
Episode Description
PPC and digital marketing are constantly evolving and advertisers must adapt to stay ahead. It is becoming increasingly important to find new ways to leverage data effectively and optimize campaigns for better performance. Understanding factors like attribution, first-party data, and automation can make all the difference in driving success.
In this episode of PPC Townhall, our CEO and Co-Founder, Frederick Vallaeys, sits down with Scott Desgrosseilliers, CEO of Wicked Reports, to explore how advertisers can critically assess their data sources and attribution models to drive more accurate, informed decision-making.
You’ll learn more about:
- The importance of reliable attribution in PPC
- Challenges with GA4 and why it’s not always the best solution
- First-party data as a competitive advantage
- Choosing the right attribution model for your business
Episode Takeaways
The need for reliable, high-quality data is becoming more pronounced as we shift to highly automated PPC campaigns. But with challenges like GA4’s limitations, shifting attribution models, and privacy regulations, marketers are having a tough time making data-driven decisions.
Fred and Scott dive into the best alternatives to GA4, how you can optimize attribution models, and the evolving role of first-party data. Let’s get started!
1. The importance of reliable attribution in PPC
We’re in an era where advertising is leaning heavily toward automation. This makes accurate attribution all the more important because you need reliable data on what’s working for your campaigns to guide the algorithms effectively.
If you choose to blindly trust platform-reported conversions, you end up allocating budgets inefficiently. For instance, you may invest in campaigns that seem to be profitable on the surface but do not really drive any business growth. This is especially true in longer sales cycles where conversions don’t happen instantly.
“If you don’t have a good attribution system in place, you’re just guessing where to spend your money. And if you’re guessing, you’re probably wasting ad spend,” shares Scott.
2. Challenges with GA4 and why it’s not always the best option
Despite being Google’s go-to analytics tool, GA4 has some serious limitations, including inconsistencies in tracking and a lack of transparency in attribution. Its reports are likely to favor Google’s own ad products over other traffic sources. If you rely only on GA4, you may find that it over-credits Google Ads while not giving enough credit to other organic, social, or email efforts.
“GA4 is not built for your business. It’s built for Google’s ecosystem. You need to question whether it’s really showing you the full picture,” said Scott
A better way to avoid potential underreporting is to validate GA4 data against other attribution models, CRM insights, and offline data. This ensures you have a comprehensive set of insights and you aren’t making any decisions based on incomplete information.
Another problem is that GA4 struggles to accurately track user behavior across multiple sessions or devices. This means you end up losing visibility into the full-fledged customer journey, once again leading to data gaps in reporting.
3. First-party data as a competitive advantage
Scott also emphasizes the importance of first-party data and how it gives an edge to advertisers. Unlike third-party data, which many advertisers have access to, first-party data is unique to your business, exclusive, and a better reflection of actual customer behavior.
“It makes sense to capture first-party data—not only because you can market to them for free on owned channels, but also because it allows precise retargeting, like uploading unconverted leads. Getting someone to submit their email is a strong conversion signal, especially with all the clicks coming in,” says Scott
This is also why it is important to invest in your own data collection process. It will give you a lasting competitive advantage, help you optimize your campaigns more effectively, and future-proof your marketing efforts against privacy changes and tracking limitations.
4. Choosing the right attribution model for your business
Multi-touch attribution is becoming increasingly important for accurately understanding the customer journey. This is why it is important to choose the right attribution model depending on your business goals, sales cycle, and the complexity of your customer journey.
Some advertisers still use last-click attribution, but it overlooks the impact of upper funnel interactions.
Rather, it’s a better option to use multi-touch models, like linear or position-based attribution, to provide a more holistic view by distributing credit across multiple touchpoints. Ultimately, the best approach is one that aligns with how your customers interact with your brand while balancing data reliability and actionable insights.
Great PPC reporting provides context and analysis
Great PPC reporting goes beyond numbers—it provides context, highlights trends, and explains performance to help stakeholders make informed decisions.
With the right tool, your reports can build trust and clearly show the value of your work. Optmyzr helps thousands of advertisers manage over $4B in ad spend and makes it easy to create impactful reports.
Not an Optmyzr customer yet? Thousands of advertisers — from small agencies to big brands — around the world use Optmyzr to manage over $5 billion in ad spend every year.
Sign up for our 14-day free trial today to give Optmyzr a try. You will also get the resources you need to get started and more. Our team will also be on hand to answer questions and provide any support we can.
Episode Transcript
Frederick Vallaeys: His name is Scott Desgrosseilliers from Wicked Reports. So I’m really excited to talk to him because in this day and age where so much of PPC and digital marketing is becoming automated, the automations do need good data about what it is that’s working for you. And so how do you get that good data in place? Is it GA4 or not. Maybe GA4 has some big issues. We’ll talk about that. What are the alternatives? What are the attribution models you should be choosing? How does first party play into all of this? So, lots of good questions, and we’ve got a great expert joining us to answer all of these today. So with that, let’s get rolling with this episode of PPC Town Hall.
Scott, welcome to the show. Great to have you on!
Scott Desgrosseilliers: Thanks for having me, Fred. Anytime I can be referred to as an expert, that’s a good feeling. Very few things I’m an expert in life, but marketing attribution is one of them.
Frederick Vallaeys: Okay. Well, you know, if, if you’re good at one thing and you make a living on that, that’s not a bad thing. Right?
Scott Desgrosseilliers: Yeah, that’s what I’ve realized.
Frederick Valleys: So talk to us a little bit about how you became such an expert on that topic.
Scott Desgrosseilliers: Well, many, many years ago, I think I’ve calculated almost 30,000 hours I’ve put in on marketing attribution.
Frederick Valleys: Of course, the guy who’s talking to us about measurement knows exactly how many hours he put in.
