
Episode Description
Join PPC experts as they delve into advanced strategies for e-commerce advertising, emphasizing the importance of focusing beyond ROAS to include profit optimization and customer lifetime value. Discover how to align your PPC campaigns with real business outcomes, ensuring your ad spend delivers maximum profitability.
Topics discussed:
- Why ROAS may be the wrong metric to optimize for
- How Customer Lifetime Value can be used to optimize PPC
- How to optimize an ecommerce account for maximum profits
- How to activate first party data in performance advertising
Episode Takeaways
Why ROAS may be the wrong metric to optimize for
- ROAS (Return on Ad Spend) often doesn’t align with a company’s broader financial goals, focusing instead on revenue rather than actual profit.
- Businesses frequently find that while ROAS targets are met, overall profitability may decrease because these targets can encourage the promotion of lower-margin products or repeated sales to existing customers rather than acquiring new ones.
How Customer Lifetime Value can be used to optimize PPC
- Shifting focus from ROAS to Customer Lifetime Value (CLV) can drive more sustainable business growth by emphasizing long-term customer profitability rather than short-term gains.
- Integrating CLV into PPC strategies allows businesses to differentiate between new and returning customers, assigning appropriate value and optimizing bids accordingly.
How to optimize an e-commerce account for maximum profits
- Incorporating profit margins, return rates, and other cost factors directly into PPC campaign management can lead to more accurate bidding and better alignment with financial outcomes.
- Tools like Profit Metrics allow e-commerce platforms to feed real-time profit-based data into PPC campaigns, enhancing decision-making processes and campaign effectiveness.
How to activate first-party data in performance advertising
- The shift towards using first-party data is essential as it allows for more controlled and insightful input into automated advertising systems, leading to more accurate targeting and bidding strategies.
- Proper data integration involves not just capturing transactional data but also understanding customer behaviors and preferences to refine advertising approaches continually.
Episode Transcript
Frederick Vallaeys: Hello and welcome to another episode of PPC Town Hall. My name is Fred Vallaeys and I’m your host today. So we’ve got the great fortune of having two brand new panelists, both of them experts in the topic we’re going to talk about today. So the topic of the day is going beyond ROAS when you have an e commerce store and you’re advertising it on PPC.
We’re doing digital marketing. So depending on what type of business you run, you may have different goals, cost per acquisition, return on ad spend, but what ends up being true is that those goals are really things that Google has put out there. These are Google metrics and they often don’t really correspond to the goals that you may have as a business.
So what I always find funny and fascinating is that Google keeps talking about cost per click and click through rate and cost per acquisition. And then I’ll ask, yeah, when you listen to an investor call do you, do you. from a public company on the stock market, they almost never talk about the same metrics.
They talk about profits. They talk about revenue, right? So where is that disconnect? And what do we do when there is actually a disconnect between the metrics that Google gives us and the metrics that we care about as a business? So that’s the topic for today. So let’s get started with PPC Town Hall.
All right. And our two experts for today. Welcome, Frederik. Welcome, Andreas.
Frederik Boysen: Hi.
Frederick Vallaeys: Good to see you again, Fred. Yeah, I know. Usually we meet up at SMX advanced in Seattle. So let me start introducing you, right? So Andreas Reiffen. So he’s the founder and CEO of Creolytics. And Andreas, I really respect because I’ve seen many of his presentations at SMX advanced.
So that’s like the smallest version of SMX that they do, but it’s also where they bring in like the world’s best experts. And not only is Andreas a speaker at that event, he’s also one of the, what they call mad scientists of PPC. So he gets to put on a white lab coat and share some of the crazy research he’s been doing over the past year.
And I’ve always really enjoyed the presentations that you’ve given and the insights that you bring about you know, how things like automated bidding are working. How you optimize for you know, for e commerce. So Andreas, tell us a little bit about yourself. Where are you calling us from today?
Andreas Reiffen: I’m in Berlin actually currently, and I missed out SMX events, unfortunately like all the different conferences.
So I’m really glad to join here remotely.
Frederick Vallaeys: Yeah. And well, you didn’t really miss out on it, right? Because it didn’t actually happen. But I also understand you’re stuck in Germany right now.
Andreas Reiffen: I’m stuck in Germany. Yes. I took a flight to Berlin where we are headquartered. I’m usually based out of New York and I expected that the travel ban would be lifted with.
which didn’t happen. And if you’re not a US citizen and neither have a green card, there’s no way back for you. So that’s why I’m sitting here and waiting for change. And eventually I’m going to take a hack, traveling to some other locations, staying for two weeks, there’s two days, and then take a trip to the US, which might work or eventually not.
Let’s see.
Frederick Vallaeys: Okay. Well, fingers crossed you come back soon. And it looks like you’ve been able to keep up with a haircut. Although I Assume Clippers were involved.
Andreas Reiffen: Still my Corona cut which was the best way to get over the crisis. But I, as hairdressers are open again my hair is going to grow back to where it was.
Frederick Vallaeys: And then I want to bring in Frederick Boisena here. So Frederick, you run a company called Profit Metrics. We haven’t really met in person yet, but I did have a talk with you a couple of weeks ago and really liked the topic that you brought up. And I thought it was very relevant to what we’re doing here today.
So you’ve been around in this industry a while too, right? So tell us about you.
Frederik Boysen: Yeah, I’ve been in the e commerce industry for like 12 years, and I run a company called Profimetric, where we help e commerce companies get some transparency in what they’re actually making in profits after ad spend and after a fixed cost, and enable them to get the real profit inside of the marketing channels so they can optimize and Measure with the real prop.
