Everything You Think About Google Ads Is Wrong with Collin Slattery

Speaker A: You're listening to the Ecommerce Impact podcast where we share the latest marketing strategies to grow your e commerce store and have more impact.

Speaker B: Mhm.

Speaker A: Hey everyone and welcome back to the Ecommerce Impact podcast. Today I've got uh, Colin Slattery to tell us about his robot. Really interesting personal journey to being one of the best Google Ads experts in the world. And I say that without hyperbole, it really is true. Colin's one of the people that I respect most in this space so I'm super excited to have him on the podcast to pick his brain. Welcome along Colin.

Speaker B: Thanks for having me, Jesse. I'm excited to be here.

Speaker A: Yeah. Why don't you tell us a little bit about your agency and where it is now, a little bit of your success story and then we'll dig a bit more into how you got here.

Speaker B: Yeah. So companies Tycoon Digital and started it almost 15 years ago now. So it's been, it's been a minute or will be as of the 28:15 full time and you know, I've seen a decent amount of growth over the last year and a half.

Speaker A: So do you want to tell us a little bit about the types of clients you work with?

Speaker B: Yeah, so I mean it really varies, I'd say like from upper seven figures in terms of like US dollars, so you know, 7,8 million all the way up to you know, the big nine figure brands best known for working with Jones Road who are one of our larger clients. So you know, definitely have some nine figure but I think that like at least the ones I find most exciting are that like high seven to like you know, 20, 30 million dollars sort of transition range which know, can be a lot of fun.

Speaker A: Yeah, yeah, that kind of truly scaling up phase is my favorite as well. You can move fast, break things a

Speaker B: little bit fit so you know that there's a customer base out there and.

Speaker A: Yeah, yeah. And you recently had Phil Keel join your ranks as kind of heading up the, the meta side of things, which makes you guys like a killer team over there. How's that been going?

Speaker B: It's been awesome. I was pretty good at meta, but not as nearly as good as I am at Google. And Phil's just incredible. It just, it's very nice to be able to you know, hire people who can run full departments. Right. And you know, I like, you know, before I even hired him like he was mentoring me on like leadership and team management and how to do those things. And so now, you know, it's actually really cool to Be able to see him like mentor junior employees on the paid social side and be like, oh wow, no wonder I was doing this so much worse. So it's, you know, I had high expectations for it and it's been even, even better than I expected.

Speaker A: Amazing. That's so cool.

Speaker B: Yeah.

Speaker A: I mean again like Phil being one of the people in the space that I respect the most about meta ads. So between the two of you, what a team. That's amazing. Okay, cool. Well tell me like what sparked my interest in like having you on the podc, as you have hinted a little bit in some of your social media posts about a kind of a bit of a rags to riches story about how you got here. So. Yeah, why don't you tell us a little bit about that because I think our listeners will be really inspired to know that you know, no matter where you start from, success can be achieved. And I, I think especially in these testing times we're in at the moment, it's always good to hear a story of someone kind of got getting through adversity.

Speaker B: Yeah, I mean like, you know, it was not super glamorous to start. I actually started the company with my last $127. Uh, you know like I worked at a bakery and walked eight miles to and from work to, to save the, I guess $275 each way on the subway so that I could start the company. Yeah, I was, I was not as, as well fed as I am these days.

Speaker A: Lucky you worked at a bakery.

Speaker B: Exactly. Take the. Take the day olds home with me. Yeah. Ah, I had a lot of, a lot of muffins for that time period. No, I, look, I didn't go to college. I uh, only graduated from high school and this, you know, dating myself. I'm um, I'm pretty old. This was the financial Crisis from like 08 to 2011 when just like everything was terrible. It's really hard to, you know, for people who are not alive during that period or like adults. Right. During that period to really kind of explain how bad it was economically. Right. And not easily employable with just a high school degree. So might as well start your own thing. And that's, you know, what I did and not just paid ads. Basically I'd take, take money for doing anything web related and make a lot of mistakes along the way and eventually I guess you just refuse to, to stop you. 15 years in should be somewhere hoped for. Boy.

Speaker A: Yeah, so I, I am old enough to remember the GFC as well. I was working in a marketing Job in Canary Wharf in London. And that's the financial kind of center of London. And so Leman Brothers was like across the, like the other building. And I remember like the lights were literally turned off in like whole sections of these buildings. It was really crazy. I managed to keep my job during that time but so many friends lost their jobs. Every newspaper was just like mass layoff. Mass layoff. It was, you know, uh, yeah, I remember how crazy it was. So you, so you were working and you were homeless at one point, is that right?