Scott Desgrosseilliers: Yeah. So I calculated it once a while ago and then I kind of keep a log and keep track just ‘cause it was fascinating. I spent that long and I was still interested in it. You know, a long time ago, a good friend of mine had started a lobster company, Get Maine Lobster where he would sell food online.
He tried out Facebook. This was back in 2006. No, wait a minute. Was it 2006? Oh my God. No, I’m sorry. 2013. And he said, hey, Facebook, doesn’t work for lobster. And I said, why is that? He said, I spent $4,000. I only made one sale. And I thought, you know, perhaps, it’s just that they’re gonna take time to buy.
Let me try to find you some software and install it because I liked, you know, analytics and I was always doing data consulting and I couldn’t find anything that was gonna track his results. You know, this is back, you know, in 2013. And so I cobbled together this idea. Very complex way to track some clicks.
And it turned out that he would break even on Facebook when he would do the ad spend. But then if we looked at it with a 90 day lens, people would buy off his emails or return visits, and he was actually making ten to one on his money. So the light bulb went off for him that, wow, Facebook could make me a lot of money, and he now spends a couple hundred thousand a month on the channel, which he only spent four grand when he started. So the data really worked. And then word spread through just a few of his friends and a few people I knew at Infusionsoft, which is now called Keep, which is now called Thrive, and all of a sudden I had a consulting business that turned into, I needed some software and that’s how Wicked Reports was born.
Frederick Vallaeys: Nice. So basically don’t measure too closely because you might miss the bigger picture.
Scott Desgrosseilliers: A hundred percent. I mean it always blows my mind when people are trying to act on intraday. Its like they’re trading stocks, which I used to be a day trader, so I do know how to do that and enjoy it.
And the data is, the conversion data is on a lag. The ad platforms update their conversion reporting up to 72 hours after the fact.
So to say that, oh, the algorithms being good to me today, I’m gonna spend more is tricky because that algorithm might be good to you because you sent an email today, people bought and yes, paid did help drive it, but it was from a day, two days, a week, a month ago, and it’s just now showing up and it may or may not show up in the reporting.
So I find that time lag is real and not a lot of people understand it. And it’s a big advantage in a, it’s a competitive advantage if you do ‘cause you can spend where audiences look like they’re not converting and, you know, ignore some other ones that are inflated.
Frederick Valleys: Talk a bit more about time lag on the different platforms. When I was at Google, the time lag was significant before things would hit the reports. But then we would say, well, you know, it may be a couple of days before the data’s there, but in reality, in most cases, the data would hit the reports within 20 minutes.
But then I know nowadays Optmyzr works with Amazon as well and Amazon does have more glitches in their reporting, so it’s less consistent and they also throw it away after 90 days. So unless you have a tool that saves it, you know, it’s kind of hard to look back at what happened last year and what do we learn from that.
So talk a bit about the different platforms that you work with and how that change differs between them.
Scott Desgrosseilliers: Yeah, well the biggest lag we see is on Meta in terms of not how fast Meta necessarily reports their conversions, but the time between a click to a conversion. So we find that sometimes it can be, you know, and we’ll have lifetime attribution forwards and backwards.
Like, you may not, that’s not always, you’re gonna use that as analysis. You know, if something happened four years ago isn’t very relevant, but we can report if you wanna say, hey, where did all the days sales come from? Sometimes we’ll show in Meta, you know, that that lead was generated four years ago on Meta that bought today, particularly around the holidays with Black Friday, Cyber Monday.
So we can see conversion lag of up to a year in terms of the click to conversion time. It depends on the price point. Maybe you just ran a great sale that finally struck a nerve or they clicked a long time and then you retargeted them on a different platform that got them re-engaged.
So Meta would be the most time lag for click to conversion.
Frederick Valleys: I don’t know why, but this morning I was thinking about cars and how nowadays, it’s the scout, the new SUV that’s coming out, they’re basically taking a hundred dollars deposits and collecting somebody’s email address. But none of these cars are gonna be available for at least a year.
And the cyber truck was the same thing. I mean, how many years between when they started collecting the “conversions” and then actually delivering the product. But I find it fascinating that nowadays there seems to be this shift to actually collecting an upfront payment as part of that initial conversion event.
And then it, there’s a huge time lag to the actual conversion. And I suppose that ties a bit into. creating the buzz and the demand, hey, it’s, it’s exclusive to be on that list. And it just like gets you higher on the priority list if you do like the product, but it’s also about collecting first party data, right?
If you have these longer sales cycles. So any thoughts on that? Like is that a shift that you’re seeing as well in the industry? How are people thinking about these long conversion cycles?
Scott Desgrosseilliers: Yeah, I mean the first party data, not only collecting it, but then tracking that first party data conversion because the cost per click just keep going up.
I mean, they’re, and they’re probably not coming down ‘cause Google and Meta and all the other platforms have to show growth. So they’re not likely to suddenly cut the cost heavily on the cost per click. So yeah, I mean, even for an e-commerce brand, I mean, you’re paying a lot for that traffic and people aren’t not necessarily gonna buy on the very first time they’ve encountered your brand.
So it makes sense to try to capture the first party data. One, because you can market to them on owned media for free, but then also you can retarget very precisely, you know, upload the unconverted leads. So, I mean, we. I find also that getting, that is a really high conversion signal to get someone to submit their email, so you have all kinds of clicks flying in.
But the ones that if you can’t get the purchase, if you can get the email submission, sometimes you can scale that ad a lot cheaper because the other people that are bidding on that audience in that moment in the purchase cycle don’t see enough sales.
So the price can stay low and then you’re bidding on it ‘cause you’re collecting first party data that you’re then closing a couple weeks or a month later, or however long it takes for your particular conversion cycle. But then you’re able to keep accumulating it if you can track that, you know, stitch that together. The first party capture to the sale or, a first party capture from paid into sale on email or SMS in particular.
