you by eComOne, eCommerce marketing agency. eComOne works purely with eCommerce stores, scaling their Google shopping, SEO, Google search, and Facebook ads through a proven performance driven approach. Go to ecomone.com/resources for a host of amazing resources to grow your paid and organic channels.

Richard Hill:
How, you doing Fred?

Frederick Vallaeys:
I'm good. Thanks for having me, Richard. 100th episode, that's amazing. Congrats on that.

Richard Hill:
Thank you so much. Thank you so much. I think when we sat here in the office literally about a week before COVID became a thing, we didn't really think we'd be sat here, I think, almost two years to the day pretty much when official lockdown happened in the UK. So it's pretty crazy, 100 weeks in and out every single week from the start. So yeah, absolute pleasure to have you on as our 100th guest.

Frederick Vallaeys:
Great to be here.

Richard Hill:
So for those that don't know, Fred is very much an absolute legend in the PPC space, to put it bluntly. Fred won't know this, but a few events that I've been to many, many years ago, I was one of many hundreds, if not thousands of people sat in the audience listening to Fred talk about different things, but particularly some of the things we're going to talk about today around Google shopping and automation. I think it was Hero Con or Hero Conference London 2015, '16 maybe, '17, two years there. I saw you do a couple of talks that really inspired me and some of the things that we do in our agency are very, very much, they're absolutely based on things I learned then and obviously things have definitely moved on an awful lot. So we're going to dive into a lot of that stuff. I'm, for one, super excited to get stuck in. So I think, how did you get started in the PPC industry?

Frederick Vallaeys:
Yeah, I mean it was a little bit random. So I've been doing PPC in some shape or form since 1998 when I was in college. And a way for me to make a little bit of money was to buy these video cassettes at Blockbuster, I know I'm dating myself, like what is video cassette? What is Blockbuster? But yes, I was buying these cassettes and then Blockbuster was basically selling them really cheap before they were allowed to sell them. These were cassettes that the consumer would have to pay 100 US dollars for, I picked them up for $5 and I flipped them for like $20, $30 on eBay, but sometimes I needed a little bit more juice to my auction. So I was like, "Well, how can I drive more traffic to this?" And I found out about [inaudible 00:03:45] and so I started buying some keywords and doing my first PPC campaigns, very small scale.

Frederick Vallaeys:
I graduated from college, ended up doing some engineering work, then the dotcom bubble imploded. I was looking for a job and it was this promising company, Google that was just starting, just starting to get some momentum. I was like, "Yeah, I see myself working there." And they happened to need someone who spoke Dutch to translate Google Ads, AdWords at the time, into its sixth language.

Frederick Vallaeys:
So I was very fortunate. I lived in the Bay Area. My dad had moved to the Bay Area for a technology company there and so I moved and was able to work at Google, worked on Google Ads from the very start really, did that for 10 years as the evangelist, claim to fame, was involved in building conversion tracking, was one of the original people on the Google AdWords editor team that built that, was on the team that acquired Urchin, which is now called Google Analytics, was on the quality score team that built that for seven years, lots of hatred from people on that one, but nowadays it's much better. So I'd say, yeah, I had a good time working at Google and that's how I got into PPC.

Richard Hill:
Oh, wow. So literally started selling a few things, doing ads yourself, picking up products. I think that will resonate with a lot of listeners because I think that's how the majority of eCom stores start, I think it's that, "I found this product, if only I could sell five or 10 of them a week or a month," and the next thing you know, you hear stories of, that's how very much a lot of businesses start and then now they're doing maybe a million pounds a week or whatever it may be. Obviously in your instance, you jumped ship on the eCom side and on the selling online and went to work for Google. Big, 10 years.

Frederick Vallaeys:
Yeah. And it was so crazy because I think of my days in college, I was very much supply limited, which is a problem that a lot of stores face nowadays, surprisingly 20 years later, but I couldn't get enough video cassettes from Blockbuster to satisfy all the demand, so PPC was great, but it was also, I wasn't using its full capability. And then when I joined Google and I was like, "Hey, I could be an affiliate and get people to sign up for eBay or to get people to buy car insurance." These are almost infinite supply type situations and that's where I got really good at PPC and all of a sudden I was maxing out my credit card buying these ads and so I had to call the credit card company like, "Hey, can you extend my line because I'm not going to get paid for a little while, but I need to run more ads because this is working so well."

Frederick Vallaeys:
And so it's fascinating too, because then I'm actually working at Google on the Ads team and I'm running all of these ads at the same time. And so eBay Netherlands reaches out and they're like, "Hey, you're our top affiliate for the Netherlands market. Do you want to come and speak at our conference and tell other people how are you so successful?" And I'm like, "Oh wait, can I actually talk about this?" Because I work at Google, some people might not be cool with that. So I go and talk to Cheryl Sandberg, my boss at the time, and so she's like, "Yeah, this is great because AdWords is so nascent, a lot of people don't know about this yet. You're living proof that this works." And so me being an advertiser, but also working in the company, working on the technology, that's really what kick started my career in terms of know how the product is built, but also know how to use it. And that's a thing a lot of people don't really have.

Richard Hill:
Yeah. Yeah. Got it from both sides. It's funny you say about asking you to speak, I remember a talk that I saw you at, I think it was the second talk I saw you at, and you had your, I think you had your iPad in the background there and you were along the lines of, "Hey Google, how are my ads doing today?" Do you remember that presentation you did, was it six years ago? Seven years ago?

Frederick Vallaeys:
Yeah. So that was Hero Con London. So I did the world premier of, sorry, of the Alexa skill-

Richard Hill:
Alexa. Hey Alexa. Yeah. Yeah.

Frederick Vallaeys:
Don't say her name because she's all over my house, but yeah, basically, and that was so cool because we took the PPC investigator capability we had in Optmyzr and we turned it into a voice command. And the whole vision, and this is where automation's headed too, but the vision is me as the human in the room with my clients, my clients are always going to ask me super detailed questions and my mind is at the strategic line, but if they want to know about, "How many conversions did you drive last month and how is that different from the month before?" Well, sure, I can find that data, but I'm going to have to poke around some reports, it'll take me 45 seconds to get that. Or I could just ask my voice assistant who instantly gets that through the API, and that was the whole vision. So you have this little robot helper sitting at the table with you and your clients and you can ask it questions.