Frederick Vallaeys: So how did you become so smart on this topic? Did something bad happen along the way?
Frederik Boysen: Yeah. Yeah. I had a a scale up an e commerce company where we scaled really, really fast and, and top line was growing. And we were like everybody else using a target ROAS that we calculated. And that target ROAS I had some customers that loyal customers coming in, taking bigger.
Basket sizes, bringing up the average was because he made more profit. So the role of target I used on Facebook and Google for new acquisitions was actually wrong. And I thought everything was good. Made face case studies with Facebook and Kissmetrics and, Suddenly I realized that I was actually losing money.
So I actually scaled into a bankruptcy and started all over, built the company up build my own little tool, which became profit metrics, made an exit on this company when it went profitable and went in and made this tool available for others.
Frederick Vallaeys: Well, I’m sorry about that that you had to take the hard way to get there, but it looks like things are going well now.
And by the way, I love the the Scandinavian minimalist design that we have with your black background. Sorry, white, white on white on black. It’s, it’s amazing. It looks great. And Andreas, love your neon sign there that says Creolitics as well. So the folks watching, if you want to say hello to the panelists or ask questions, just put that in the chat box.
In fact, if anyone wants to say where they’re calling in from today, we’d love to see that and put it up on the screen. Andreas tell us a bit about your story. So the what was your intro there? We didn’t really talk about this, right? You’re kind of like a rough path to get here to become the expert.
Andreas Reiffen: So what happened? I was in the marketing lecture at university and the professor had talked about Google Edwards and it’s so amazing. You can bid now on keywords and you can position your your ads on the first position. You generate a lot of traffic and that you might not be profitable, but you can bid down and just pay a little bit and then you don’t get traffic.
So somewhere in between there’s this sweet spot. And then I started reading and it was the times of these File sharing services, emule and BitTorrent and all that stuff. So I scrolled around and came across a book, which said, get rich in 30 days with affiliate arbitrage. So open up your Google AdWords account sign up in the Amazon partner network, buy the traffic from Google, shift it over to Amazon.
You’re going to be rich. So I tried and when all my pocket money was gone, I decided to apply for an internship went to South Africa to a company called clicks to customers. And that’s where I actually learned how to properly do it. Is that Vinnie Lingham’s company?
Frederick Vallaeys: Sorry? Is that Vinnie Lingham’s company?
Exactly, Vinnie Lingham, yes. Oh my god, you have a Vinnie connection too. So Vinnie was actually my neighbor for many years. So after he moved from South Africa. For those of you who don’t know him, so Vinnie was like one of the mega affiliates back in the day. He started a whole company in South Africa around it.
And then eventually moved to the US. And when I lived in Palo Alto, he moved two blocks away from me. Oh my god. And then I moved to Los Altos and then he moved as well to be two blocks away again. If you, if you
Andreas Reiffen: meet him, say hello. He knows me certainly. It was the early times, 2006, I went to South Africa and actually interesting and very related to this topic here.
Then he had a huge block, really successful block at that point of time. And I published an article saying ref share and affiliate is stupid. You should share profits because it’s better for the search engine, better for a publisher and better for the affiliate. And this really made it I think into the number one blog post in five years or so.
Research guys from Stanford and Harvard who commented on this because everybody was researching keyword auctions and all it was the new kid on the block at that point of time and and this is how I got started and I really got excited about the whole thing and then founded Creditex and I’m continued with this profit driven thinking, which is probably the same rationale as Frederick and Boyson has in mind.
Frederick Vallaeys: Yeah. So yeah, let’s shift into the topic, right? The profit driven as opposed to the ROAS driven, but you wrote this post here it’s a couple of years old now, but I guess it’s kind of teeing up what we’re talking about. The vicious cycle of ROAS targets is killing your business. So Frederick, I guess the exact story, right?
It literally killed his business. It did.
Frederik Boysen: Yeah, the problem is just with the ROAS that return on ad spend has become revenue on ad spend in e commerce. It’s, it’s not return. It’s, it’s, yeah, it’s wrong to use it, in my opinion, the ROAS, because you cannot really see what you’re making from, yeah, revenue.
Frederick Vallaeys: Right, and that was always the interest. So I worked at Google, obviously, and I would run into like disagreements with Hal Varian, the chief economist on what or how that was supposed to be defined. And was it like the, the profit that you made or was it the revenue of there’s just like what was in the basket?
And obviously there’s a huge difference between the two, right? So explain a little bit like what is the absolute worst way that you can measure? Return on ad spend to eventually kill your business.
Frederik Boysen: Me? Sorry. Yeah. Yeah. But, but in general, it’s just to use an average of all the different combinations with different margins on the product, discount codes, free shipping, to use an average year.
You’re like testing on one side of it in your. Marketing platforms. And in the other end, you are using the revenue and just using an average to calculate back if it’s profitable, it actually don’t make any sense, but it has just become the standard in e commerce today.
Frederick Vallaeys: Right. And so I guess one thing that you started alluding to is the fact that you take averages, but you have different margins on different products. And that’s sort of the underlying problem here, right? Is that if you. If you look at return on ad spend and you look purely at the value that’s in the basket, what it could be a huge sale.
There could be an item that’s being returned. It could be something that you actually literally don’t make a profit on. So yes, you’ve generated a scale and it looks like perhaps that sale is within your return on ad spend target, but you haven’t made any money for your business. Exactly. Andreas, how would you explain this one?
Andreas Reiffen: It’s been a while that I published this article, but let me maybe. briefly walk you through and I can tell a brief story about it. How we experience this. So as I mean, Fred, or the two of you, you probably know this usually I’m working with someone and ROAS is the target. And sometimes it feels like it is what it is and just continue doing this.