Frederick Vallaeys: And I think there’s an underlying, really core key point that advertisers need to keep in mind, which is sometimes you look at your CPCs and you’re like how is it that I’m paying so much for a click and I can’t get a positive return on ad spend? Like how is it that everybody else seems to be bidding so much higher and making money on this or are they just throwing money down the drain?
But I think what it comes down to is, like you’re saying, they’re just more holistically measuring what’s happening. And so they do have a better understanding that it’s not just that one time intraday event that leads to the conversion. And because they understand where that click fits into the broader picture, they’re able to bid more for it ‘cause they know it does lead eventually to more value than you might think at that instant in time.
And so the risk is that if you don’t put these measurement models and attribution models in place, you are sort of playing on a different field than smarter competitors, and you’re always gonna be behind them in terms of bids. And if you’re using Smart bidding, then smart bidding is also gonna bid lower.
You’re not gonna have the position, you’re not gonna get the volume and your competitors are gonna win and you’re gonna lose because you focus on measurement.
Scott Desgroseilliers: A hundred on all that. I mean that people aren’t, sometimes people are throwing away money ‘cause they have huge, huge budgets.
Particularly the bigger the company, the more likely they are just bidding on so much they can’t keep track of it. But anyone spending, you know, under a million a month usually has a grip on what’s going on or is trying to manage that. And if you see that you’re getting out bid on things, yeah, it’s likely they’re either taking some concept of LTV into account or some wider measurement window, or their offer is just working better.
Now, I mean particularly a subscription brand, let’s just say you’re selling a hundred dollars a month. A really brilliant marketer we had that was doing supplements and he would start bidding, raising his bids and budget on Meta when he was still showing a negative return on ad spend in the first 30 days.
But if the trend was good because he knew he was gonna get so much renewals. He was trying to get gobble up this audience before any ‘cause supplements is cut throat. The bidding, I mean, everyone’s selling powder and so he would go in and bid it up and try to clean up the whole audience before anyone else could get there, because once he’d found us before it was even fully profitable because he was gonna look at it with a nine, not just a, a wider click window on the attribution, but the LTV, he was gonna take 90 days of LTV into account, you know, account for his churn, but then assume that one sale was really two and a half or two and three quarter sales.
So instead of bidding and saying, I’m gonna try to purchase a customer for $50, ‘cause I need to immediately make, I’m selling it for a hundred and I wanna try to make 50 bucks, he’d say, I’m gonna bid 120 because I’m gonna actually make $270 on the average customer in three months.
And so that’s rough the first month you do it, but you do it a couple months in a row and then it’s like earning compound interest. You’ve got all those recurrings happening, plus you’re bidding enough, you’re having to flood a new subscribers coming in and they scaled to a nine figure brand, you know, with a couple strategies.
But one of them was really intelligent, high bidding for the initial purchase. That works if your business model supports it.
Frederick Vallaeys: So how do you know that your business model supports it? How do you have the confidence to make that bet? And do you think that’s different if you’re an agency doing it on behalf of a client versus in-house?
Scott Desgrosseilliers: Two things on that one to know is you look at your historical customer cohorts. So you take all your customer, if you’re doing it, it doesn’t matter if you’re just getting repeat buyers or you hope you’re getting repeat, or heck, you just had, you know, black Friday, cyber Monday and all the Christmas stuff.
You pull out your revenue data and you look at the very first purchase date of each, just say customer email or customer name, and then you aggregate all the other. Then each month after that month they buy, you append any additional value. So you look and see historically, how much lift have I gotten from my customers in the past?
Because if you haven’t done it in the past, it’s unlikely you’re gonna just magically do in the future. You should work on it if you haven’t. But that’s the way you use history. And then ideally you have history with attribution. So, you know, by channel or by audience which ones are more valuable because it’s likely the, the higher cost per click audience is because they’re more desirable.
You know, everyone’s making, there’s a pool of people making a half a million dollars a year, and Meta has figured it out. Well, they’re gonna charge $10 a click for that audience. ‘cause everybody wants that audience. So they might cost more because they have more money to spend or they’re in high know, more lucrative careers or in a more lucrative zip code.
They’re gonna cost more to get their retention. It’s just supply and demand. What was the second part of your question’cause that was an interesting one too.
Frederick Vallaeys: Yeah. So it was also agency versus in-house.
Scott Desgrosseilliers: Okay. So the big thing with the agency’s clients, you know, we work with 400 agencies now and it’s getting alignment of expectations with the customer.
So I have a whole course coming out on my measurement in data decision philosophy. But what I found was that you have to agree. Get aligned on expectations before you start just running out and throwing out bids and just trying to drive revenue with your proven playbook. You know, agency, oh, I got a playbook, I just closed this customer. They’re gonna follow my playbook, I’m gonna make ’em all this money.
You gotta say, hey, we’re gonna use this measurement tool, whatever. It could be anything, but we’re gonna agree on a time period and a way to measure, and then for a sale, are we gonna include any LTV or not? Because that can affect my bidding.
And then how much an average customer has been worth to you over, not a lifetime value, but a defined period of value. So short term value because everyone’s gonna say, oh, I have $8,000 customer LTV, you can bid 800 bucks on a customer, then you bid 800 and well, the customer is only worth a thousand dollars in month one and there’s something like, wait a minute, we’re losing money after we fulfill.
So you gotta get really clear with a client on how much value am I allowed to accrue because it’s gonna affect everything in terms of the go to market, the the paid go to market strategy.
Frederick Vallaeys: So that raises two questions for me.
The chicken and the egg problem. So you said look at your historical data to make decisions on how much you can invest in the ad platforms, but oftentimes the ad platform, I believe is the way that you get that new product into market. ‘cause nobody’s gonna buy your powder in such a competitive landscape unless you’re doing something right.
So what is that something? What is, what is the flywheel in this case? How do you get that data?
Scott Desgrosseilliers: So you’ll still have your sales data to show you the value of the cohort. So you’re just trying to say, regardless of source, when people buy, are they buying again? And if so, how long does it take and how much more money are they spending?