Richard Hill:
Yeah. I remember seeing that. I was like, "Wow, that's so cool." You literally, you imagine getting up in the morning, you've just had your orange juice and your whatever, and you're just saying, "Hey," I won't say it out loud, but you know, "Hey, how did we get on yesterday? What were the sales? What were the ROAS? What was the conversion value, the key things we want to know?" And as you're just having a little bit of orange juice and whatever, your Weetabix, it's like, "Yeah. Okay. We had a pretty good day." Saving that 45 seconds of logging into your ads or your report tool of choice sort of thing. Yeah. Fantastic.

Richard Hill:
Okay. So eCom, eCom@One, obviously eCommerce stores. Should every eCommerce business be running ads? What would you say to that?

Frederick Vallaeys:
Yes. I mean, so listen, but there's different kinds of ads, right? Should you buy a Super Bowl commercial? No, probably not unless you have a lot of money, but when it comes to Google Ads and search ads and shopping ads, well, yes because, and this is the fundamental thing about Google, it has been about answering consumers' questions at the moment that they have that specific need. And if you have that product that satisfies their needs, why wouldn't you want to be in consideration at that point.

Frederick Vallaeys:
But then it does get a little bit more complicated, obviously, if you have a newer product, something nobody's searching for. Yes, you should advertise, but you're probably not going to see that much success buying keywords because nobody knows what to look for. And so there's different avenues and channels, but yes, absolutely, advertising will grow your business.

Richard Hill:
So no matter who's listening to this episode, Freddy's saying you should be running ads. But let's dive into that a bit more because obviously you touched on branded, unbranded products, big discussions there on ads, obviously huge volumes for your Nike trainers and size 12 Nike whatever, whatever your flavor is, but obviously unbranded shoes is a lot harder to sell for example. But I think AI, that's obviously a very hot topic, has been for, I saw you talk about that back then sort of thing and it's very much, our agency very much uses your tools, the suite of Optmyzr and obviously a lot of those things, I think when I first looked at what you were doing there, obviously saw your talk and been looking at the different tools. And then I just looking at some of the scripts back then, what we could do, I was just blown away really.

Richard Hill:
It's just... And that was however many years ago, seven years ago, but what are your current thoughts on AI? Where are we now? Fast forward, I was impressed six, seven years ago when you're talking to Alexa and you could run a script that would automatically create a scag from a keyword that converted two hours or 10 minutes ago, and that was great seven years ago, but now fast forward, what are your current thoughts on AI in the PPC landscape?

Frederick Vallaeys:
Exactly. I mean back then it was really very transactional. It was like a script. So if this, then that. The AI in machine learning is much more about finding that signal that's just hidden in massive amounts of data and that takes massive amount of computing power. And so obviously the computing power keeps going up, the amount of data that people and consumers and searchers put into the system, which Google then has access to, that's growing. So you put these two together and Google can do some pretty amazing things.

Frederick Vallaeys:
So I mean, I'm a fan of AI. There's no question about that. I'm a fan of technology, but the scary thing is that, does AI get to the point where it makes the human no longer necessary in this space? And so we see that a little bit in terms of the movement that Google's having towards, while we have a shopping campaign, but now we have a smart shopping campaign and now the smart shopping campaign's going away and it's going to be a performance max campaign. And by the way, a performance max campaign is not just shopping ads, it's every ad, it's YouTube, it's Discovery, it's Maps, it's everything. And Google's fundamental ask for these new campaign types is, "Well, tell us your budget, tell us your goal, and then just go away. We'll handle it from here." And that's a little bit scary, right? And that's where I think a lot about AI, how do we help the AI not fail and actually deliver on the promise that Google puts out there?

Richard Hill:
Yeah. Yeah, because I think very much that will resonate with a lot of agency. I think we'll probably go back three years ago and smart shopping, we're like, "Oh no, no. It's not going to be very good. No, no, no, no. It's not what we want to do." And then it's like actually, it's actually quite, fast forward, I don't know, a couple of years ago, the difference, I think, I don't know what your thoughts are, but the difference in the last 18 months or so with smart shopping is almost night and day.

Richard Hill:
But obviously layering in then, okay well, are we just going to set it up and leave it? If everyone's doing that, then that's a certain playing field out there, but you talk very much about automation layering. So, I think it'd be good for you to focus on that for us, adding in and layering in different things over and above just pressing the button and letting it go.

Frederick Vallaeys:
Right. And you just said it levels the playing field, and so that's the title of my new book that just came out, Unlevel the Playing Field and automation layering is a big part of that. And so exactly as you pointed out, the problem is that your average newbie, small advertiser can actually be very successful with Google Ads in a way that they were never able to until AI got to be as good as it is today. So anyone can come in say, "Hey, here's my product catalog, my budget, go, be good results." And by the way, these good results, are they really good results? Or is it just a lot of remarketing that Google's hiding in the mix which is actually your SEO team did all the hard work and now you're just-

Richard Hill:
Yeah. [crosstalk 00:14:11]. And they're losing their budget, yeah. Yeah, probably, yeah.

Frederick Vallaeys:
Exactly. But so okay, say that you do smart shopping campaigns, we're now going forward, you're going to be doing performance max campaigns. While an automation layer is something as simple as bringing a little bit of intelligence to the mix here, do you sell different kinds of products with different margins, different seasonalities? If that's the case, you probably want to unlevel the playing field a little bit by having some sort of an automation layer that puts these products into different performance max campaigns. And that doesn't require automation. That could be as simple as saying, "Listen, I'm going to just filter by category for one campaign and another one," but where it does get into automation layering and some of the capabilities that Optmyzr's working on is, "Well, let's look at the performance, right? So let me put all my products into a performance max campaign, but then let's monitor each of these products and if we see that certain products are not driving individually a good return on ad spend, we can actually filter those out. We can automatically apply a label that says, exclude or whatever it is."

