The Unofficial Shopify Podcast

The Facebook ROAS Death Spiral

Episode Summary

Scale profitably again post-iOS15

Episode Notes

This episode is also available on YouTube:

If you advertise on Facebook (or other ad platforms) you need to understand Media Efficiency Ratio (MER) vs. Return on Ad Spend (ROAS) and how to think about advertising incrementally.

Our guest today is William Harris, founder of Elumynt, a growth agency for Shopify stores. In the last 9 months, they've helped six clients get acquired, with the largest selling for a whopping $800M.

Show Links


Never miss an episode

Help the show

What's Kurt up to?

Episode Transcription

The Unofficial Shopify Podcast
William Harris

Kurt Elster: Today, on The Unofficial Shopify Podcast, we’re talking about how you’re probably doing your Facebook ads wrong, or at least thinking about them wrong. Certainly, you’ve heard of the issues with return on ad spend lately, and the new method for looking at it, media efficiency ratio, and to talk through some of that, I had breakfast with a gentleman, William Harris, who’s the founder at Elumynt, at growth agency for Shopify stores, who’s helped six of his client stores get acquired, the largest for $800 million, all just in the last nine months. So, certainly he has some experience here, knows what he’s talking about, and convinced me that there were at least three ways I had wildly screwed up my approach to Facebook ads, and so he’s going to join us today and talk us through that. I’m looking forward to hearing how I am painfully, sorely wrong.

I’m Kurt Elster, A.K.A.:

Ezra Firestone Sound Board Clip: Tech Nasty!

Kurt Elster: And I am joined by Mr. William Harris on the phone. William, how you doing?

William Harris: Good. How about you, Kurt?

Kurt Elster: Good. Well, we’re doing this over Bluetooth and a phone. Let’s test our levels here. Check, check. Check. No, it’s brutally bad. We gotta figure out something else here.

William Harris: Okay. I think I know how to fix it. Hang on. Is this better?

Sound Board Clip:

Kurt Elster: What? Oh my gosh. The man was right outside in my hallway. This was quite the elaborate bit we set up here. Why? What was the point of that elaborate bit? What point were we trying to make there?

William Harris: Yeah, I guess the point is just that there’s something more enjoyable about the real, the tangible, the here and now. I think we’ve all had enough Zoom calls lately and kind of wanted to just be in person. There is something, though, about how that relates to Facebook ads, and I think that’s what I’m kind of calling the ROAS death spiral right now.

Kurt Elster: ROAS death spiral. Oh my gosh.

William Harris: Yeah. So, okay, the ROAS death spiral, I look at it as this idea of ROAS is fake. It is made up. It is monopoly money. It’s a metric that we’re looking at that isn’t actually even a real thing. It’s all based on attribution, right? So, the ROAS that you have depends on, well, are you looking at 28-day click? Which we can’t do anymore, but it was a thing. 7-day click, one-day view, there’s a lot of different ways that you can look at this. Are you looking at first click, last click? There’s a lot of different things that get into making this into a metric, and that’s even then based on if it actually can track what’s happening or not.

Kurt Elster: Let’s back up.

William Harris: Okay.

Kurt Elster: We’re talking about Facebook ads, here?

William Harris: Facebook ads, but this goes beyond Facebook ads.

Kurt Elster: And we’re talking about if I’m spending multiple thousands of dollars on advertising with Facebook.

William Harris: Right.

Kurt Elster: I need a way to determine if that is successful.

William Harris: Right.

Kurt Elster: And there really was only one metric everyone used for all time.

William Harris: It was the gold standard for a while.

Kurt Elster: Yeah, it was it. Just return on ad spend.

William Harris: Sure.

Kurt Elster: But there was, and this return on ad spend worked by when someone made a purchase, it looked at… It recorded that purchase data, and it knew, all right, did they look at… Did they see an ad in the last seven days? Did they click an ad in the last 28 days was typical, but you could mess with this? You could change your definition. So, not all ROAS was made alike. And then the larger issue we had with ROAS as a metric was who takes credit? Because Klaviyo took credit for it, full credit for the purchase. Google took full credit. Facebook took full credit. Who gets credit?

And of course, the metaphor I loved that I heard not long ago was it’s not the last beer that gets you drunk, it’s all the beers. But all right, so what I’m hearing here is with the iOS 15 update, we lost ROAS as a reliable metric, but it’s starting to sound like you weren’t actually a big fan of ROAS.

William Harris: I wasn’t a big fan of it even from the beginning, and part of the reason is because I think it allowed people to invest in some of the wrong techniques that weren’t necessarily as incremental to the business. And one of the ways that I try to explain that is let’s imagine that you own a coffee shop, and I said, “Hey, Kurt, I’m gonna help you get more customers for your coffee shop. I just want you to pay me $5 for every customer I send.” And you say, “Okay, great. That sounds like a good idea.” And then I go to the people that are sitting in line at your coffee store, right? At your coffee shop, and say, “Hey, Billy, you should buy this coffee. It’s really great.” And he’s like, “Yeah, I’m in line. I’m gonna buy it. This is good.” And at the end of the day, maybe you were gonna have 10 customers, you were gonna make $10 a customer, so you were gonna have $100. But now you have to pay me $5 for those customers that you already had in line anyways, and so you’ll actually only make $50 off of that, so you lost money. You didn’t gain any incremental revenue from this.