Frederick Vallaeys: But I’ve got some powder in my kitchen. I’m gonna put that up on, I’m gonna start selling it with all these health benefits after this call here. How do I get to anyone? How do I get these initial sales, right? Like, I’m an ads guy, so I’d be like, well, let me go put it on Meta on Google. Or is the assumption that you sold it through stores
Scott Desgrosseilliers: Google first likely. So you’d spend a ton of time on conversion rate optimization, getting that perfect page. Then, I mean the reason why all these things are out there is the powder is quite cheap to get. You know, you can buy it at bulksupplements.com. You can go buy ’em in bulk there, mix ’em up.
Frederick Vallaeys: New business idea. Okay, so that’s interesting. So basically you’re saying you still have to use the ad channels and your bids will be imperfect in the beginning, but that gives you the data. And now with two, three months of that data on what’s actually producing sales, you get a better understanding of what tools customers are worth, and now you can move towards correct bidding, where you’re optimizing your business.
Scott Desgrosseilliers: Yeah. I mean, there’s always that leap of faith and that intuition and the, you know, particularly when starting, there’s no like, foolproof way to get going.
Assuming you’re gonna get repeat buyers, which if the products, you know, not every market is, but a lot of ’em will, then you want to calculate back your break even and make sure you’re not bidding more than break even after fulfillment. But that would be your point where if you’re a cost to acquire the customer is higher than that, then you’ve gotta go readjust.
But you don’t want it, you need to give yourself enough timeframe, particularly if you’re a new brand going to market. But any brand, you can’t launch an ad and say, hey, my target is $50 to acquire the customer. And then if day one is 70 bucks, you’re like, oh, the ad didn’t work. Like you gotta give it a couple of weeks. You gotta have the budget for a couple weeks to run a fair test.
Frederick Vallaeys: Okay. And so I said there were two questions. So my second question that you made me think of was the conversion lag could be a high number, right? So it takes a while for the conversion to happen. You then said, well, as you start to see the repeat business append the value to your data so that you start to understand that a certain customer had a higher lifetime value.
My question is about bringing that back to the bid management systems. And so let’s take Google as an example. How long do you recommend you update your conversions? ‘cause the offline conversion import functionality does have a relatively long window. It’s not super, remember it’s 90 days, right?
Scott Desgrosseilliers: So we have an integration with that. We built the very first one with Google, so I’m very intimately familiar with it.
They have a lag on that, right? You don’t. So, and what we do is we’ll store the, you know, not everything has a GCLID, but if we have a GCLID or another identifier, we’ll hold onto that and then check your shopping cart every day, or your Stripe or you know, Shopify, whatever. And each day if there’s a sale from a customer, when we work our attribution back, if we say, oh look. they have Google, offline conversion tracking.
I’ll call it OCT for now. Google, OCT enabled. And we have an ID. We’ll go tell Google, hey, guess what? This click’s actually worth more money. We’ve actually, we do a lot there with Google, OCT, we allow you to only send first clicks attributed. We allow you to exclude campaigns.
If you don’t want Google bidding more on brand, you can exclude brand conversion. So we won’t send those because you want, you’re always trying to tell the, the black box what you want. Yeah. Which is easier said than done. But for the lag, I mean, one way we have this report already, but you could do it on your own, is you export out your CRM emails and create dates.
Then you export out your shopping cart. First purchase dates and emails and cross reference to get the date diff, and then that will tell you how long it takes people to buy after they’ve opted in. You’ve captured the first party that’ll give you an idea of the lag after you’ve earned the lead, which may or may not come on the first, second, third, fourth, click whatever it takes.
So that gives you an idea of a ballpark of how long if people didn’t convert right away. They took more than a day from joining my Klaviyo until they showed up in Shopify. How long did they take? And then you get like an average of that out, right? Or a mean you get a distribution and account like every week and then you gotta double it because if, let’s say you’re bidding on, let’s say it takes a week for people to convert.
Well, day one, you capture those people. That’s one day that’s gonna take till day eight for that many people to convert. And then day two. So you gotta like treat them as little mini cohorts. So it’s challenging.
Frederick Vallaeys: And that’s exactly one of the reasons. So in our rule engine or reporting, we have custom date ranges, and one of the things that we allow is how many days should we push back your reports so that you’re not.
Making that mistake that you just said, which is on day two, you assume that your conversion rate is complete when it’s gonna take another seven days past that, for that day to be reported, right? So it looks like yesterday was really expensive and didn’t really lead to a lot of conversions. But you can probably already predict sort of how many conversions will come off of that from historical insights.
But if you’re gonna be reporting it back to a client, say that you’re an agency, I mean, if you send reports on the first of the month, well guess what? If you have a seven day conversion lag, then those last seven days were incorrect within that. Right. And that’s gonna make it look like you actually did a really lousy job during that month. And so I completely agree with all the things you’re saying there. It’s so important to understand conversion lag and then put it into your attribution and your bidding and your reporting. It plays into everything.
Scott Desgrosseilliers: Yeah. The biggest lag I’ve seen was a company that does cosmetic procedures and they run ads to get people to come get free, like, I don’t know, makeover or something.
And then they want to get them in for the real, like, you know, cosmetic, whatever they’re gonna do. And we have a feature where you can just track the ROAS and the cost to customer, then the month of the click, and then you can see the next three months what those clicks value did and it would triple their return on ad spend.
So if they just looked at it even from Meta or whatever it looked like, you know, not bad, but when they added the next three months, when the people got the free service or whatever, and then went in and redid their face or whatever they were doing, they were, they’re making a fortune.
And that’s where measurement was a key factor. I mean, they still have a good product and they have a good funnel. Like you still need to be good at marketing, but this way data unlocks a massive channel that otherwise would look very so-so.