If you then take a look, what truly happens is the following. There’s finance. And they set they define the budgets and they set the targets as they don’t really fully understand performance advertising in detail. They have some profitability metrics, new customer acquisition targets in mind.
But they give marketing, advertising a ROAS target. What they, what then happens is that on the marketing side, you try to achieve exactly this target and could be maximized revenue out of a given budget. So if you just do this blindly, you follow this target. What you tend to do is you tend to sell products which have low margins.
products which have high return rates and you tend to sell all this to your existing customers. Why
Frederick Vallaeys: is that usually the case?
Andreas Reiffen: That’s the case because there are smarter advertisers out there and this is the cheapest traffic you can simply buy. And if you sell to existing customers, for example, incrementality is pretty low.
And and then what happens is you have this, let’s say one year cycle. And towards the end of that budgeting cycle, finance will analyze the numbers and they will realize. Oh, my God. We achieved our target. Raw s is fine. But for some reason, we do not understand profitability is down. We had calculated with a higher new customer acquisition rate.
We have calculated with higher margins and lower returns. And why is this wrong? So the problem here is not the marketing side. The problem is the target itself. Whenever you put out a raw s target, right? You only can do the right thing if you have people who kind of move away from that pure target and do things which might make sense for the company, but actually make the numbers worse.
And that’s,
Frederick Vallaeys: this is interesting too, right? So you’re saying you’d have to have people doing the right thing. Like what about automation? Does Automation fail equally much? Does it work better or worse? I think
Andreas Reiffen: the only way to keep up a system which is based upon ROAS is that you start layering over hundreds of different rules.
Product category A needs to have a ROAS of X. Product category B or geolocation B or whatever segment you want to take needs a different ROAS target because new customer acquisition rate is higher. And then you create this really difficult monster and once people leave, it’s just impossible to maintain.
So you’re
Frederick Vallaeys: saying, and just sorry to dumb this down a little bit, right? But you’re saying for different geo locations, you have to put different rules in place because you might find that in New York, you have higher net new customer acquisition versus maybe in San Francisco, which is more. Returning,
Andreas Reiffen: I think Fred, if I recall right.
At that last SMX so basically one and a half years ago, a little bit more than a year ago, you said this, that it’s hard to aut, automate things fully because you might not want to have the same target across all. different geolocations. And you mentioned one example. If you know that in whatever San Francisco customers are way more loyal than in New York.
You want to have a different ROAS target in place and you can have two separate ROAS targets or you work with a different data ingestion. So you could reflect in the numbers right away, that we, that San Francisco is more profitable that New York. And then you, you could. You operate on a true, whatever, lifetime ROI or something.
So the question is, do we fix things through a few different rules? Or do we fix them through data in a generic way?
Frederick Vallaeys: Exactly. And I think what you’re saying here is you have two places where you can fix the problem, right? So you either, Build this perfect conversion tracking attribution system that takes these things into accounts, takes these things into account, or you fix it on the back end and you run like the generic ROAS, but you have all these conditions on it that kind of like tweak things along the way.
Both of them can be complicated.
Andreas Reiffen: And the difficulty I mostly see with that approach is that let’s take this example again. San Francisco highly profitable customers, New York, not so profitable. It’s just a random example for the sake of making my point. In this case if you run it in an automated way and you aim for the same ROAS, you get the highest revenue because it would level the ROAS across these two locations.
If you now have a smart guy sitting there who knows that San Francisco is much better, what would he do? He would lower the rawest target in San Francisco to push harder while he would increase the rawest target in New York. So overall his numbers would look worse than before while he has made progress.
And this is always very difficult to maintain on the long run if your progress is not reflected in your numbers.
Frederick Vallaeys: A huge problem, right? And Frederick, I want to bring you in here. But so Michael is asking some really good questions here. So, I mean, it kind of starts with how ready do you think the market is for not just looking at ROAS?
And then I think the problem is like Andreas is explaining, these are the metrics that the business tells us we need to go after. How do you explain to someone what we’re doing here? I mean, it’s hard to explain this between the three of us because it’s such a
Yeah, but in
Frederik Boysen: regards to, to address also like return rates, for example, he’s coming from Germany. It’s not it’s normal for like a clothing company to have like 30, 40 percent returns in Germany because they come from a past post order catalogs where they buy big orders and then try the clothes on and send it back where in Denmark it’s maybe 15 to 20%.
So there’s 70 percent in Germany
Andreas Reiffen: many times percent, 70
Frederik Boysen: percent yeah. So, so, so that could be, but, but what I’m thinking is you need to calculate the real gross profit on every single order and then send it into the channels. And then of course you need to do the attribution and look at the customer lifetime value, but on the first attribution.
You need to work on profit. So what we do is we use like, it is actually return on ad spend, but we use POA as profit on ad spend just to make it like, you know, it’s profit. So we calculate all the orders. The profit on them, the cost price, the shipping cost and payment fees. And then we send it into the platform so you can make like decisions that are transparent and then you can make the decisions of like what Andreas is saying with, with locations and stuff like that.
But it’s, it’s, it’s like a transparent KPI where revenue is just. In e commerce. Yeah. It’s too dangerous to work on. Right. So let’s break it
Frederick Vallaeys: down into the two stages, right? So let’s put better information, better data into the ad engine. First of all, so that we have better understanding of what’s happening.
And that Frederick, that’s what your company does, I believe. Right. So how do you go about a project like this? How complicated is it? Who’s buying? Do you need?