Frederick Vallaeys:
And so they're coming out of this campaign, and an easy way to understand this is I was working with this client a really long time ago now, but they were selling T-shirts. And if somebody did a search for blank T-shirt, so T-shirt with no logo, no nothing on it, blank, T-shirt, the orange T-shirt, every single time it was the orange T-shirt that showed up. And I was like, how many people need an orange T-shirt when they generically search for a T-shirt? That's not normal-

Richard Hill:
No many.

Frederick Vallaeys:
You want black or a white or some standard color, right? Google keeps preferring this one, and so what's the work around? Well, you got to take that orange T-shirt out of the mix because it's just taking up all your impressions, not delivering the conversions, and that's tedious. And so one of the roles that humans play in this new world is really about monitoring. It's the pilot flying the plane, is the plane getting the right inputs or do I need to tweak some of the dials you do with that T-shirt? But to do this across a catalog of, say 100,000 products on a daily basis, that's tedious. I mean now we're spending more time managing this automation than we did [inaudible 00:16:30].

Frederick Vallaeys:
And that's what blew my mind, right? That's why I started writing scripts, but then I also realized a lot of these automations that you put in place, they take up as much time unless you automate stuff. And so that's why you have to get to the concept of automation layering, which is take what you know you should do to help the campaign be better, but then find ways to do those things in a very easy and a very simple automation. So it could be as simple as an automated role, which you can set up for free in Google or it could be an ad script, or it could be a tool like Optmyzr, or it could be really high-end sophisticated machine learning that you build, but you have a spectrum to choose from. And so Google does the high-end, very expensive machine learning work. And even with simple things on top of that, layered on top of that, you can make those things perform better.

Richard Hill:
Yeah. Yeah. I mean that's pretty much what we stand for as an agency. This is very, I think if you're listening now and you're setting up performance max or setting up smart shopping and you're letting Google do its thing, yes, you probably are maybe doing better than you were a year ago. Well, depends, there's other factors there, but if you're not layering in different elements, you've got layers in there, especially on eCom, especially on eCom. There's so many variables. If you're talking about a 100,000 SKUs, 50,000 SKUs, a lot of SKUs, each SKU having literally dozens plus variables.

Richard Hill:
And I think with Google shopping, feeds, obviously feeds is a very, very hot topic. You can't do one without the other. So what would you say about shopping feeds? What are three shopping feed automations that you'd say is really, really key? Because obviously if you are using smart shopping, and maybe not doing any layering and any other things, it's literally, you're doing the same as everybody else potentially, but with the feed you could really get a headstart, kickstart, jumpstart, whatever you want to call it on everybody else by having a really, really strong feed. What sort of things would you say most people miss in a feed, maybe two or three things?

Frederick Vallaeys:
Yeah. I mean the first thing is obviously the title. The title and the price, right? Those are the two big ones because that's ultimately what shows up in the shopping ad. And that's a funny thing, so a lot of people say, "Hey, we're losing control." And yes, you're losing control in Google Ads, a shopping ad, you can't make text for it, but you can control that text through the feed, through the pricing. I call this concept like you should be managing Google Ads at the periphery of the system. So the details of what happens, that's up to Google, but the periphery, where you put in the feed, where you put in your goals, your targets, that's still really important. And then Google will handle the little details associated to that. But so title and a title tip, think about how consumers search. If you think about automotive, I think my most recent car, the color is obsidian.

Richard Hill:
Yeah.

Frederick Vallaeys:
And I'm like, "What's obsidian? Oh, gray." I think it's gray. Right? But so how do people search reflect that in titles?

Richard Hill:
Yeah. There's probably no mention of gray in the default manufacturer listing, is there? Just it's obsidian and obviously everybody knows what that is.

Frederick Vallaeys:
Because to the manufacturer, this is boring. Gray, who wants to buy a gray car. That's not sexy. But to me as the consumer, well I want a car that's gray and if I'm shopping between Mercedes and BMW, while one calls it this thing, one calls it the other thing, how do I find the thing that I actually want? So put yourselves in the shoes of the consumer and use terminology that is reasonable for your prospect.

Frederick Vallaeys:
And then price. Price is the other great thing. So I used to throw the slide out, and basically it was a bit optimization question. And the point was simply that there were two competitors, one was a penny more than the other. Now my argument was, take that penny that you're more expensive and actually reduce your price by 10 cents, so now you're significantly less expensive than the next, the top one. And that is actually going to make your advertising so much cheaper, right? So you're potentially making less money on the sale, but you're going to have less of an uphill battle to get these clicks because consumers are so price sensitive. So look at price as a great thing. And again, this is-

Richard Hill:
Yeah, that's-

Frederick Vallaeys:
... an AdWords optimization, but it is something that influences.

Richard Hill:
You've got to know your numbers for that, haven't you? Well, obviously anything AdWords wise and when you're selling huge volumes of product.

Frederick Vallaeys:
Right. And that's where you want to work with, I mean someone who has these integrations across, for you I'm sure this is a challenge sometimes because you can't really control the pricing of the vendors that you work with, but at least can you get them thinking in those terms, right? Because they come to you, they come to me, and they're like, "Hey, fix this AdWords campaign." And I'm like, "Sure. I can spend $5 more on this click, or you lower your price by 10 cents. Which would you rather do? Which makes more sense strategically?"

Richard Hill:
It is a conversation we have with the majority of prospects that are coming through our agency, where obviously if we look at our ad account, we can see that, we can look at their feed, we can look at everything. Okay. There's definitely opportunity in there, but if you're 5%, 10%, 15% dearer than the nearest two or three people, then that is a challenge no matter what we layer in, we could be layering all week and we might not move the needle sort of thing. Yeah.