Frederick Vallaeys: Now, and this is a very specific question, but I get conflicting answers from even Google on this.
So with the 90 day period in which you can restate conversions, it sounds like you are restating it every time you see new activity happening, Some people like Google say it’s more important to just consistently do it on the same schedule. So, every, like two weeks after the initial conversion, update the value and then don’t do it again.
Because that way they’re smart bidding. Our algorithms get to understand, oh, it looks like there’s sort of this gap of two weeks after which something happens and it’s, they say it has an easier time to process and understand that than if the values keep changing on a daily basis. I’ve also heard them say, really focus on the first two days, that’s when you need to restate it. And if you don’t have the actual values, tell us a predictive value because that short date range, there’s almost like a time decay model inside of the algorithm updates, right?
So it’s gonna weigh the recent changes much more heavily than something that’s reported 90 days after the fact. So in your experience like. Have you seen a trend? Is there anything to any of this?
Scott Desgrosseilliers: So what we found is when people have OCT turned on, they, and this is over, you know, hundreds of customers, their average, they’ll do 9% better than those that don’t. So nine percent’s not insignificant. It’s not like the rich, but 9% better is a lot.
Yes, that’s generally how they’ll do. And that’s based on how we do it, which is we do. We only up update once a day. We don’t do real time. Once a day, we say, hey, OCT here, hey Google, here’s all the conversion data for all the clicks, for all the new sales conversions, for any clicks we’ve tracked in the past.
So that’s just how we do it. Once we detect it, we send it up. And it’s working some, I mean, 9%. But yeah, we, when we built it with them, that was the strategy we had. But it, you know, then someone new takes over the bidding algorithm and tweaks it. It’s frustrating. ‘cause I mean, I would, I would hang on those Zoom call, you know, I’m not one for hanging on a lot of webinars, but I’d be on theirs every month if they were telling us how they changed things and what we should do.
It’s tricky to get two different people in Google tell you the different stories, and then you’re like, well, geez, I don’t know what to build as long as the customers are getting value. And then I’ll like, you know, I know I’ve done what I can, so yeah, I, I don’t know what’s going on in that.
I just know when we have revenue, we tell them as soon, you know, when we can. And we hope that they’re making the right decision with the algorithm, but, and well then we allow influencing of it, of only sending certain types of conversions if you wish.
Frederick Vallaeys: Yeah, so let the data speak for itself, right?
The people at Google may say a variety of things, but I think oftentimes, it’s a generalization of what’s happening. And do remember every advertiser is different and what works for the average may not work for you. So by having a good strategy around data, and like you said, test, including, excluding different things and, and see what works and what gives you the biggest lift.
Alright, great stuff Scott. I did also have a few questions related to some blogs that you recently wrote. So, the first one that I saw was in the last quarter you wrote about email data as first party data. So email marketing benchmarks and basically I think you were saying use benchmarks to guide your strategy.
Can you talk a bit more about both, I think the first party data component and how benchmarks fit into all of this.
Scott Desgrosseilliers: Sure I have strong opinions on benchmarks. So the benchmarks I don’t really love are industry or channel wide. Like we still report them because people love to read them. But just because you, let’s say you’re selling food online.
If I say the average food cost per click is $2 and the average CAC is $50 and yours are different. Then you start thinking, oh, I need to do something. And I don’t think that’s relevant. I think the benchmarks you need to use are your own. I like to think of the analogy of weight loss or getting in shape.
Like if, if my, 90-year-old grandmother can walk around the block, that’s a great achievement for her. That’s a disaster if that’s all I can do. Everything’s relative. And so I like to train our clients to use benchmarks to then to first of all, get a realistic thing of what they’ve done so then they can set a goal that’s actually possibly achievable.
And then, measure them against themselves. They’re trying to beat themselves based on how they’ve done in the past, just like you do with a diet or in the gym or, you know, running or whatever. And so, specifically with email ‘cause where it’s, you know, your own media, it’s a free conversion if you can get it.
You have this wealth of data. And so like if you send an email out, you can learn a lot based on your benchmarks. If there’s a lot more clicks than your normal email to that segment, hat means that email subject line, that email message, if the clicks that email offered was good. And if there’s less than normal, then that means something was off.
And then the same thing with the sales off an email. If they click the email and buy, which happens a a lot. Email’s a great channel for people. If you have a low amount of clicks and a high amount of sales, then that tells me as a data person, I have a great offer. I have to optimize this email to get more people to use it, but I should store it and reuse this email ‘cause I found something that sells.
Or if it’s I can’t, ‘cause it was a one-time sale we’ll analyze the structure and the image and what was my hook like, learn from the past what type of emails. It can be less creative, but it’s more valuable. I got this idea because back when I was, well, this is in 2013, Infusionsoft now, it was a big CRM would fly me in to be the data guy for their accelerators. Companies would come in, pay 15 grand, get a whole marketing thing done for them, but they’re already successful and they’d have these beautiful, sometimes one guy had this beautiful set of emails written, he’d paid 10 grand for them.
And I tracked him for him and I was like, and he, I was like, I got bad news. I was like, you’re only selling like a two or three emails with these pretty emails and your text emails are selling 10 plus sales. And he was like so upset. He’s like, but I paid $10,000 for this. I go, dude, beautiful. No one cares though.
I’m sorry you didn’t pay me them. So it’s not my fault. It’s not that. I mean they look great and that shows you gotta test the data. But then I knew that they weren’t good ‘cause I had a benchmark for ’em of how his average email would do, given how many people it was sent to and how many clicked on it. I was like, you gotta stop sending these right away because it’s just like that’s where your revenue dips cause you know it’s at an accelerator trying to improve his revenue. And I was like, you gotta go back to your old emails. He didn’t like that. I don’t know. You know, he didn’t necessarily like that news, but it was, that was the story.
Frederick Vallaeys: Yeah. And I think that also speaks to, you know, teams and people who want to be busy.