Frederik Boysen: Yeah, but actually we, we, we, for example, have modules for most the e commerce platform. So we get the cost prices in. Then we get shipping cost payment fees.
And when an order comes in we calculate the profit in real time, and then we send it to, for example, into Google ad server side or Google analytics server side or Facebook ads. So you have a profit with that
Frederick Vallaeys: revenue, but you take out the shipping cost, the promo codes that were added. Yeah. You can take out like the affiliate commissions that may have been paid somewhere.
Frederik Boysen: Yeah, yeah, exactly. Yeah. And, and, and, and discounts when you run sales and stuff like that, people often tend to not. Tell their, their agencies that the role should change in that. If you’re using a ROAS in that period of time. So they’re discounting heavily and still using the same ROAS targets, and then they’re losing a lot of money on the sale where with, when you optimize for profit.
It will be taken into the calculation that you have given discount codes or sales and
Frederick Vallaeys: stuff like that. So your system takes basically the shopping cart system and layers and all the costs of that sale. And then tells you what was still left after that, which hopefully is the profit, and then you take away from that, how much you actually spent with the ad engine to drive that sale.
Yeah. But you were left with a positive number. Hopefully. How do you account for a return? Like Andreas was saying, 70 percent return in some categories in Germany. Okay.
Frederik Boysen: We are, we are working on returns because we are going server side now, because I also see there is a big problem regarding client side tracking where people are missing, maybe 10 to 15 percent of all their orders in the platform.
So the transparency of the data is, is actually flawed. So, so we are going service out now, so we can also do some of this return and also try to forecast it because I also see it as a problem that the return maybe happens 30 days, 60 days later. Okay. But the guy sitting with the marketing have already done the optimization.
On either the profit or the revenue right now. So at some point it should be forecasted on, on there to be really right. It’s really
Frederick Vallaeys: cool stuff that I’ve seen and not specifically in retail, but is building correlation models and regression models. That take today’s signals of like, what was this user like?
And what were the attributes around it? And then let’s compare it to other users that we’ve seen over the history of the company and how likely they were to return and how similar this customer is to those people. Likely to return and so by by that way, you don’t have to wait 60 days for the return to happen But you can make a prediction today and send that to the engine so that even starting tomorrow The ad engine is going to try to give you more of those types of customers that have the good signals to good attributes
Frederik Boysen: Exactly.
So so just like like for example, you can take with returns. If people have like two sizes in the same order, there’s probably a big, big chance of 50 percent of that is going to be returned. So you can already calculate it from that or, and maybe on the brands or product, if you have the data there. So it should be possible to forecast that that’s something we’re working with now.
It’s not something we have, but it’s something a lot of customers have.
Frederick Vallaeys: Andreas talks a little bit about how you guys at Crealytics deal with you know, looking beyond the raw as I’m putting the right information in the system.
Andreas Reiffen: So, I mean, what you just said, how you phrased your question, putting the right information into the system.
It’s a bold statement because I think that’s exactly what needs to be done. And it sounds so simple, but I think, Things are changing fundamentally if you compare what we did like two, three, four, five years ago and what we have to do going forward. That’s the fundamental difference. Before we were optimizing keywords and ad copies and landing pages.
There were lots of things which are now fully automated by AI. But AI needs data input, so you manage everything through data. So providing the right information back to the system is key. And what we see that there are many more predictive elements in there. Because I mean, it’s very obvious that revenue doesn’t make sense.
If you have different return rates, different margins, revenue is just a very silly KPI, which makes things simpler, but it’s, it’s never as good as working on profit unless you have an algorithm with the same margins and same return rates, which is totally unrealistic. You never find that. And then you have the predictive element when it comes to returns, as you just mentioned.
So the way we are solving this is that we take a look at The product itself and historical return rates, if not enough data, we go to the parents queue and take a look at these return rates and then we move layers up and get more information in and try to come up with a prediction when it comes to the return rates and then it’s Ultimately, we need to put back a value per click when the click happens.
And this must be reflected in the system. And what is the value of a click? That’s the profit margin of the product sold minus the expected returns. Plus, what I expect to come in, in terms of profits over a certain period of time. This could be just a day, then I’m optimizing just for profitability.
This could be a year, then I have some portion of a customer lifetime value reflected. And if I make this visible to the bidding, AI is so smart already today that it will react based on what you put back. I think we won’t have to care about too many other things. But if we get this right, I think we can actually benefit a lot from the tech out there.
Frederick Vallaeys: Yeah, and so that’s a point that I see quite frequently is people use automation of some sort, and then they complain that it doesn’t work. But it’s not that the AI was broken. It’s that the. Their source of truth for that business was not the actual source of truth. And, and then there’s this funny example too, right?
So I mean, I use this one quite often, but as far as last click attribution versus something more interesting, like data driven, right? So, If you have a last click attribution model and you’re selling sneakers, all the credit is going to go to the very specific search right before the consumer buys that sneaker.
No value goes to the brand searches, the generic searches like sneakers. If that’s the data you feed into an automated system, the automated system is going to say the keyword sneakers is not relevant for a company like Nike or Adidas, which is completely incorrect, right? Because it’s super relevant.
Like that’s a huge part of their business. But if you didn’t get that AI system insight into that, like, mistakes are going to happen. And the reason that this was fine for a lot, many, many years, we all ran on lab click attribution. And it’s because people like you and me, we were looking at making the bit changes.
It wasn’t the machine doing it. And we’d see the keyword stakers and we’d see very low returns on it. And we’d be like, well, that’s fine. That’s probably like an upper funnel, currently consumer journey keyword.