Speaker B: Yeah, briefly. Well, a couple, couple stuff, yeah, it was, it was, it was bad.

Speaker A: Yeah. Uh, and you started this business in like out of necessity basically and during the global financial crisis. And how did you like. So you're offering anything web related because you're uh, as we, before we turned on the recorder you said, you know, you're a self professed nerd. So I guess you've always been into computers and technology and that kind of thing. So you felt like you had something to offer that people needed. Is that kind of where it started?

Speaker B: Yeah, well, like in high school I started dabbling in this stuff too. I mean I took computer science in high school. I thought about going to college. Computer science. This was, you'll, you'll probably remember this too, like the, the wild west days of marketing. In high school I had a, like a Canadian online pharmacy. Right. Because like pharmacy advertising was not restricted at all. There were no, no restrictions. So you know, I'm a 17 year old kid with a discount pharmacy like doing like black hat SEO to get traffic to it. Yeah, I mean I had the, I mean you look back now and you're like what skills? But like at the, you know, had had the skills and there was a demand for it and I could be cheap because I needed to eat. I uh, yeah, hosting, web development, email marketing, like content creation, you name it. If you would pay me, I was going to do it for you.

Speaker A: And how long did it take you to be able to quit the bakery and do uh, what you were doing full time?

Speaker B: I mean I, so I quit that job pretty quickly, pretty much right around the time I started the company. And then I took another full time job which I had for a couple of years. And then like I quit that probably too early. But I was like, yeah, it's funny because I was like, oh, this is really cushy. You know, I'm making like 50 grand a year and like I don't want to get comfortable. This is more money than I ever made in my Entire life. And I quit when the, uh. So the first year of the company, when I was running it full time, we did $57,000 in revenue, so probably quit a little prematurely, but wanted to push myself against.

Speaker A: Yeah, I mean, who, who's to say if you hadn't done that, my. Maybe you just ended. Would have ended up falling into a corporate career, like. Yeah, it's funny sliding doors things sometimes, isn't it? And so your, your. What was that job, that corporate job? What was that, that. That you were doing? Was it anything related to marketing or websites?

Speaker B: Yeah, so I read. I ran the marketing department for a regional golf club retailer.

Speaker A: Oh, wow.

Speaker B: And they are still actually a client.

Speaker A: Wow.

Speaker B: Oh, they're our smallest client, but yeah.

Speaker A: Were they Ecom then and now?

Speaker B: So they were majority retail, but did have an E Comm component at a time. And so that was like my initial responsibility was managing the site, growing the site, doing advertising, email marketing, all that.

Speaker A: Yeah.

Speaker B: Yeah.

Speaker A: Wow, cool. So, so you. So you kind of went out on a limb, launch your agency properly went full time on it, and the rest is history. Been kind of plugging it away at it for 15 years. When did you sort of specialize in ads?

Speaker B: 2018 is when I decided to stop doing the other stuff.

Speaker A: Yeah. Okay. So relatively recently, I guess. Well, that's halfway. Halfway through. Yeah.

Speaker B: Seven, eight years ago. Yeah. Yeah.

Speaker A: Feels recent to me when it's like.

Speaker B: Me too.

Speaker A: I kind of think of the pandemic as like not that long ago, but it's super weird to me that it's like half a decade now. Cool. So. So tell me then about. Let's, let's check Google Ads because I've been dying to pick your brain on this and like, I, I'm not even sure where to start, but let's start with what's working recently and how is it kind of changing at the moment in terms of the way you're running ads and running ad accounts?

Speaker B: How much time do we have? I mean, it really depends on client size, I guess, and the, the media mix generally. You know, I mean, I think two things are always going to be true, right? Like there's going to be upward pressure on your costs, right? That's, that's true on meta, that's true everywhere else. Like death taxes and CPMs going up. Right. Like set your watch to it. And then, you know, one of the things that you can always say is true is Google's always going to be trying to put their hand in your pocket and pull more money out for Themselves. So you know, those are constants. But yeah, other stuff kind of changes around it.

Speaker A: Yeah, yeah. So CPMs are always going up and they're usually outpacing inflation. Right. So what's an advertiser to do about that? What do you think is the solution? How do you grow when your cost base is going up more than your prices?