So they keep doing efforts in the same areas, but ultimately you do hit a point at which sometimes these experiments or innovations are actually gonna hurt your performance. And so then the question becomes, well, listen, if I’ve done AB testing on my ads, and I’ve done five rounds of optimizations and every time I do a test, it reduces the results.
You’re actually doing yourself a disservice, right? Because that 50% of traffic you put into the experiment has gotten fewer sales, and if time after time it keeps losing, like what are we learning here? Are we learning that you’re bad at it? Experimentation or are you learning that you’ve kind of reached the limit of what’s possible in your space?
But that becomes the more important question to answer at that point, rather than, hey, what’s the sixth test that we’re gonna do? And again, kill off a bunch of conversions that we would’ve gotten if we hadn’t done anything right. But going to your boss and saying, I didn’t do anything and we had good results again. That’s not gonna fly right?
Scott Desgrosseilliers: Yeah, because there is that case, I call it finding the Meta sweet spot. So these ad sets only have so many people in them. They generally aren’t infinitely expanding. In some cases they are, but if, like, let’s just say you do a basic lookalike, or if you got a keyword search, you only have so much impression share.
Maybe there’s only so many keywords that you’ve dialed in that are gonna work, and there’s only so much impressions you can bid for. Still, it’s not profitable. So people just wanna scale, scale, scale. But sometimes you gotta keep, it’s more like plateaus. Yeah. You get to the next plateau. Meta will be like, well, I can spend 400 a day, but when I try to scale to a thousand a day on this ad set, I don’t make any, it goes down.
What’s wrong with the algorithm? It’s like, no, there’s just no more people in there that are willing to buy. Dial it down to the, the dollar amount per day that makes your profit, and then you move on to the next one. You just gotta get a bunch of those going.
Frederick Vallaeys: So I have a thought on incrementality here, right?
So you could go from a $400 daily budget to a thousand dollars daily budget, and the ad platform may tell you that you’re still gonna get conversions below the cost per acquisition target you have, or within the return on ad spend targets that you have. But think about the incrementality in this case because that average comes from taking everything together and maybe that additional, that incremental one sale that you got comes at a higher price point than you could afford. So going back to all the examples about nutritional supplements, if you know you’re gonna make $800 on that customer a lifetime, that next click could come at, it’s unlikely, but in say, $900. So that was a wasted click, right?
Even if that still gives you more volume, that your average cost per acquisition is still good, that new click was just too expensive and, and that then leads into a question, which I know you do have thoughts on, which is profitability, right? How do we get to profitable campaigns as opposed to just using the ad platform metrics like TACoS or ACoS, ROAS, which are not really business metrics, right? These are advertising metrics, how do we tie that back to something a business cares about, which is something like profit?
Scott Desgrosseilliers: Yeah. So it’s good to have a North star metric that say, before we get into all the nitty gritty of the optimizations, which I mean, I do love to do that. What’s gonna be our North Star business metric? Blended ROAS is a good one.
Blended ROAS looks at all your ad spend and all your revenue, no matter where it’s coming from, and saying, what, where do I need to hit, where I’m gonna be? And you gotta align with the business stakeholder. Where is it gonna be that I’m like ridiculously happy and want to increase ad spend?
Where is it gonna be? Like your job’s at risk and where is it like business as usual, it’s good enough, but please try to improve it. And you try to get those base guidelines in place before. Or if it’s not blended ROAS, maybe that that’d be one example that I like. Then if you’re selling on Amazon, which many brands are, and it’s a black box over there, but you’re running a lot of Google and Facebook, and you’re convinced that they’re bouncing to Amazon ‘cause of Prime, then if you do blended ROAS over all the channels, you can still show that it’s working.
I mean, it’s a high level working because they’re Facebook. Then you can say, well, my Facebook, I’m gonna use Meta. I say Facebook’s still a lot. I’m old school. If you’re gonna use Meta to like drive new traffic, and then you’re gonna use your search to drive the immediate demand, but then they’re gonna bounce to Amazon.
You’ve gotta have a strategy in place that measures that loss of signal. And so MER covers that. Well, MER and blended ROAS covers that. that’s one I like. And then another one could be like, well, I only want to use paid for new customer acquisition. In which case, first of all, you gotta be able to segment between new versus repeat, which not a lot of tools can do.
And then you gotta use your benchmark to say, well, if you can get it, what has new customers been to acquire? Because if you run a normal like, oh, I had 20 sales today and it cost me two grand, I’m paying a hundred bucks a sale, a hundred bucks a customer. No, ‘cause a bunch of repeats are in there.
And so really the cost to acquire the new is a lot more than you think. ‘cause if you see your cost to acquire a customer without knowing new versus repeat. The odds are fair that at least half of them are repeat if you weren’t segmenting well enough, or in Meta’s case, the AI is ignoring your ad sets and and retargeting.
Anyway, I don’t know if you’ve seen that. It’s been like an epidemic.
Frederick Vallaeys: I haven’t seen that. But if you say it’s happening, then I trust you. So we’ll take a look at that. So Wicked reports, I assume has that distinction between new and returning. That’s one of the great features there.
So you also kinda spoke about like the loss of data signal. And I think what you’re talking about there is the fact that say you advertise on Meta and then eventually the person ends up on Amazon and the conversion happens on the Amazon platform. You don’t have your conversion tracking pixel on there right.
And so that usually ends up breaking data. And so we live in a world where we have the large ecosystems like Amazon, Microsoft, meta, Google, they each have their own walled garden as far as protecting that data. And they have to do it. Partly because of privacy, but also really because that keeps you locked into their platform and they’re gonna be able to show better measurements as far as like what they contributed.
So what are some of the solutions? And I know there’s clean rooms coming out from different platforms. Are these giving us ability to look holistically, or is it a situation where you really do have to have expertise in looking in each platform and doing some handwaving and maybe some estimations to figure out what really is happening?