Andreas Reiffen: Whoever is curious what the impact of What we would call data activation is in fact should do one experiment.
I really recommend doing this. You let T ROAS run and in one instance you provide first click data and the second instance you provide last click data. What you will see, and then you analyze the retargeting share of the first click setting and the last click setting, what you will see is that you will have a tiny little retargeting share if you do first click, because credits are all allocated towards the first click and the engine learns This retargeting stuff just doesn’t move the needle because there’s no credits coming in, so it pulls back and vice versa.
So it’s very clear you use the same bidding engine, the same tech stack, everything is the same, but you just flip the data input. And I mean, if you test for incrementality, what is very clear, the difficulty in general we have today is that the more inclined users already are to buy, the better the results will of course look like, but the lower the incremental impact will ultimately be.
So, nobody will can today answer the question whether you should buy or not. Actually build up on your retargeting stuff or whether you should rather bit down because you have two effects into into different directions, right Last click kpis if you do lower funnel retargeting but low incrementality, if you don’t, nobody really knows what the true the right strategy would be.
Frederick Vallaeys: Yeah, and incrementality is this huge topic and I know you talked about it quite a bit last year at SMX Advanced. And the problem I think is with automated campaigns like smart shopping campaigns, which are fully automated. A lot of customers use them. A lot of customers say the results look great, but the key word here is they look great.
But they can’t look inside. Like how much of this is being driven by retargeting versus you know, actually new customers. And that’s the problem. Say again, Frederick.
Frederik Boysen: Yeah. Branded searches.
Frederick Vallaeys: Often that those credit for that. Yeah, exactly. So, so how do you, I mean, how do you go to a customer and make them understand that, yeah, the numbers may look great, but you got to look deeper.
I did you explain that a lot or do you like what’s the education portion here? Because it’s so easy to gloss over the numbers and just kind of like, take what Google is saying. Because what both of you are saying, these are complicated projects, right? But these are the right approach to doing things, but it’s not the easy way.
Andreas Reiffen: I personally believe that attribution is dead because it has just entirely failed. There are actually, I unfortunately don’t have these articles at hand right now, but there’s one about eBay, another about Topshop, and a third one about Adidas. And they’re all along the same lines. Basically they stopped advertising and they realized nothing happens.
Nothing happens. It’s just continuous business as usual, and they haven’t lost anything, and they believed in phenomenal numbers. And ultimately what attribution does, it assumes that advertising was responsible for the sale. And then it allocates the credits to the different clicks based on whatever time decay and whatever models you might find.
And I mean, if you have 10 retargeting clicks and someone has just recently visited the site and put 50 products in the basket, It’s quite likely that this person used a shortcut and would not buy it. So the only thing which can prove value in my opinion is running isolated increment, dedicated incrementality tests, audience based where you can truly find out what the impact is.
And this is what I feel the only field which will never ever truly be tackled by the large companies. It won’t be tackled to the full extent by Google or Facebook, because there’s just no interest in making fully transparent what the true impact of everything is. And that’s the job of the marketer to figure out.
I really get it.
Frederick Vallaeys: It was crazy because at the Google marketing live event last year, I was talking to some of the product managers on smart shopping. And I was like, well, one of the reasons we don’t want to use it is because we don’t have insight into the incrementality and like, what are you putting too much into remarketing?
And just the way that they looked at me, they were like, wait, that’s, that’s a concern. And meanwhile, like search engine land was a blaze about this topic, right? So there’s sometimes such a huge disconnect. And I don’t think Google necessarily wants to hide it or if they are trying to hide it.
They’re doing a very good job looking surprised what we did. It just seems like sometimes,
Andreas Reiffen: I mean, Hey, we need to understand this as well. They’re running a business. And if it says in the end it’s all about ad sales. And it’s fine. If you buy ads, you need to know that the numbers you see. have nothing to do with reality.
It’s some random number you get, and the job is to really figure out what needs to be done. And I think that’s, that was from the very beginning, the biggest issue of all the different bidding tools out there. Because you, they based decisions on some numbers which were tracked without even attribution in place, without incrementality measurements.
So all the decisions were based upon a random number. And then you have a better algorithm operating on some random number, which, Doesn’t really make sense in all the cases. I think the market is changing in that sense because there’s so many things have gone away that what stays as the measurement piece and really understanding how much money you put into a different channels or even sub segments within channels.
Frederik Boysen: Well, what we do actually is, is we have a dashboard where we We take away the attribution fully. So we actually look, what are your gross profit? What are you using on ad spend? And what are you making in gross profit after ad spend? And what is the gross profit after ad spend ratio? So you can actually see, okay, do I use 10, 000, more this month or the last week.
Can you see, will our gross profit increase? Will we keep those ratios the same? So after that, you can start trying to allocate it with some attribution, but actually looking at what is, what is happening to the company, what is happening to the bottom line. without looking at that attribution that look really great in Facebook and really good in Google.
And there is some mismatch over the orders. So you can actually just look at the financials instead of looking at the attribution.
Frederick Vallaeys: So I think we understand the problem. The measurements that we’ve been using are wrong. The data that we’re sending into the systems are wrong. Let’s solve the problem here, right?
For the people listening in obviously they can. Use your tool, Frederick, they can work with Andreas and Lytics, but how big of a client do you have to be? How much do you have to spend? And where do you start with fixing this problem? What’s step one? If you haven’t done any, if you’ve just been using ROAS as the target, where do you start in?
You start,
Andreas Reiffen: doesn’t matter. You started, so go ahead.