Speaker B: Yeah, so I mean it is a multifaceted thing, right. I mean some of it is in platform. Right. So, so if you're able to squeeze more waste out than other people because you know you're better about negative keywords and running N gram analysis, um, that's going to be part of it if you know you're able to convert at a higher rate on your website. Right. So you know your traffic converts better. You know one thing that I think we do and this will actually, this is put maybe like a secret sauce alert on this one. But you know, like shopping ads, like you don't necessarily have to run those to the pdp. You can actually create landing pages or use landing pages and that'll increase the efficiency of your, your shopping ads. And so you can just like bid higher than other people so you can get more share. Right. Website related activities that are going to improve your conversion rate and your revenue per session so that you can, you know, you're just more efficient there. So yeah, getting rid of waste and improving how the post click experiences are two of the main things.

Speaker A: Interesting. Yeah, so it's not so much about like hacking the ad account but like ensuring that the traffic that you're getting has a really, really good chance of converting.

Speaker B: I mean you still have to, you have to make sure that like the stuff that you do in the account is correct and not just like, you know, it's always don't trust Google. Right. Like in, in a way that is not true with Meta in my experience like Google is a very adversarial partner and that's the whole like oh, use pmax, like use demand gen and drop all your assets in there and just like assume it's good and it's like

Speaker A: definitely don't assume that and, and trust Google Analytics.

Speaker B: Yeah, don't do that. I'm.

Speaker A: Because it's a secret. It's not using last click. It's using our uh, special algorithm that attributes your sales completely neutrally. Right.

Speaker B: Data driven. Yeah, but what data?

Speaker A: Yeah, exactly. I say that with. For those who are listeners who don't understand my sarcasm, I'm being very sarcastic about Google Analytics ability to neutrally tell us what's happening with our campaigns. Yeah, that's really interesting. Really interesting insight you said, about them being more of an adversarial partner than meta. So what has led you to that conclusion?

Speaker B: So, I mean, I think it's just basic market dynamics. So if you think about Google, right, they are a search monopoly. They have percent in the US somewhere around there of the search market, and they can't make more searches, right? The amount of people, the amount of times people search is the amount of times people search, that's a finite number that they really don't have any control over. And so the only thing that they have control over is how much they're monetizing each of those searches. Right? So the monetization, that's, that's you, that's the advertiser, right? You're paying for that. And so, you know, if, if Google needs more revenue and they're a publicly traded company, so they're held to their quarterly reports like they need to show more revenue. And the only way they're going to get that on search anyway is by increasing the number of auctions that you participate in, whether they're good or not, and then increasing how much you pay per, you know, auction. That's the only way they make more money. Like meta, on the other hand, like they have control to a degree, right? They surface more interesting content. You spend more time on the platform, you scroll more, you look at more content. That's more ad inventory that they can potentially sell. They can obviously adjust the amount of advent inventory that they have, right? So like, if it's one every 10, maybe they do like one every 9.3. Right. And they've just manufactured additional inventory. Right. Like, uh, and that doesn't necessarily impact you from an advertiser standpoint. Like they can sell more ads, whereas Google, like really cannot. They're, they're stuck with what they've got

Speaker A: and they can, I guess, like, be more competitive against TikTok or the other, like, competitors for our eyeballs and be better at that. Whereas Google, they're the only one, they've got no one to beat, but they can't make it any bigger, uh, than the monopoly they have. That's really interesting. Yeah. Okay, let's discuss pmax. So it's my hypothesis and what I kind of, this kind of sneaking suspicion I have is that PMAX is really claiming is basically a way for Google to spend more of my money, make it look good because of like, the brand value I've built up in my business and me Being the ecom brand, for instance, while making me feel like I'm getting a good return on my dollar and it's a black box, that means I can't actually tell which parts of it are going towards like my, the traffic that would have come anyway versus traffic that's like prospecting. Discuss. Am I right? Am I wrong? What's your view on it?

Speaker B: Yeah, I mean look, that's, that's, that's baseline. Exactly right. And that is exactly right when you're setting up PMAX the way that Google wants you to set it up. Right. And the way that uh, the patterns and the sort of workflow in works, I mean it's, it's only been in like the last two months where you can actually add negative keywords to PMAX without having to like send a request to a rep to put them on there for you.

Speaker A: Wild.

Speaker B: Um, yeah. And that's kind of step one, right? Is like get brand out of there. Because that's ultimately what they want, right? They want, you know, brand is going to be 10% of the spend but 80% of the revenue generated and then they can waste the other, you know, 90% of your budget. And still on a blended basis it looks great.

Speaker A: And like in your experience, does that do those brand negative keywords in pmax, like is that a foolproof way to stop Google from doing that or are they finding ways to leak through?