Frederick Vallaeys:
Exactly. And then maybe the third one, so custom labels, right? So we are huge fans of more complex account structures. So rather than having one campaign, have multiple campaigns and those campaigns can have different targets. But that's sometimes complicated to do, and this is where automation layering is brilliant because you can set labels, custom labels, and then based on those custom labels, you can split stuff out. Now these custom labels, they don't have to be static. So we've actually done experiments where we take the current return on ad spend of a product, and we put that in as a label. So we sort of bucket it, so we say this one's between say 100% and 200% ROAS, 200% and 500%, etc, etc.

Frederick Vallaeys:
And so now this sits in a campaign with a target return on ad spend that's close to that bucket, because if you put the target return on ad spend, the target too far from the current performance, it usually doesn't perform that well. Now, as the performance changes because of some externality in the market, we actually pick up on that, we rewrite the custom label field and so that product will move between campaigns so that its bid strategy stays where it should be at.

Frederick Vallaeys:
Now one thing, and I'll just probably most of your listeners are really familiar with this, but supplemental feeds. These are great because I've run into customers and I've said, "Hey, this is what you should do," what I just explained. And then they say, "Oh yeah. But every week or every day we get a new feed from the merchant and they override everything that I've fixed." That's fine, use a supplemental feed that you control that layers on top, again it's a layer, on top of everything else.

Richard Hill:
So having multiple feeds, multiple supplemental feeds for different, whether that's products that have got a certain ROAS, whether that's products that you don't want in the feed because they may be problematic or in and out of stock or... Yeah. Okay. So I think, yeah, the ROAS custom label, I would imagine there won't be many people doing that. I would've thought that would be, that's a good one. That's a real good one, really good one.

Richard Hill:
Okay. So we've got the feed up and running. Now we've got our, maybe, 10,000 SKUs. We're selling, let's say we're selling branded product for now. What would an ideal account structure look like? We've got maybe 10,000 SKUs. It's a difficult question, because there's a lot of variables there, but what would an ideal account structure look like?

Frederick Vallaeys:
Exactly.

Richard Hill:
For shopping.

Frederick Vallaeys:
In PPC it always depends, right? No, and listen, I think rather than focusing on the 10,000 products, it's really about the variability in performance between these products. And so one example I like to give, if you're... Sorry, I'm trying to think of a British brand, is it Dixons? They sell all kinds of-

Richard Hill:
Yeah. Yeah, yeah, yeah. PC World, well it used to be, it's called Currys now actually. Yeah.

Frederick Vallaeys:
Okay. Yeah. See, I mean I wish that I was able to travel once in a while. I was actually going to be back in London in two weeks.

Richard Hill:
Yeah. It's been moved, hasn't it?

Frederick Vallaeys:
It has been moved. So hey-

Richard Hill:
July the 18th now. Yeah.

Frederick Vallaeys:
So eventually I'll see what happens-

Richard Hill:
Are you going to be there? Are you planning on being there, yeah?

Frederick Vallaeys:
Yeah, yeah. I'm doing a keynote at that one again. So yeah, a couple more months to figure it out. But yeah, I mean in terms of figuring out you're a big retailer, think about your seasonality, right? That would be one big factor for how you split out your campaigns. I'm not a fan of the single campaign structure under almost any circumstance, unless you literally only really have one type of product you're selling. I just like the control that you get from being able to set different targets, play on seasonality, play on externalities. I mean I keep giving this weather-based example, but it's really critical.

Frederick Vallaeys:
I mean say, okay, you sell automotive parts, car batteries will sell in higher quantities when it gets really, really cold because car batteries tend to die at that time of year. That's something that you know, Google's machine learning probably doesn't account for that. So this is, again, an unleveling tactic, but you can say the day before I know it's going to be freezing, I'm going to boost up my target CPA or I'm going to lower my target return on ad spend because I know it's going to be made up for by unusually high performance in terms of conversion rate. Google's not going to know, but I'm going to trick the system into doing the right thing for me. You can do that if you put everything together.

Frederick Vallaeys:
I also have often talked about the GRIP structure or a SPAG structure. So basically GRIP means groups of individual products. So the idea is that you put every single SKU into its own biddable product group and the idea from that stems from the fact that, like your example, Nike sneakers size 12, right? I have Adidas sneakers size 12, I have Adidas sneakers size nine, I can run a report and Optmyzr has great reporting capabilities for this, but I can say regardless of where stuff is put, actually give me a report based on the size of the shoe. And now I realize that size 10 is relatively common for men. Okay. So that's a really popular shoe size, so maybe that's where I need to focus my bids a little bit higher. I can't do that if I have my products all mixed together, I need to have them individually split up. So that's why I'm a fan of the GRIP structure.

Frederick Vallaeys:
The other thing I'll say about structure is that Google will always tell you simplify, simplify, simplify. And a lot of people understand that to mean have a single campaign with as many things in it as possible. Okay? But Google is talking to millions of advertisers and these millions of advertisers are not as sophisticated as you who chose to listen to this podcast. For you, what Google will actually acknowledge is if you have business reasons for structuring into different campaigns, that makes total sense. So the whole example I gave about seasonality, about weather impacts, about things that might be on clearance or that have different product margins, put it in different campaigns. But for your average low-end advertiser, put it all together because it's just going to confuse you less and that's good.

Frederick Vallaeys:
Now, the other thing that plays into this is conversion tracking and people think, "Oh, if I have five campaigns, my conversion data is split between those and that's bad." But Google's machine learning looks at the conversion action and so long as that conversion action is connected to all five campaigns, it doesn't care. It says, "Oh, we know this is the likely conversion rate for this conversion action." And no matter what campaign something's in, it's going to update that correctly. So you don't have to be worried about that.

Richard Hill:
Yeah. That's brilliant. That's brilliant. When you talk about, I was thinking a lot of people that are selling shoes and things with size, I think quite a common thing I see is, let's say you're selling a certain type of shoe and that shoe's available in size six, seven, eight, nine, 10, 11, 12, but right now you've only got nine in stock, size nine. If somebody types in Google just that shoe name and they end up on that page, clearly there's only a size nine available, you probably don't want that ad showing to everybody because the likelihood of everybody landing on that page who has a shoe size nine is very unlikely. So, what's an easy way to deal with that or what would you say about that?