Frederik Boysen: For us, you can start with 5, 10 hours a day, up to as many as you want. It doesn’t matter. It’s just to get that. I think it’s really important to get that right. KPI, that transparency all the way from the start when you scale, but of course it gives you a bigger impact of the larger customer you are.
But, but it’s the same problem, no matter how big or small you are,
Frederick Vallaeys: you’re saying it doesn’t matter how much you spend, you can put your tool in place and at least start sending in the right data. And then it just seems a matter of over time building up data points so that the system can start optimizing against it.
Frederik Boysen: Yeah,
Frederick Vallaeys: by the way, what’s your entry price point for your software?
Frederik Boysen: It’s around 20. 200, 250 orders included, but we go up, we charge per order. You get in. Okay. So that’s great. Yeah.
Frederick Vallaeys: And that, and that works with Shopify,
Frederik Boysen: Magento, WooCommerce. Yeah. All the platforms. We can also do universal integrations through tech managers and stuff.
So it’s really easy. As long as you have the cost price on the products.
Frederick Vallaeys: Like a good starting point. Doesn’t seem to be too expensive. How long does it take to set up an integration on a Shopify
Frederik Boysen: or a Magento between 15 and 30 minutes? That’s all we have one click installers for everything.
Frederick Vallaeys: It sounds like everyone should just go and do that today.
Then, of course, of course, Andreas, what about in your case?
Andreas Reiffen: We have a totally different focus. We are rarely managing accounts less than 10 million a year, so it’s rather enterprise level and totally different approach.
Frederick Vallaeys: 10 million a year. Are you talking ad spend or revenue?
Andreas Reiffen: Spent 10 million per year.
That’s our sweet spot. It’s interesting where we invest most of our work because it’s always perceived as being a very technical problem. And in my opinion, we are solving to a great extent an organizational problem. Because there’s a marketing department, there’s finance, there are budgeting processes there are ROAS targets across the organization and now everything needs to change.
It starts from the, all the data delivery, data ingestion and activation, which is the technical side. This can work very fast to very long depending on what setup clients have and you never know what you find. We’ve seen this working in a, in a few days, and we have seen projects where it takes a year or longer because they have no way to access the data and until you get everything a year has gone by, but essentially there must be a lot of education across different departments to be able to actually implement this.
And what we’ve seen is. Some, let’s say shadow methodology just in marketing, which isn’t understood by other departments. And this never really works. And you need many very senior stakeholders to buy into this. And sometimes you can make it work technically, but then there’s someone, yes, but now budgets are down because you don’t hit the certain ROAS target.
And you have to say, stop it. We don’t have ROAS targets here. We consciously decided to go after. long term profits, castable lifetime value. And this is in large corporate organizations, sometimes the most difficult piece. And it’s very, very time consuming. And we have delivering this piece as well, which is by the way, something which is more and more tackled lately by the large consultancies like McKinsey.
They go they are going and review is marketing aligned with the overall business strategy and the overall business strategy in retail is Fully focused on customer lifetime value. Why market marketing is doing something else. And that’s how we connected.
Frederick Vallaeys: That’s interesting. Right? So, and we have a lot of PPC agencies and consultants who watch.
So as more tools from Google are becoming automated and bidding is automated and targeting is automated and ads are being put together on the fly based on the components that you have, and then you’ve got the McKinsey’s coming in and helping people understand the bigger business problem. Like how would you recommend someone shifts from being a PPC consultant, PPC agency?
To being relevant, say three years from now.
Andreas Reiffen: I think it’s the strategy part is because as I, I think we, as we agreed upon initially, if the target is wrong, if there’s a misalignment between the overall company’s strategy, what the business wants to accomplish and the marketing side. That’s, that’s a fundamental mistake.
And no matter how good you are, your operational day to day work, you will never, ever close that gap. I think having the strategic perspective is highly important. And the second piece is, I think it’s, it’s gonna purely shift to measurement. The large publishers won’t want to fully tackle this problem, even if they want it.
You don’t want to let them do all this because in the end, this It’s your money and you need to understand what you are getting back. It will become the number one capability going forward.
Frederick Vallaeys: And how much of this measurement is because of the walled gardens? And so Google is not going to share data with Facebook, it’s not going to share data with Microsoft, it’s not going to share it with all of the others.
Is there a place where you are aggregating that and then making decisions across the board?
Andreas Reiffen: I mean in the end there isn’t any Reason why you should operate just a single channel, right? Because what we are doing is we are let’s call it assessing the business value of Advertising clicks as they happen and it doesn’t matter whether this click happened on facebook on instagram on google or somewhere else.
We just need to put, put that value back. And this is why I think we, you have this kind of Whole layer of measurement, which spans all the different channels. And then you adjust it into the different channels if possible. And there are of course, massive limitations on Facebook, for example, it isn’t that easy.
Also Google, it’s not as easy. If you want to attribute a specific value to a specific click in the journey, it’s a question mark, whether this is still going to work going forward. It’s smart shopping. No way you, you just can’t do that. So. you lose control to a certain extent, but how do you get as close as possible?
How can you kind of hack it? Because in the end, I mean, I think it’s going to work through incrementality testing across all channels and then measuring progress in terms of incremental impact. It looks easier, but the complexity actually is even getting higher. My opinion.
Frederick Vallaeys: Okay, so buckle up and be ready for a significant project, but it’s worth it in the long run.
Okay, Andreas, Frederik stick around. We’re going to take a look at the news here, so everyone stay tuned.
All right. How do you like that music there? Okay. So first the news article that I pulled up here. So search engine landed another study asking people when they’re going to be traveling back to conferences. And the majority right now says no in person events until the second half of 2021. So, summer of next year, basically Frederick, you’re in Denmark, everything’s back to normal, right?