Speaker B: Oh, I think there's always going to be some leakage. And like this is something that I think happens when you get into the kind of like the other keywords challenge. Right. Where Google hides a lot of the search terms is that I do think there is leakage. Ideally it's not a lot. Right. But I'm sure there's going to be like misspellings. And like if you've ever looked at like a, you run like a brand search campaign and you look at the search query report and you're just like the kind of things that people type. You're just like, how did you even get there? Like what, what resulted in this query? Like, how, how did this happen? And I'm sure those end up in, in the pmax just because you're never gonna negative out like the world's worst misspelling.

Speaker A: Yeah, okay, so that's like crazy misspelling of your brand is gonna sneak through and then meta's also gonna kind of like cover up some of the others that are sneaking through potentially on uh, the other end of it. Like how much is PMAX just bringing you random crappy traffic that's only loosely related to your brand. Because I mean the example when I was running Google Ads, more hands on, you know, back in the day, one of the biggest things I was always doing was like being really tight with the negatives, uh, and using phrase match and exact match to make sure that I wasn't getting like these really weirdly loosely related terms coming into my, you know, and paying for that traffic. However, that doesn't seem to be the way it's done anymore. We are trusting the algorithm more using, you know, what effectively is broad match, which is performance max. But how do we, is there a way or do we just have to trust Google will find us the most relevant traffic and is there just naturally going to be a bunch of junk in there and we just have to accept that or what do you think?

Speaker B: Well, I mean I'll preface this by saying that we do very little pmax. I mean it's something that we're starting to retest again and ultimately, especially with brands at this size, we're testing for incrementality and you know, getting net new purchasers into the funnel and what is going to do that most effectively. And you know, we typically see that standard shopping is materially better than pmax at that job. So the vast majority of our clients run on standard shopping. But I do know Google's been making improvements to improvements to pmax, which is just giving us back stuff that we always had in the beginning before they took it away with pmax, we are going to be retesting PMAX for, you know, clients who run on standard shopping and see if it is, you know, good from an incremental volume standpoint over standard shopping. Yeah, but you know, from the shopping side of things in pmax like that, that's usually fine. That's just kind of like focused on your feed. We very, very rarely do like a full asset PMAX with like videos just because, you know, I think you're going to do a better job targeting on YouTube with, with YouTube if that's where you want to target. And I think that's certainly true with, with search as well. Yeah, I will be sad when they sunset DSAs and we have to do that in, in, in PMAX. But for now we still have DSA,

Speaker A: so I'm still getting dynamic search ads. Do you want to tell us just about that format, how that works?

Speaker B: Yeah. So uh, basically you're not targeting based on keyword. Right. You can target based on like landing pages and things like that. And you're like here's these landing pages and you're like, Google, you figure out what search terms this should serve against. And it serves as like a nice sort of like hammock under an account to catch stuff that you're not targeting. And also it's good for, um, keyword research because you're like, oh, well, it's in here, it's performing well, we'll pull it out, we'll manually target it.

Speaker A: Yeah, cool. And you mentioned incrementality. So I feel like this is like really the hot topic of 2025. And the good agencies and the people that are like really care about the growth of their clients are focusing on this. How do you think about testing for incrementality and how do you, how do you prove it with Google Ads? Like, how are you designing those tests to figure out that for your clients?

Speaker B: So we use holdout tests. Right. So you. Yeah, and there's. I'm going to very much simplify it, right, because there's data science that actually goes into it. But you show your ads in a certain number of regions. You don't show your ads in other regions, you know, geographic regions. And then you measure the delta in sales in the regions where the ads showed versus the region regions where the ads did not show. Right. And that's going to show, you know, how much of an actual impact your ads had on your sales. M and the reason why this is good is because the user journey is complex. Right? Like, uh, pixels aren't necessarily going to track everything that happens. You know, I'm going to type it in on my computer and then I'm going to go to my phone or, you know, I'm going to see the ad on Google, but I'll purchase on Amazon. And so there's just so much in the journey that gets lost. And so doing revenue against advertising in a region, like, really gives you a, I think a better measurement in terms of like what the performance actually looks like. Yeah.

Speaker A: Cool. Are there tools that you use? Is that a Google, excuse me, Is that a Google based tool that you use to do those holdout tests or are you doing it with your own like, technical setup?

Speaker B: So we work with a third party provider to run these tests. The providers and the tools get cheaper every year. You know, this is like, this is what like Coca Cola and Procter and Gamble used to use to measure like television ads in like the 70s. But they required, uh, you know, a team of data scientists with slide rules in order to do it. And now, you know, you can pay A few thousand dollars a month to a software provider and they'll be able to help you set up the tests.

Speaker A: Cool.

Speaker B: Yeah, I wouldn't trust any of Google's data.

Speaker A: Like I was about to say, we just can't trust Google at all, can we?

Speaker B: I mean, I wouldn't recommend it anyway.