Frederick Vallaeys:
I love that example. I mean it's a variety product and extend it to cars, if somebody's looking for an SUV, but you only have it in one model or one color, one version of it, that's less likely to sell than if you have it in every variation. It is a bit more complicated. But again, you could use your feed and use one of the custom labels to generate variety, a variety classifier. And only if the variety classifier is labeled as high, I have all sizes in stock, then you could say, "Okay, I want to put this in my high priority shopping campaign where I have a more aggressive bid because I know I'm likely to convert."

Frederick Vallaeys:
Maybe you don't want to turn off the ads for the size where you have fewer sizes available, but maybe you don't want to spend as much on it, maybe you want to prioritize it. And again, in the case of shoes, there's probably two variations of that shoe. Is it more likely people care about the size or the color? While the size is must have, the color, that's a negotiable. So then show me the one that I can actually buy in the color that maybe wasn't my favorite, but that I might still end up liking.

Richard Hill:
Yeah. Yeah, because I think that is quite a common challenge and people don't know how to do it, especially fashion sizing is quite a challenge full stop for eCom stores, obviously people buying the wrong size or maybe buying two sizes, but if they haven't got the closest size in, they might not buy anything rather than one thing. Yeah.

Frederick Vallaeys:
One thing I see, and I'm sorry, I know you should be asking the questions, but I'll interject, I'm really curious what kind of returns do you see in the UK [crosstalk 00:31:16]?

Richard Hill:
In terms of-

Frederick Vallaeys:
In fashion.

Richard Hill:
Do you know? I don't know the numbers off the top of my head. They are high. They're in the, I don't know.

Frederick Vallaeys:
I've seen like 70% in Germany, which-

Richard Hill:
Yeah. I seem to think it's like 35%, but I wouldn't know exactly, but somewhere, I think it's pretty high. Yeah, yeah.

Richard Hill:
Going back to the weather example, I think we actually have clients that sell batteries and obviously when it gets cold, your car typically may have challenges. But also when it gets really hot, we have obviously a slightly different temperature gauge to where you are in the world right now, California. But obviously when it gets hot, we find that maybe bring out your little toy out the garage, whether that's a motorbike and you need a motorbike and it's been stored in the garage for the last 10 months and it's like, "Oh, it's not starting." Or a little MG Midget sports car, whatever it is, it's been in the garage for 10 months so we've got that other end of the spectrum with the weather. Obviously when it goes above maybe 23, it's like, "Oh, let's get the soft top out the garage." Obviously when it goes cold, so obviously tied into actual weather, that's something that we've been able to do for many, many years here. Is that something you seem more and more prevalent now, ads relating to the weather, the temperature?

Frederick Vallaeys:
Yeah. I mean, and I love your example there and it's such a cliche example of bidding by weather, but ultimately it's so relevant in so many scenarios. But be flexible. I mean our job, our unleveling is the fact that we can take this example and we can say, "Well, okay, the weather doesn't actually impact my business, but I sell allergy medications." So I could look at the pollen count and that's just a different input to a similar system that makes similar decisions. What is that for you, right? Because that data's out there and you can do amazing things with it.

Richard Hill:
Yeah. I think, yeah, mind starts to go just when you're saying, we've got a client that has about 300 and I'm not sure exactly, but about 300 veterinary stores and one of their things is fleas, fleas treatments. So when spring kicks in, the weather starts to get, it's nice and warm and then certain animals and a lot of cats and dogs start to get fleas more. The chart for certain products goes, vroom. So obviously, it's got that use case there where... Yeah.

Frederick Vallaeys:
And I think what's fascinating about all of this is global warming. And so I'm in California, January, February, it did not rain. That's fairly unusual. Usually it rains at least a bit here. And that's different, but next year, who knows what next year's going to be? And then we have the pandemic related uncertainties and now we have the Ukraine invasion and the world is so crazy in so many ways. We almost need these systems, these automated systems that can say, "Hey, while we didn't expect flea and tick season to start until, say March, but yeah, we're having a hot January and all of a sudden the fleas are laying eggs. So let's spin up that campaign."

Frederick Vallaeys:
And the more you can have that automated and these rules codified in some way through a script, yeah, it'll look at some source data that you have and it'll take care of that for you. And then if it gets cold again, well it'll turn it off and that's the beauty because you now can be more focused on the strategy rather than everyday monitoring, like, "Oh, are we there yet? Are we there yet? Are we there yet?"

Richard Hill:
Yeah. Waiting, waiting, waiting, automation has kicked in while you're working on something else. Yeah. Yeah. Amazing. Yeah. And I think that was the big aha moment when we started using some of your tools, really back in the day, many, many years ago.

Richard Hill:
So obviously we've got a lot of account managers, strategists that are listing in and we've got many, within our agencies, and obviously in-house people managing their ads themselves. But so, what would you say to those guys about routines? We refer in our agency to some daily, weekly, biweekly, monthly, and that's how we refer to, and we've got a list of about 250 things from over the years and obviously that's changing and moving down. And obviously that's a crazy amount. I think it is over that now, but there are there that you can adjust, but what would you say a good monthly routine looks like on shopping?

Frederick Vallaeys:
Yeah. I mean the fundamentals, right? And I highly recommend people look up your 200 point checklist if you've shared that anywhere. We have these checklists as well, but again, I'm not going to say anything surprising here, but figure out, what are your search queries? That is one thing you can still look at with shopping. So you can make negative keywords for that. You can't target positive keywords, but you can do negatives. Are your bids performing? Is your target return on ad spend delivering the right sort of volume? Target return on ad spend too often misunderstood, ROAS is not a goal, right? Profit is a goal. Revenue is a goal. ROAS is a mechanism to help you achieve that.