Scott Desgrosseilliers: Yeah, it’s number two. So like for example, with the Amazon loss, you’d have to use MER or blended ROAS. And then you still benchmark how you are doing in Meta in Google ‘cause even though there’s some loss of sale, you still benchmark that your direction is still directionally accurate.
Let’s just say you’re losing 10 sales a day from Meta that’s bouncing to Amazon. If MER is still good, then you just benchmark your Facebook where you’re at and then use that as your baseline and then make, as budget decisions that way. And as long as the numbers, even though you have some signal loss, are moving in the right direction, that’s what matters.
What matters is that you’re, you know, improving your business by using the data. Not that every last conversion is accounted for. It’s really an optimization tool is what it should be.
Frederick Vallaeys: There another blog post today I wanted to talk about. So exploring attribution models for improving return on investment.
Attribution models obviously have shifted quite a bit over the last couple of years. Hopefully nobody listening today is still doing last click. I’m suspecting you’re also not a fan of that one. But talk a little bit about what are the modern attribution models and, and what’s working these days.
Scott Desgrosseilliers: I like last click for two things—email, SMS conversions. ‘cause they should be the last click. Or if you want to just check if your tracking’s working from the previous day.
But otherwise, yeah, you can’t. You know, I know of some brands that actually do pretty well based on last click and it’s mind boggling. That’s just what they like to do.
Frederick Vallaeys: I do think if you are a big brand, things are generally easier because you have the brand, right? And I think it’s oftentimes big brands come in and say, oh my God, marketing is so easy. And like we just, we never got into a different attribution model because we didn’t have to.
And sure. But you have a luxury that most of the people listening today don’t have. Not everyone has that kick-ass brand that everybody already knows.
Scott Desgrosseilliers: Yes. And you’ve got the great creative team that has viral creative. So then your brand is top of mind, that’s really hard to do.
So one thing I noticed when I first started building some models out was that if I did a pure multi-touch model only and just shared the credit across every point, it diluted the impact of the more powerful conversion points along that path with the more, more important clicks.
And so my task was how do I give you data that’s attributed accurately that I can transparently show who it was and what they did. But when you make changes based on the data reported, you’re gonna grow your business versus everything looking somewhat bland. If you have like 10 touch points, you saw something for a hundred bucks, everything’s worth 10 that can be not look very profitable in some cases.
And then you’d be like, I know I’m making money, but now everything you’re telling me looks bad from a pure multi-touch shared credit. So, this company, well, they’re now called Tier 11, this media buyer over there, Ralph Burns. He was scaling this fitness lady at home called the Betty Rocker, and the strategy was, we’re gonna target cold traffic to go to her block then the people that view our blog, we’re gonna retarget them to opt in for first party data, and it was just called get their email. Back then they didn’t have the fancy first party data. We’re gonna get their email to join the challenge, and then on the challenge we’re gonna try to get them to buy with email.
So it was like a perfect thing. So they needed three different attribution models. You needed first click to say what drove first clicks and then hold onto them and then if they buy down the road after these 30 day challenge, track back to these cold clicks. And then what is working best to get people to join the challenge ‘cause that’s a big conversion right there.
Okay. I’m going to listen to this lady on my phone 30 days and do jumping jacks and pushups and whatever. I mean, she has a great program. I’m making light of it. And then okay, she didn’t have SMS back then, what email is gonna work? And that was how I originally started thinking about, I need to score.
I need to basically score, it’s keeping score. How am I gonna keep score so that this guy who knows how to buy media is gonna be able to scale this brand? And he scaled her like 20X. She went from like a lifestyle business to like a staff and office and the whole nine yards. And, it was good advertising, but the measurement really guided them.
And so I like to look at it almost like a sports analogy. If you win the game, well overall, that’s the point of business to win. But then if you win, American football, 41 to 38, well, your offense had great measurement, but your defense is terrible, so you gotta go fix the defense. The same thing with attribution.
Like is my top of the funnel working? Is my middle funnel first party data capture working, and then is my bottom funnel working? You apply the attribution models based on just like a coach on a team. Where am I trying to improve? Is my offense weak? Is my defense or is my kicker? That’s how I try to apply it.
Because the math is adds up, but if I just try to be a mathematical purist and say everything’s gonna get 0.12 credit or whatever, you’re not gonna get enough insight. So we have different models based on what your intention is, what’s your intention behind this campaign. Are you trying to get leads? Trying to just find new eyeballs. You’re trying to, I mean, everyone’s trying to sell, but is it just please sell some stuff Google. We have an attribution model for that.
Then we use Multitouch. ‘cause you don’t know. So we match the intention and strategy to a specific model. Then that gives you the right measurement to determine how to optimize.
Frederick Vallaeys: That makes sense. Yeah, and so like have these multiple attribution models and different ways of measuring, and in Google ads you can set that up and, and then the decision becomes, like in your sports analogy, Hey, we’re playing a team next week that actually has a pretty weak offense, so maybe we’re gonna put in place the attribution model based on, you know, we don’t need to be that strong on scoring points. So maybe you rest some of the players that will be more critical in the week after. Right? But you have all of these data sources available to deploy them at the right time. When you think that makes strategic sense. I don’t know if that analogy makes sense.
Scott Desgrosseilliers: Yeah. Because certain keywords like brand would look good at last touch but brand wouldn’t be good for first touch. That’s for sure. Unless you’re like household name and then you don’t really need to worry about a lot if you’re household name.
Frederick Vallaeys: Exactly. Hey, there’s another blog post I wanted to talk about.
So, GA4. After Universal Analytics, which was so easy to use and just had a completely different methodology of looking at things. I think people still struggle with GA4, and you’ve pointed out in your blog some of the challenges around that. But talk to us a little bit about what are those biggest challenges and what are the alternatives and maybe where does Wicked Reports fit into that?