Like, yeah, they’re starting
Frederik Boysen: off in August, August and September, there is some events that are supposed to be scheduled there. So I think it’s allowed now to do it. I think it’s up to a hundred people now, and then it will become 500 people. But. They say in August, they should be open for, for as many people.
Would you
Frederick Vallaeys: guys as speakers go back to an event in 2020? I don’t know. Andreas? When I have a speaker
Andreas Reiffen: gig in 2020,
Frederick Vallaeys: yeah, I mean, if something came up, would you actually go and speak in person?
Andreas Reiffen: I would I’m, I don’t have a reason to be scared. I went through COVID already, so I’m fine. Whatever is needed, I can do it.
It’s unlikely that’s gonna happen. I mean, as everybody knows, we’re also running our own conference New York. It’s called New York. No go about retail media. And it’s always hard to to get the attendees and everybody who’s running conferences already struggling. I’m bringing attendees in and now it’s probably almost impossible.
So you might end up with a quarter of your attendees with a by doubling your effort. And I think there’s no point about running a conference this year. And I hope it’s going to pick up at, again, at some point, I feel it’s working and we can have these just virtual conferences, but it’s also a different situation because we still know each other from.
physical experiences, right? And at some point, I think it makes sense to see people in person again and have a real touch point and not just yeah remotely. So I think it’s okay for a while, but I’d like to yeah, get in touch again with people from the industry. It’s a different feeling being in the same room, but it’s still possible to get the content across.
Frederick Vallaeys: Yeah, I agree. And I’m happy. We have a Frederick on the call today. We haven’t met him at SMX before or any of the other events. So I’m starting to build new networks. It’s it seems to be possible even in the virtual world, but Something to be said for having a beer, carrying a little bit too much information and you know, getting some cool details about how stuff actually works.
All right, so the second article I pulled up here Google ads increased visual options for shoppers and brands. So this was announced just about two days ago, but they have image ad extensions. So that’s what they might look like on mobile devices. So put an image next to your ad. So obviously shopping ads have been huge with images included in them.
Consumers love them, but now there’s more images as well coming to text ads. Smart shopping, let’s actually talk about smart shopping for a second here. So this is evolving. Google is automatically pulling in data like free shipping, special offers that you have maybe mostly for Frederick. Smart shopping, people using it.
Do you like it?
Frederik Boysen: I see different results right now. I’m actually seeing it’s working pretty well when the jump to using cross profit in the target robots inside of Google, but, but as far as I see it, the main problem is you don’t know what is going on, like we talked about before, and Sometimes Google predicts to maybe take some products that would maybe not have a really good margin on, but they can make a good revenue and you hit those raw raw target that your financial department.
But you’re actually not making a lot of money when using smart shopping. It depends.
Frederick Vallaeys: Andreas, I’m going to assume if you have 10 million of ad spend a year in an account, you’re probably not running a smart shopping campaign. Correct. So what does the ideal structure look like in your mind for shopping?
Do you split out campaign by ROAS targets? I said by product lines. How do you do it?
Andreas Reiffen: So what we mostly currently focus on is actually on the data and just data activation piece. We’ve started tech levels. One is what we discussed initially. What is the actual order margin excluding expected returns?
The second piece is we calculate or estimate a customer lifetime value. What we do is we, if we know it’s a new customer, we put it into three buckets. High value, mid value, low value.
Frederick Vallaeys: How do you split, sorry, I’m missing how you split your campaigns around that.
Andreas Reiffen: We don’t split our campaigns based on that.
We split by existing customers and new customers. So we have two buckets. So audiences, exactly.
Frederick Vallaeys: Okay. And, and what do you, so how do you start with your campaigns though? I mean, do. Tend to group products by category. Actually, I haven’t done
Andreas Reiffen: this in quite a while, Fred. I’m not sitting in front of the computer anymore.
So I what I know before we switch to fully automated TROAS, what we did is we split by query type. And we did this very, very early on. So we split by
Frederick Vallaeys: Sorry, Martin Richarding principle.
Andreas Reiffen: Yes. So I mean branded terms with a specific product name in their second category, just branded terms without specific product and generic terms.
And then we differentiated the bits in order to get the most specific traffic with higher bids in. This is no longer necessary, the structure, and we see that TROAS actually doesn’t benefit from having this structure. I think the whole how we structure campaigns in Google shopping isn’t as relevant anymore since TROAS is operating on this and they have all the information anyway.
So as soon as you flip, I think this becomes a secondary problem and we’re rather trying to fix it by. Data ingestion,
Frederick Vallaeys: data ingestion. Okay, cool. So you campaign structure doesn’t matter that much anymore But then you do have different audiences based on expected returns expected value And that’s what you overlay on top of those campaigns.
Andreas Reiffen: No, we just we don’t split this by audiences we what we what we do is we just segment between existing and new customers because Different. I mean, there’s different value. A new customer always has a customer lifetime value on top. And we, we’d like to just segment this to specifically target these guys.
The whole rest. Of course we set up audiences. For example, whatever cart abandoners. What we then do is on the, on the attribution side, if we find out that this is low incremental value, we can manipulate the conversion values, so the whole audience. segment gets low credits and the bidding pulls back.
Yeah, that’s how we do it. I just want to bring in another screen here. So Google just launched predicted predictive metrics. This is only about one or two weeks old. This is not showing any pretty screenshots, but we’ll share it in the show notes. But so if you don’t have the really high end capabilities that, you know, Craylytics and CoffeeMetrics have, this is something that’s baked into Google Analytics.