Frederick Vallaeys:
I've told this story many times, so I apologize to anyone who's heard it before, but there was an agency that was tasked with a 400% return on ad spend after the client came to them for the first time. And six months down the line they're like, "Well, you know we could get you more revenue at a 300% at a lower return on ad spend? Why did you ask for 400?" And the client was like, "Well, the last agency gave us 350 so we figured we should ask for more." But that actually led to bad-

Richard Hill:
A reduction in conversion value. Yeah.

Frederick Vallaeys:
Exactly. The volume just goes down because you're not as competitive, so figure out what that sweet spot is. And then the other thing, so a lot of people, impression share is a metric that a lot of people looked at during manual bidding, but impression share still exists even when automated bidding is running for you. So, do you have campaigns on a target return on ad spend with a high loss impression share due to rank for example? Okay, so that tells you that your target return on ad spend is too conservative. You could get more volume by being more aggressive. Now it's a question for you, can you be more aggressive? Do your margins support it? Do you need to do something to your landing pages, to your shipping providers? But these become really interesting conversations to have with your client.

Frederick Vallaeys:
But the fundamentals that you look at, it's still bidding, it's keywords, placements more and more. So as these performance max, it includes remarketing for you, it shows up on different websites. So negative keywords are just as important as negative placements, but hey, you can put a negative placement in, well, yes, actually you can, you can have an account level negative placement. So, leverage those 65,000 or so negative account placements they can have to read out the bad traffic. If you don't know what those are, Optmyzr's building a tool that will look across our, we manage something like $6 billion in ad spend, so we will look across all of the placements that we see and find you the ones with high cost, low conversions. Boom. There you go. A list of 65,000 good negative placements where you shouldn't be wasting your money and we'll update those continuously. So...

Richard Hill:
Yeah.

Frederick Vallaeys:
Yeah. And then feed optimization, right? So look at disapprovals. Again, this is not directly AdWords, but it's still what feeds into AdWords so it's hugely important.

Richard Hill:
Yeah. It's always one, I think, gets missed. You look at an account. That feed was oddly set up, it's not obviously always, but it's very common, isn't it? You think, "Oh yeah, somebody set that feed up. It doesn't work here anymore." Or somebody set... We're using obviously there's a lot of different tools, we used that tool when it was set up, but then adding in those additional keywords into the titles, or the custom labels. Actually, we've only got one custom label and you may be getting that, A, return that's profitable, but if you're not spending a certain amount of time on that feed, checking certain things, especially disapprovals, you are missing a trick kind of thing. It is one thing that I definitely see, just gets left, left, left, every now and then there might be a little mini meltdown because Google sent an email saying, "You've got 30 days to fix something apart from that." Quite often it does get missed I think. Yeah. So-

Frederick Vallaeys:
I mean some of the things that I wish I could do a little bit better through automation, but I've struggled to get them to be automated is, say with the titles or the categorization of products in a feed, doing a A/B test between that, right? So, okay, change the title from one thing to another and then-

Richard Hill:
Put it in a feed. Yeah. Yeah. There's a couple of feed tools that do do that that I'm aware of. I know one that definitely does that where it split tests within their tool. Yeah.

Frederick Vallaeys:
Exactly. And then images too, right? It is a funny thing, so I was talking about the orange T-shirt as being a bad thing. 10 years ago, that was a bad thing, nobody wanted to see the orange T-shirt, but nowadays companies like Nike, they have whole departments dedicated to colour optimization. And so it's funny because they're not necessarily looking for the colours that will look good on some of these sneakers when they wear them, but they will look at, "Well, people they scroll through on an app with all the shoes, what's going to stand out? Well, it's not the white one, it's the garish one with all sorts of patterns. That makes you stop." So that optimization's being done. And that also informs, I mean how do you stand out against your competitors with the image that you show in a shopping ad?

Richard Hill:
Yeah. That's so true, isn't it? Purple seems to get me every time. "Oh, what's this?" I end up looking. I've got a bit of a thing for purple, my purple notebook is right here.

Frederick Vallaeys:
Nice. Purple is a royal colour, right? Are you related to the royal family in any way?

Richard Hill:
What can I say? No. I'm a huge Prince fan.

Frederick Vallaeys:
Nice.

Richard Hill:
I think that's one of the originals back in the day. Remember Let's Go Crazy?

Frederick Vallaeys:
Yeah.

Richard Hill:
I think it's literally, you've just had the Super Bowl, haven't you, a couple of weeks ago? And I assume you may be familiar with the very famous Prince halftime, he was about to go up, he was doing the halftime show and he was absolutely caning it down with rain and the guys came over to him and said, "Prince, what do you want us to do?" And he goes, "Make it rain harder."

Frederick Vallaeys:
Love it. His show was amazing.

Richard Hill:
So yeah, yeah. So, okay. So, obviously you talked about there's ROAS, obviously the conversion value ultimately, ROAS is a thing that gets bandied around a lot and we talk about it every day probably in our agency completely, but without a certain level of conversion value, obviously you're not shifting stock, you're not paying the bills potentially. So we've got conversion value, we've got ROAS, but what other metrics do you think are there that really merchants should be looking at to see the success of their Google Ads campaign?

Frederick Vallaeys:
Right, and this is not the easiest one to implement, but ultimately what a business cares about is profits and revenue. And so that requires bringing in margin data. And so, when you're running these shopping campaigns, and this applies to any sort of campaign, could be lead gen, could be shopping, but we'll give the shopping example, right? So too many advertisers, they pay for the ads, the ad leads to a click, the click leads to a purchase, and that's it. Okay? So you reported the value of that basket and then nothing else. How you're going to win in PPC going forward is to look at what happened after that. Sometimes referred to as the offline conversion. In retail, it's the value adjustment mechanism that you have.

Frederick Vallaeys:
So what you can do is simply, with that order, you pass in the order ID, which is a unique ID to Google. And for the next 55 days, you get to communicate back to Google and say, "Hey bot, I told you it was 100 quid of a purchase, but half of it was returned so let's adjust that down to 50." Or, "Actually that customer came back, they signed up for the email list, they responded to the offer, and they bought another 200 quid's worth of stuff. So let's adjust that." And that wasn't from PPC, but PPC initiated that, so let's increase that value.