Speaker A: Yeah, yeah. Cool. And I guess when you're running a test like that, as well as like using a tool to kind of control the data and the uh, statistical significance, you also have to pick a time of the year and a situation where you're gonna like, everything else is gonna be pretty stable, which is a bit hard at the moment. Right. With everything in the macro environment going crazy. Do you think that has a big impact or if it's pure geo, uh, holdout, you, you think you can kind of still get a good read on the data?

Speaker B: I mean, it should be fine. Right? Like, I probably wouldn't recommend doing it over like your biggest, like I probably wouldn't do it over Black Friday. Right. Or you know, if you're in like swimwear, like Memorial Day weekend sale. Probably not when you want to be running it. But you know, I think the, the nice thing is, is that in general, like, with the way that it's structured, like it shouldn't really matter too much when you're doing it. Again, I think with some, some exceptions. Uh, but in general I think it should be fine because again, like you're not, you're not comparing across like multiple time periods, you're just comparing across GEOs. And as long as the GEOs are selected in a way that makes sense. And again this gets into the data science part, like how are you selecting the holdouts versus how are you selecting the treatment? Um, it should be fine.

Speaker A: Cool. You mentioned a little bit a few minutes ago and I wanted to pick up on this because it sparked my curiosity. You said about negative keywords running, was it N grams? You said

Speaker B: an N gram analogy of that. This is something that we've done for a very long time and I think it's more important now than it used to be. Especially with, as you said. Right. Like broad match. Like Google puts a thumb on the scales for broad match. So most of our search campaigns are running on broad match, which can match to literally anything in existence. And Google hides a lot of search terms that they didn't used to hide. And again, like I say recently. Right. But I think they changed this over in 2021, so it's been a couple years. So N gram, essentially what happens is an Engram is like so one gram would be one word and like the instance of uh, every single like instance of that one singular word in all search queries. And then like a 2 gram. So like cold plunge. Right? Could be a 2 gram. And then it would be like best cold plunge. Like that has cold plunge in it. But uh, you know, cold plunge for sale. Right. Like that also has cold plunge in it. And the goal is to aggregate all of this data from the account onto each of these frams. Right. And so you can get ideas for how specific words and specific phrases perform at uh, an account campaign or ad group level. And it gives you the ability to gather more insight because, you know, when you have 500,000 different search queries, that's a lot of data that's spread across all of them. But if you can roll every instance of the word like free into one result, you can see, oh well, free actually performs terribly. Like, let's set that as a negative. And that's, that's the process. And so it does, it makes it so that it's easier to find large amounts of waste in an account that you can then negative out.

Speaker A: Right? Yeah. And I presume you're kind of, you've got like your own system for doing that, which means you can do it like at scale for your clients rather than it being like a really manual process.

Speaker B: You know, it's really not that hard. I mean you run a script on the account and then as long as you can do a little data manipulation in Excel, you can knock it out in 90 minutes pretty easily. You know, we, we can do it a little faster but you know, the script is free and um, yeah, it's really not too, not too hard to cool.

Speaker A: Well one, to Google if people aren't doing it already or chat to chat gbt about it, I guess. Or get Colin to run your ads. What was I going to say? You mentioned scripts there. So let's nerd out a little bit. Like how do you want to tell the audience a little bit about what scripts are and how they work with Google Ads? And like how kind of what are the main ones you like using and why?

Speaker B: Yeah, so I mean it's just software that you set to run in Google Ads and I forget exactly which submenu it's under now. I think it's like bulk actions and then scripts. And I'm, I'm definitely not the script guy, but like that's Nil's Rougemon's. Like if you want to, you want to really nerd on scripts. Go, go look up Nil's website. He'll set you up. And we use a lot of them, both ones that we found, you know, on the Internet on Nil's website. Some are paid like Mike Rhodes. PMAX script is, you know, something that we have in accounts where PMAX does exist. You know, it's great for things like alerts, right, if stuff changes easy. Like if you let's say are looking at L7, right, and like, or conversions like drop aggressively yesterday, you know, it's not going to impact your L7 numbers too drastically for a few days and like, you know, you could have a few days where just like conversions were non existent and your L7 numbers still look good but you know, scripts will be able to catch that instantaneously and like notify you. Right? That's a basic use case. We do have, you know, more advanced implementations internally where you know, like one of the things that I think most brands do the worst is brand bidding which we can get into the whole. Is that incremental or not, that's typically one of the first holdout tests that we run with clients. But um, you know, a lot of brands are running it on like a Target ROAS or a Target, you know, CPA or Max conversions or something. And usually that's just a recipe for a lot of waste. And so you know, we typically use a manual CPC structure and then we have our own like internally developed script that is able to adjust those bids multiple times a day based on you know, how aggressive like competitors are bidding in the auction and like what our you know, absolute top search impression share is. And so we're able to kind of just adjust those numbers to you know, maintain what we want while minimizing what we're going to spend on it, which is the ultimate goal.