Frederick Vallaeys: It will tell you audiences based on predicted churners, predicted high value shopping carts, and those types of things. So it’s a really good way to get started with some of the things the panelists are talking about today. All right. This one’s sort of near and dear to my heart. I know it’s not as important for both of you, but I do want to talk about it because it is PPC news.
So privacy shield, which was the data transfer mechanism for GDPR for companies that need to get that data into the United States has been invalidated. So that’s kind of a big deal. So any American company that works with European customers kind of has to revisit how they handle privacy related issues.
So that that’s something that’s, it’s brand new. If you haven’t heard about it, do take a look at it. It is important. User privacy is obviously only gaining more and more momentum. And this is changing some of the ways that we’ll have to take care of making sure our users data. Is safe and secure. Also in the news this week, Microsoft advertising now gives you free stock images for audience ads. Now Andrea is probably, you know, making different image assets or ad assets, not a big deal. I spent 10 million a year. But Google and Microsoft are all doing a great job for those smaller accounts that want to run video ads that want to run image ads.
Making a lot more of those resources available at no additional cost to advertisers. And then let’s see, the last one. Google adds new features to responsive search ads. Have you guys used responsive search ads?
So responsive search ads are where you give Google different variations of headlines and description lines, and then they automatically put together the ad based on the machine learning and what it thinks is the best variation of the ad for that specific user in that specific scenario. They were restricted to static text, less dynamic keyword insertion.
So ad customizers did not work, but now they do. So now you can automatically put in the location that a user is searching from in a, in a, Dynamic ad basically you can put that in. You can also do a countdown timer. So if you have A special offer a special sale that’s going to last for a few days You can actually put in the end date so you can say only six hours left with our huge summer sale And that can automatically be shown up and that six hours obviously is going to change depending on how close we are to the end date of that sale. Do you guys have any impact, any thought on the impact of putting the promotion in the, in the ad text, putting specific discounts, shipping offers, like what’s the, what’s the most impactful thing that you can put in an ad to.
Andreas Reiffen: I would say it’s really hard to say up front. I mean, it’s, I recall the old days of our affiliate text ad campaigns and you it’s, it varies based upon specific keywords. One can work better for one keyword. The other works better for the other keyword. I think it’s just having the variation in there.
So. multiple variations Google can choose from and then let Google just take That’s the only thing you can do ultimately.
Frederick Vallaeys: We have an audience question here. I don’t know who wants to take this one oh and the text is getting a little bit jumbled. So i’ll read it. But if you have a product where remarketing and repeat purchasers are an extremely small part of the sales mix with smart shopping be a perfectly viable strategy
Andreas Reiffen: I think the problem with smart shopping is mostly that you don’t know exactly where your traffic is coming from and whether this is prospecting or website visitor retargeting.
And if it’s heavy, heavy, heavy, lowest funnel retargeting, The value is just really small, and if there’s a heavy prospecting part in there, then it’s fine. I mean, you could try running this with a first click setting where you don’t incur the risk of getting too much retargeting in. But that’s the lack of transparency.
We don’t like about it because I mean, it’s apples and oranges, right? Whether you get or retargeting and you get one KPI, which is not very telling if you don’t know where it’s coming from.
Frederick Vallaeys: All right. So wrap up here. What’s the the big thing that I didn’t talk about you guys think we should tell me one more thing before we sign off here.
So Frederick, how about we start with you?
Frederik Boysen: Yeah, I think we got
Frederick Vallaeys: through it.
Frederik Boysen: Every, we got through everything or most of it. I think it was really good talking about that attributes and thing with Andreas that it is getting harder and harder and taking a step backwards and, and, and looking a little bit more at.
How the business is actually doing. That’s also a way to figure out what to do with your budgets and not like overspend because using smart shopping and stuff like that, you don’t know how much is retargeting. So you could use a lot more spend that you’re actually actually was necessary to get those sales.
Frederick Vallaeys: What’s a, I know it’s hard to predict things in uncertain times, but what’s your prediction of what’s going to be the most important thing in 2021 for e commerce advertisers? I think it’s profit. Oh, it’s profit, huh? Well, we all like profit.
Frederik Boysen: Yeah.
Frederick Vallaeys: Andreas, I think it’s game here.
Andreas Reiffen: I can only repeat myself what I said before.
I think it’s, everything is on the measurement side going forward. And it’s really hard to truly figure out what you are getting. And as AI is taking over, it’s, The only thing to manage AI is through data input. There is no other way. And I think that’s the key reason why we should refocus ourselves away from the many operational, executional things towards finding a solid, bulletproof methodology to Be as good as possible when it comes to measurement there.
It won’t be perfect because the large publishers won’t allow us to But the one who is mastering measurement is gonna be the best marketer around. That’s what I truly believe in
Frederick Vallaeys: Right. I think that’s a really good point. So don’t don’t wait for perfection to at least put something in place. And I think ultimately all these measurement systems are just trying to do the best possible job of correlating what happens in the real world to what you see in your measurement systems.
And so there is no right or wrong way to measure. They’re just ways to incrementally get better at it. And use that to drive more business value. Good. Well, both of you, thank you very much for joining us. I’d love to see any blog posts that you write on this topic. I know for people listening and viewing, these were some complex things and it might be easier if we put some visuals around it.
But do reach out to Andreas Reiffen at crealytics. com and Frederic at profitmetrics. com if you want to follow up. And We’ll take next week off, but then we’ll be back the week after with another PPC Town Hall. So, thank you all for watching. Thanks for having us. Bye. Bye.