Frederick Vallaeys:
And that's how you win, because the more that you use automation, whether it's... And automation has many forms, so it's bid automation, but the way that Google selects what queries to show your ad for, that's automation too. If it knows that you actually drove a higher value, it's going to be more willing to show you on more expensive queries. And that's okay for you if you actually drive more value from it, right? It's still going to be profitable. So I think that's the key next wave for all advertisers is figure out how to communicate to Google what happens after the initial conversion that you report. They call it value based bidding, we've written articles about it on our blog, but that's the next wave of optimization.

Richard Hill:
Yeah. That is brilliant. That is, yeah, we need to do more. We need to talk more about that another time. I think that's a whole... Yeah. We've got-

Frederick Vallaeys:
And it's a hard one. I mean I think everybody understands this is what they should do, but like, how do you get to it? Google has luckily made it a little bit easier, especially for lead gen. So it's called enhanced conversions for leads, just came out a few weeks ago. Look that up if you haven't, but it doesn't require the GCLID anymore. GCLID was this Google Click ID that you had to capture and put into your CRM. No longer necessary. And then I'm always surprised because I think a lot of eCom advertisers don't know there's a think called value adjust, which is a variation of offline conversion import and is actually much easier.

Richard Hill:
Yeah. We've got a few clients that have got quite a few clients that are bringing the margin through, and you don't see that very often in reality.

Frederick Vallaeys:
That's brilliant because, hey, you guys actually [inaudible 00:45:32] profit, that's a big deal. You're-

Richard Hill:
Well, this is, it always throws me because sometimes I'll forget and I'll look at the account and it's like, "Okay, they're getting two ROAS. That's not very good." But actually it's very good. It's very, very good. Yeah. Yeah. It always throws me. We've got some very big account, well some of our biggest accounts have got margin coming through and then every now and then I forget and it's like, "Oh, crikey. Oh, that's okay."

Richard Hill:
Okay. So, where should we go next? So, I think experimentation. There's a lot of things we've talked about out there, new things coming through, layering in this, layering in that. How important is experimenting and where do you think somebody should start? So maybe they've got their smart shopping campaigns running and then they're thinking, "Right. Do you know what? We need to try this." And what sort of percentage of budget do you think somebody should try certain tests with? Is there a gauge that you use or a percentage that you use and where should somebody start?

Frederick Vallaeys:
Yeah. I mean I think experimentation is one of the key things that will set an agency apart from its competitors. There's so much new stuff being thrown at you. He or she who tests the fastest and finds out what's successful the fastest is going to win this game. So absolutely, experimentation is foundational. As far as the budget that you put towards it, it depends again, how much budget are you starting with? You really want to have enough budget dedicated to it so that you can find some meaningful insights in a matter of one to a couple of weeks. Never look less than a week because you get day of week patterns that may not be accounted for, but you also don't want to be waiting months.

Frederick Vallaeys:
Another thing people do wrong in experiments is sometimes you have experiment that's running at 50/50 with the control and it stays like that for months and months and months. And it's like, "Well, you probably, based on the time factor, could say, this is actually not winning." It's not losing, but it's also not winning. So it's just hogging space that you can use to do a new experiment. So have a framework in place for figuring that one out. And then sometimes with experiments, especially now as we go into performance max campaigns, how do you test a performance max campaign versus a shopping campaign? Well, you really can't because if you have the same products in both, the performance max, just like smart shopping, will always take precedence. But then you get into these really cool and smart techniques, and again, it's about splitting out your products in a sensible way.

Frederick Vallaeys:
This is an example I give in my book, but we've seen clients and they sell different types of bathtubs, like clawfoot bathtubs and more modern sleek bathtubs and they wanted to split them out because one of the types of tubs was actually in their warehouse, the other was being drop shipped. But the problem is, even if you split that out, if somebody's looking for a clawfoot bathtub and Google sees while you have a smart shopping campaign with modern bathtubs which you want to deprioritize, it's still a bathtub. So Google says, "It's still close enough that we're going to show it." So you have to find these real, clear delineations where there's no chance that this query is going to go between your smart shopping and your regular shopping campaign. Only then can you start to really do a good test on that.

Richard Hill:
Yeah. No, that's brilliant. So, crystal ball time. We've got you back on in, let's say we've got you back on in 18 months to start with, and then we'll talk about agencies in the future as well. But we're sat here in 18 months, Fred, what are we talking about on Google Ads? 18 months, what's going to be, you've talked about there, the new enhanced value coming through and those adjustments that can be made around returns and things and that sounds amazing, but what else is coming through? What do you think will be coming through?

Frederick Vallaeys:
Listen, I think in 18 months if I have my crystal ball, we're going to talk even less about details like keywords and bids themselves and we're going to talk much more about the inputs that we give into the system, like feeds, like setting targets, like connecting things with your business insights, like how does the weather impact stuff? So we're going to be much more talking as marketers about high level stuff, as opposed to managing these details. In 18 months, we're still going to have frustrations that these details are being taken away from us. These levers are being taken away. But I think we're going to transition to a place where we're okay with that because we're actually going to see better results and we're going to learn how to live together with these machines and actually make them do the work for us.

Richard Hill:
Yeah. Yeah. So if you didn't quite catch that, we've got to embrace, we've been embracing very heavily the last few years in our agencies, and boy, am I glad we did, automations and different high ends to different tools, Fred's tool being one of the things we use, we wouldn't be still in business I don't think if it wasn't for embracing these things. And I think that obviously goes with every, you guys that are listening in, in your business, things change and if you try and stick 100% with what worked permanently, obviously things do change, especially in the majority of eCommerce industries, whether that's product trends, whatever it may be, we know we've got to embrace that change. And I think when it comes to AdWords, more than ever.