Speaker A: So you're kind of creating your own automatic bidding based on your like your criteria rather than Google's. Again it's like, and I would say

Speaker B: I would only use this for brand search. Like I would not use this for non brand but I think you know, brand is an area where Google's gonna want to spend as much as it possibly can because it looks great but you manually can do a much better job than Air systems Canon. So this is just, you know, building our own system to do it even better than I could, you know, do it myself because I can't sit there

Speaker A: 24, 7 much as you'd like to. So I wanted to ask you then, and those holdout tests that you run on Brand. I'm super curious, what percentage of them roughly would you say the brand is? The spending on the brand keywords. So like, yeah, for the audience who don't, who aren't completely clear on that, if some, if your brand is called Frank Green and people are typing Frank Green into Google and coming to your site, that's brand traffic. So how often in those tests is that worth still continuing to do versus not 10%. Wow. So 90% of the time it's a waste of money.

Speaker B: Yes.

Speaker A: And in that 10%, is it generally because they for instance, have wholesale partners who are stocking their product and therefore they need to kind of like hold, like make sure they win the click versus people buying their product elsewhere or is there other reasons why you think it is incremental?

Speaker B: So I think that's, that's one of the potential cases. Although in those cases I'll also say it's like you can also just let a, let your retail partners have the win or just tell them, look, if you want to carry our products, you can't bid on the brain terms. Right. Like that's pretty common too for manufacturers. But that is one instance. Another instance is when your brand name is basically the product category. Um, and so it's actually very hard to parse exactly brand versus non brand. So like if your brand is like, it's called Dog id, right. And someone's going to Google Dog id, like that's a, that's a generic query.

Speaker A: Right.

Speaker B: But that's also your brand name. And so you know, it is hard to fully split like non brand intent versus brand intent. And so those are our instances where it can also be wow.

Speaker A: And then so you're always doing it as a holdout. You use it. You're not saying, I know this to be true. You're always doing a test. And when you've done those tests, 90% of the time the data is saying to you that you might as well keep that money in your pocket and not have it in the account, not have it paid for. Yeah, you've blown my mind.

Speaker B: I, I do.

Speaker A: You've confirmed what I thought was true, but I didn't have proof of.

Speaker B: Yeah. And, and again like it's, it's not 100% of the time, but it's, it's pretty, it's pretty close.

Speaker A: Yeah. And so it must be interesting as an agency, uh, so I could tell you a real example of a multi billion dollar company I used to work for who I wanted to do a holdout test like that. And they, the senior management didn't want to because they didn't want to have to explain to the board why the roas had dropped so much when we did this, even though it would be a net cost saving for the business and like a net better use of our, uh, ad budget if the test was successful. Have, has, has this lost you clients? Have you. How have you kind of like navigated this conversation with clients when they're Google ro theoretically or um, on paper drops because brand is no longer there to kind of like put the thumb on the scale, as you said,

Speaker B: running into that challenge now with a large legacy brand. And it's a challenge, you know, like at the end of the day, like, I treat client ad spend like it's my own money. And so I actually, you know, get pretty animated about it. I'm like, look, this is, we're wasting money. And like, I don't care if it like makes me look good. Like, I'd rather it be good, right. I'd rather be good than look good. Yeah. I mean it's, it's definitely a challenge. You just, I mean I always just like keep bringing data to it and just hope like eventually it's like, look like we, we can test on a small scale. All right, Prove it on a small scale. It's test on a bigger scale. Like, how can I help you educate other stakeholders? Right? Because ultimately I want revenue to go up faster than ad spend goes up. Right. And if that, this, this helps achieve that, even if it looks first. Right. So it's, it's, it's, it's a frustrating challenge because you're like, we can save you money and make you more money if you'll just let us do it.

Speaker A: Yeah. If you'll just step away from the vanity metrics.

Speaker B: M. Yeah.

Speaker A: Crazy. Um, switching tack a little bit. Like, how do you think about attribution? What kind of tools do you think are good or do you not use tools? What do you think's the best way for a brand to navigate? Like which channels are the biggest growth levers and where they should be apportioning their spending.