Scott Desgrosseilliers: Sure. Back when we had Universal, we used to pull in some of the data even way back when we first started, but now I wouldn’t trust GA4. Not that they’re intentionally trying to mistrust you, just the way it’s set up. It’s just I don’t, I don’t get it. I don’t get the decisions.
Well, the biggest thing that we usually find is that people, if you go into the acquisition report, these brands that otherwise no one would’ve heard of, have a ton of direct traffic conversions. So that’s just useless. And it can’t be true. It can’t be true that if you have some mywebsite.com that no one would know.
It’s not like you’re viral and some, and you’re getting a hundred sales a day. 80 of ’em are direct. It’s just not, it’s no premise, no hypothesis. But that’s accurate. So that one always is great for us because we can say, well, we can pull up our clients and say, okay, they’re showing that same direct, but we act ‘cause we track direct if it’s actually direct. And we’ll say our direct is like 5% is those the people and those directs, you click on ’em and you can see, no, they really came from advertising. So as a marketer, if you see a lot of direct, that should give you anxiety because if your brand, your owner or your boss is in there seeing direct in the acquisition they’re gonna be like, well, why do we need you?
The other thing with Google it’s off, but with Facebook, it heavily discounts the most important piece of Facebook, which is top of the funnel. Top of the funnel is how you fill your funnel in the cold traffic prospecting.
The way that they measure for the acquisition report, it’s always massively underweighted to the two. It’s sometimes like 10 to 20 x less revenue reported than what’s actually there. That’s consistent. That’s not just like one high, one, like massive one that happened. The more revenue you do the, and the more spend on Facebook at top of the funnel on cold traffic, the more the disparity in the amount of revenue that the acquisition GA4 is going to miss.
And so that’s trouble because you need the top of funnel. The top of funnel already has the time lag. So the bosses and the purse holders and the budget people are already like, you know, why are we spending on this? And you know, as a marketer, you need new eyeballs in and need cold traffic. It just takes time.
So if they’re in there poking around at GA4 and seeing this low in their, in their minds low return or negative return, that’s just a firefight that you don’t want to have to battle. So it goes back to expectations. As an agency, you gotta like set, we’re gonna put X percent of the budget into cold traffic you cannot look at GA four for these results. You need to give us this much a timeframe and then judge us. You need to get those like captured on like a recorded Zoom in like a decision log where everyone’s who’s at the meeting, we all agree this is how we’re gonna do it because then the people get amnesia.
Sorry, I’m getting off GA4 here. It’s just that’s the pain point that people have ‘cause of GA4 is they got direct problems and they got way under attributed Facebook, and then new versus repeats off because they use new verse returning user, which I think is two months.
Sometimes it’s less. Sometimes you can go into a setting and make it more. You’re gonna think you’re driving new people to the site that aren’t new. They might have just gone away. Like, I’m not, I’m not rifling through the, you know, 500 different places. I might buy stuff all at Christmas. But then it comes Christmas and I could show up as a new user because I hadn’t been there even if I buy the stuff every year, or I buy it bi-annually, and there’s like, you know, thousands of the same type of people return buying that are showing as a new user and then you think, oh, I’m driving all this new stuff and you’re not.
Frederick Vallaeys: I mean, it comes back to understanding your users, right?
Because at the same time, that user who comes every Black Friday, like, is, is that really the type of customer you want that’s buying all of your deep discounts? And there’s nothing wrong with that necessarily, but just understand it and split it out again, like in that sports analogy, right?
That’s a different target market that you’re gonna have a different strategy for based on how you measure profitability. For that specific segment or cohort. And again, it all comes back to having good data, different ways to look at data, a nimble platform to bring that data into the decision engines, which are usually AI based and with the ad platforms.
So, I think Wicked Reports probably fits into that pretty well. And then to plug my own tool, with Optmyzr you can use all these different attribution windows and models and plug ’em into your rule engine strategies and automations and bring your profitability into the equation so that the target return on ad spend, you said is actually driving profits as opposed to just more ROAS but not necessarily profit.
Scott Desgrosseilliers: Yeah, you have a lot of slick automations in there and because a lot of time you get on and analyze an account that is not automated and they miss opportunities a lot. A lot of keywords that are like, they’re either getting outbid or they’ve got so many campaigns with keywords overlapping, you know, and then they’re only getting like $10 of ad spend on a keyword that brought in two customers.
You know, I’ll see that consistently. People be missing opportunities. And why is it. They’ll be like, well, you got the same keyword floating around in five different campaigns where your performance max is stealing it, or you know, all sorts of things that automation can help them catch and fix without even knowing.
Frederick Vallaeys: Yeah. Well, this has been great stuff. I think I’ve learned a lot. Hopefully, the listeners have enjoyed this as well. If people wanna learn more about Wicked Reports or stay in touch with you, what should they do?
Scott Desgrosseilliers: So go to wickedreports.com and, you know, you can click over on demos from the homepage, menu header, and we got an interactive demo you can look at with that before you even talk to anyone.
And then I’m on LinkedIn, Scott Desgrosseilliers, you can find me there. If you can spell it, or maybe Scott Wicked Reports probably would be easier.
Frederick Vallaeys: You know what? I’ll be honest, that’s exactly how I went to your LinkedIn profile.
Scott Desgrosseilliers: I just thought of that right now. I was like, boy, no one’s gonna type all that in.
Frederick Vallaeys: Yeah, well, Scott’s name is, right there at the bottom of the screen.
We’ll also put it in the show notes and we’ll put in the links to our Wicked reports as well as to his LinkedIn profile. Scott, thank you so much for sharing all your wisdom. I hope people stay in touch with you and take a look at your tool. And then if you’ve enjoyed this episode, please hit the subscribe button at the bottom.
Go ahead and give this a like, share it with your friends and we hope that you’ll watch the next episode as well. So with that, thank you for watching and I’ll see you for the next episode. Take care.