Richard Hill:
So, okay. So agencies, obviously I know a lot of agencies do listen to the podcast as well, and we have agencies, I have two agencies. So what does agency world look like in a few years' time? Will a typical agency exist? And I think it's a similar answer to, we've got to embrace the AI, but agencies are very much changing, aren't they? What would you say about agency life in a few years' time?

Frederick Vallaeys:
Yeah. I mean I think agencies will exist, just your role within the agency will be dramatically different. And so again, my book, it's not an expensive book, it's available on Amazon, but it-

Richard Hill:
I have it right here, Fred.

Frederick Vallaeys:
Oh, thank you. So it's called Unlevel the Playing Field. The first one was called Digital Marketing in an AI World. But the first one talks about it much more, the second one kind of restates some of the facts, but your human role in an agency is going to be, you're going to be the PPC doctor, the PPC pilot, and the PPC teacher. The biggest shift between book one and book two is I explain some of the specific ways that you can be these three roles and how you can automation layer them, how you can build your smarts, your knowledge of how to work together with the machines, but make sure you don't have to do this manually, that you can actually scale it. And so I think that's the big shift and if you've asked me this question two years ago, I would've said like, "Hey, a lot of you need to become really developers, you need machine learning." And then I realized that's a big ask. Okay? That's hard.

Richard Hill:
Yeah. I'm definitely changing my career if that's the way it's going.

Frederick Vallaeys:
Exactly. But you don't have to be, and that's the beautiful thing that I've started to realize was it's these lightweight, small automations that you build through simple tools that keep the machines in check. It's about checks and balances. Again, Google does the really heavy lifting with machine learning that costs millions of dollars, but if you can literally have a rule engine or an automated rule that says, "Hey Richard, this campaign is on automated bidding, but did you see that it just spent $200 on a click?" And you're like, "WTF?"

Richard Hill:
Yeah.

Frederick Vallaeys:
What do I do? It's like, "Okay, I need to go and look at that. I need to understand what happened. That seems too expensive." Right?

Richard Hill:
Yeah.

Frederick Vallaeys:
So these are the types of things that I think we're still going to be doing for a long time. Creative stuff as well, Google is doing more of RSAs and then putting text together in its own ways, but at the end of the day, humans communicating to humans are still doing an amazing job.

Frederick Vallaeys:
There's also, I'm reading this book right now, it's called The Loop and it talks about this self-reinforcing mechanism, how all people are getting sucked into these machine learning systems and machine learning, by and large, looks at the history of what's happened to predict future things that will happen, but its predictions tend to be limited to the things that have already happened in the past. So yes, a machine will be able to write really good ad text for Google, but it's not going to get creative. It's creativity, it's actually making mistakes that forces innovation. And I've given this example too, but I think it was an engineer on Facebook, they made a mistake in the video encoding algorithm for ads and one of the ads started jittering a little bit and it was that jitter that actually got a huge click through rate. People were like, "What's that?"

Richard Hill:
Yeah. Yeah.

Frederick Vallaeys:
[crosstalk 00:54:29], we need to build animation and we need to have video-

Richard Hill:
Where's the jitter button? Where's that jitter button? I can't find the jitter button.

Frederick Vallaeys:
Exactly. But it was a mistake that was like, "Hey, that's maybe something we should be doing." And that's the role of humans, right? Push the envelope, figure out what's next, be different, stand out. The whole example about Nike and the shoe colours, sure, white shoes are best sellers, but if you have 10 white shoes on a page and one that's purple, hey, I'm seeing the purple one, I'm going to that one out.

Richard Hill:
Well Fred, it's been an absolute pleasure. Absolutely has. I think there's literally so many, so many takeaways in our 100th episode, so it's been an absolute pleasure having you on. I really appreciate it.

Frederick Vallaeys:
It's been really fun. Thanks Richard.

Richard Hill:
No, I like to end every episode, as everybody will know, with a book recommendation. Now, obviously, I will recommend this till the cows come home, so go and buy this. But if you are buying this, but you're also buying another book, what else would you recommend Fred?

Frederick Vallaeys:
Yeah. I mean listen, the recommendation, I'm not sure, but I'm currently, like I said, reading The Loop.

Richard Hill:
Loop. Yeah.

Frederick Vallaeys:
And it's a very fascinating story about human behaviour, the patterns that we get into. It has a lot of applications to marketing. As a person, I'm also starting to think a little bit more about the impact marketing has on the world. We want to be helpful to people, Google became successful because it was helpful to people in answering questions. Scarily, the world, there's overpopulation, there's too much consumption, are we getting people to buy things they don't need? So I'm not necessarily a huge fan of Facebook ads, Instagram ads, even Discovery ads. I see they have a place for introducing people to new things, but I think we also need to be cognizant about there's a difference between helping people and getting people to believe they need something. Right? And so those are the types of books I'm reading right now, just to make sure that I help people, I make people's lives more efficient, I give people a platform to be more successful, but we also don't ruin the world.

Richard Hill:
Yeah. That's brilliant. Well, thank you, Fred. For the guys that want to find out more about you, more about Optmyzr, what's the best way to do that?

Frederick Vallaeys:
Yeah. I'm on Twitter @siliconvallaeys, silicon and then V-A-L-L-A-E-Y-S, like my last name. Can connect with me on Optmyzr, frederick@optmyzr.com. And hopefully I get to meet a lot of you in person at one of these events once they start spinning up again. So-

Richard Hill:
Yeah. They're all springing up now, aren't they? Yeah. I know the UK dates. I will see you there for the Hero Con in July.

Frederick Vallaeys:
Awesome.

Richard Hill:
That'll soon fly by, so I will see you then. But yeah, thank you so much for being on the show and I look forward to catching up with you again.

Frederick Vallaeys:
Thank you.

Richard Hill:
Thanks, Fred.

Richard Hill:
Thank you for listening to the eCom@One eCommerce Podcast. If you enjoy today's show, please hit subscribe and don't forget to sign up to our eCommerce newsletter and leave us a review on iTunes. This podcast has been brought to you by our team here at eComOne, the eCommerce marketing agency.