Speaker B: So multi, multi part question. The first is it's incrementality testing, right? And like getting a read on what a, uh, like a, ah, incrementality multiple might be. So like, you know, you run a holdout test on meta and in, in meta, right, it says it's running at a 2x, but the holdout really says it's running at a 3x. You're like, okay, well you Know, we can have a 1.5 multiplier on it. Right. And then you know, you can work against the end platform data. Especially you know, for brands that are in that like upper seven transitioning to eight figure where like the media mix gets a little more complex. I do think like an MTA tool is very helpful. You know, multi touch attribution. So Triple Whale or North Beam and then you know, using those numbers as like optimization levers. So you know, combination of MTA and holdout tests is really what I would consider to be the gold standard.

Speaker A: Yeah, interesting. And how do you like a uh, tool like Triple Whale for instance? There's like a multitude of different attribution models you can then apply to the data. How do you go about picking? Would you have a favorite or do you just, does it depend on the brand? Like again, it just feels like as marketers we can pick the one that makes us feel best. But how do we actually make that decision? Like objectively?

Speaker B: Yeah, I mean, marketing answer, right? It depends. You know, I would say like pick one and like stick with it. Right. And I don't think it necessarily matters. I mean that's not true. It definitely does matter. Right. But this is a challenge you run to. I think there could be data paralysis, right. Where you're like, I have three different data sources. They're all telling me three different things like what do I actually trust? And it's like well you pick one. Right? Like, you know, I, I like, I, I like, I like north beam data. Right. Like a North Beam one day click. I think is a great, great tool with a great uh, attribution model. Right. It's, it's quick, it's, it's, it's agile. I just think it's, it's a great place to optimize from. But some people think that's not what I want. You're like, all right, well you, you look at what you, the, the client wants to look at. But yeah, I mean, I think, you know, again, you take that, you take that MTA number and you get the sort of like multiplication effect on it from the holdout and then you'll rerun those holdout tests from time to time. But sort of like day to day optimization happens within the MTA tool.

Speaker A: Mhm. Okay, cool. Awesome. Um, that's really interesting. So before we wrap up, I've just got kind of like a final question which is like you mentioned you're not running performance max for a lot of your clients. You're running generally shopping. Like what does the structure look like and what does the typical structure look like? And, and does it differ for like size and maturity of client? Yeah, if you can share that, that would be amazing.

Speaker B: I mean the most basic structure is manual CPC brand search. Right. For the ones that have brand search running and if you're not going to run an incrementality test on brand, that's fine, but move it to manual so you can cut your costs on it. A branded shopping campaign because there is branded shopping inventory and you don't want that to go to PMAX or anything else. And then standard uh, shopping campaign, that's just your non brand shopping and then you know, non brand search and a uh, dynamic search ads campaign. And then you know as clients get bigger kind of the complexity of okay, how many shopping campaigns do we actually have? Right? Like how are we bucketing shopping in terms of like top performing SKUs versus other SKUs and, and so you know, scale dictates I would say like number of campaigns but structurally it's, it's just that it's the same structure with like added complexity and then you start layering in you know, demand gen for video action campaigns and other YouTube type stuff. But the uh, yeah, the core foundation is the same.

Speaker A: So Google. Sorry. So Google is similar to meta in the sense that at a smaller budget you want to be more consolidated and you don't want to split it out too much until you've got enough budget to make that split make sense. So if you're, even if you do have like a large number of SKUs and different price points and different kind of product categories, but the budget is relatively small or the revenue is relatively small, you would still kind of lump those together until the budget justified splitting to allow enough data to go through each of those split up shopping campaigns. Is that fair to say?

Speaker B: I mean I might pull out a lot of the SKUs in time. Just like start with a fewer SKUs. Especially like budgets, you know, budgets limited but you have a large SKU library. Like I probably wouldn't drop all of them in. And you, you do want to do some segmentation within the campaign for sure. Right? Like you don't want, you know, something that needs, you know, a uh, 5x row has to be in the same ad group as you know, something that needs a 2x. Right? They're different products, they have different targets. So you definitely want those segmented out. But yeah, I mean if you're really budget limited and you have a large SKU library, I'd probably just start with bestsellers and

Speaker A: cool and. Sorry, one final question around that. I see a lot of like advertisers, especially ones with like a larger SKU count, seem to struggle a little bit with or there seems to be like work to be done inside Merchant center, which is where like the product kind of catalog sits and making sure the products are correctly categorized, labeled, et cetera. Is that true that a lot of work is required there? Uh, is it manual work? Are there tools to help with it? How should brands be thinking about like maintenance of their catalog?

Speaker B: In Merchant center, it's 95% of the work.

Speaker A: Wow.

Speaker B: The entire benefit from a shopping standpoint or close to it is what happens in Merchant center in your data feed. And there are tools. I mean Feedonomics is I'd say like the most expensive option and like pretty much like a done for you option which is nice. We use data feed watch for our feed management. I typically would not recommend like a straight like Shopify to Merchant center connection. I uh, would want to go through some sort of feed management third party tool. And yeah, that's, that's where you're going to get all of your, your growth, your improvement is through the feed for the most part.

Speaker A: Yeah. And so uh, when you say like that's where all the work happens so that your agency team who are running ads for your clients, their kind of like day to day work revolves around like with those tools optimizing the catalog within, um, Merchant Center. Is that right?

Speaker B: Yeah, I mean usually the bulk of it's done at the beginning, right. Where you're like, okay, well there's 80 hours of work that we have to do to clean up this massive feed. Right. And then you do do ongoing testing or when new fields get introduced. Right. Where you can be like, oh, let's test a new, you know, title structure and you can apply that and do tests in there or. But you know, a lot of the initial cleanup is just, you know, it's, it's one and done for the most part and then just, you know, regular optimization and testing.

Speaker A: And so it's mostly around product categorization, product naming and then any metadata about the product being correct.

Speaker B: Yeah, I mean it's, it's, it's, it's the title. I mean titles are big, right? But like images, like how many extra images do you have? Are you actually in the right, you know, Google product category, which you know, I see all the time where they're like, oh, they're in like the parent category. So like shoes. But it's like, okay. Which if you go two layers down, like, you're selling ballet slippers, right. So you don't go to shoes, go down into valet slippers. Uh, it's all of those extra fields that are like, oh, these are optional.

Speaker A: Yeah.

Speaker B: You know. Yeah, it's like, optional, right? Like, yes, you technically don't have them. You will still be able to run, but.

Speaker A: But Google won't know what this product really is.

Speaker B: If you're only getting, you know, 40% of the volume you could be getting by just adding them. How optional is it really?

Speaker A: Yeah. Surprises me that, like, technology or AI hasn't solved this. Something that feels really manual.

Speaker B: I mean, it's not crazy manual in the sense that, like, you can use. And now we're getting into, like, crazy nerdies. Uh, like you could use regex, like regular expressions so that it can be done in a bold manner. Like, all of the data feed tools are starting to introduce, like, AI to do stuff, but have largely not been impressed with the results. And my experience with AI and, you know, we use it a decent amount, is that it's not actually. It's great with stuff that doesn't require structured data. It's very bad with, you know, spreadsheet work and like, data analysis is just. And, you know, restructuring data. It's, you know, if you ask it to write headlines, it'll do a pretty good job. If you ask it to restructure, you know, a massive, like, spreadsheet. Gonna do crazy.

Speaker A: Yeah, yeah, It'll probably get there in the next few years. Uh, hey, but we just have to be patient. Uh, cool. Love that. Thank you so much for sharing. So much incredible wisdom. I learned a ton. I'm sure our listeners did too. Is there. Sorry, what's the best way for people to follow you? I know you're like, you're active on LinkedIn. I'm not sure if you're active on Twitter as well. Anywhere else. Any freebies you want to give away to the audience? I know you put out some pretty amazing lead magnets, so.

Speaker B: Yeah, I mean, so LinkedIn active on there. You can find me Colin Slattery, Twitter, CJ Slattery. A lot of hot takes there. And, uh, yeah, I mean, like, if there's. We do have a ton of lead, like lead content, brand search guides, all kinds of stuff. So if people want those, you could probably just send me a DM and I'll just. I'll send you the PDF. So have happy to do that. And lots of paid social stuff there. That's mostly again, a, uh, Phil thing because lucky to have Phil on the team.

Speaker A: But yeah, go give Phil a follow as well. He's, he puts out a lot of great intel on, on meta ads, so

Speaker B: he's forgotten more about meta than I've ever known. So I love that.

Speaker A: Cool. It was great to have you on and I'll catch you soon.

Speaker B: Yeah, this has been a lot of fun. I appreciate inviting me on.

Speaker A: Thanks so much for joining us here on the E. Commerce. If I can ask you one favor, can you please make sure you subscribe and if you can leave us a review, it helps us have a much bigger impact with what we're trying to do here at the Ecommerce Impact Podcast. Now, if you're ready to take your Ecom store to the next level, then go to www.ecommerceimpactpodcast.com and click on the button to book a strategy call with me on my team. We offer a free order of your advertising and a custom growth plan so you really have nothing to lose by getting in touch and jumping on a call with us. See you soon and watch out for the next episode in two weeks time.