There’s a big difference between that and if I actually walked outside of your coffee shop and I just grabbed somebody off the street and said, “You gotta come in here. This is the world’s best coffee,” and they say, “Great, sounds good. I’m gonna come in.” And I grabbed five new people who come in. But in Facebook, or in Google, or Snapchat, or TikTok, or any other ads platform, when you do that, the ROAS of those customers, of the net new, truly net new people, is gonna be a lot lower than say retargeting would be, or even then say targeting your existing customers. And so, by focusing on ROAS, a lot of times it allowed us to focus on just getting the same customers we already had versus going after net new people.

Kurt Elster: Oh, okay. So, it was skewing toward… Essentially, the system caused us to look at and skew toward where we could see the attribution. Everyone wanted to make data-driven decisions. Facebook and other platforms wanted to justify your ad spend, rightfully so, and this seemed… We really didn’t question this too much other than like the multiple beer scenario, where like who gets credit, how do we look at that.

But beyond that, we didn’t think that hard about it. We really didn’t.

William Harris: Because it feels good. It’s easy to get addicted to it like any other performance metric that we have, social media, if you get addicted to likes, and followers, or whatever, right? It feels-

Kurt Elster: Well, what’s the ultimate metric to get addicted to? Return on investment.

William Harris: Right.

Kurt Elster: Everyone is… Money is always seeking alpha, right? We want that ROI. And that’s what ROAS was a measurement of. And an addictive one. It’s like, “I’m getting…” I mean, it’s a metric for the sky’s the limit on my business. “Oh, I’m getting an 8X ROAS. I’m getting a 10X ROAS.” Keep cranking money into it and money comes out the other end. Oh, my gosh. I put in a dollar; it gave me eight. I like this vending machine.

William Harris: And that’s where I think that breaks down. You put a dollar in, and it said it gave you eight. Just like the coffee shop, you paid me $5, and I said I gave you $10, but you were already gonna get that $10. I didn’t actually give you $10. I actually-

Kurt Elster: And I didn’t necessarily put in just $1.

William Harris: Right.

Kurt Elster: All right.

William Harris: So, this is where MER differentiates itself. MER is looking at two facts. The fact is how much money did you spend. That’s a fact. That’s not based on attribution. There’s no windows to consider. You can look at your credit card statement and decide, “I actually spent this much money.” Revenue, the actual amount of money that came into this, that people paid you, that’s also a fact. That’s not based on attribution. You can look in the backend of your Shopify and see, “This is how much money we actually brought in.”

Taking those two facts, revenue, total revenue, divided by total ad spend, gives you MER. And what’s nice about this is it eliminates that duplicate attribution thing that you were talking about before. Klaviyo took credit. Google took credit. Facebook took credit. It was one sale, you got three platforms taking credit, saying that it’s three sales. You still don’t know which one you need to invest in again. And to your point about all the beers, they all had a role in playing in it, so you can’t just say, “Well, I’m only gonna give the credit to Google because they were the last click.” That’s also not effective, because that last click would not have happened if the first two things didn’t happen, right?

So, what’s nice about MER, though, is it helps you to break down some of those issues to the point where if you say, “Great, you spent $10,000 and you made $100,000. You’ve got an MER of 10.” Right? $100,000 divided by $10,000. If you spend $20,000, do you get $200,000? Do you still keep that MER? Does that still happen? Versus if you invest in something that’s less incremental, it might say that you got a high ROAS, let’s say retargeting. Maybe you spent another $10,000, but it was retargeting, and you only made $10,000 more. So, you went from $100,000 to $110,000. That’s not the same thing as going from $100,000 to $200,000 when you increase from $10,000 to $20,000 in ad spend.

Does that make sense?

Kurt Elster: Yes. Well, and I think the other issue with ROAS is it is, as a metric, is too laser focused, right? We’re not looking at the whole sales funnel. We’re not looking at everything that happened to contribute to the sale. Really, it’s a matter of how many… Marketing essentially is unchanged now as 100 years ago. If I get more impressions, I am more top of mind and more trustworthy, and a purchase is more likely to happen. It’s not a particularly complex scenario to work out. And so, I like media efficiency ratio from that standpoint that, “Okay, we’re taking a holistic view of this marketing funnel.”

I like that. What I also like, my inner bookkeeper likes it. Because ROAS, again, it’s limited in the impressions and attributions. It’s too focused. But it’s also doing the same thing with revenue, right? As an example, you take someone like our client, Hoonigan, who both advertises on PPC networks, but also is publishing organic content daily, multiple times a day, across multiple channels, and especially like YouTube. So, they’ve got combined across all channels, it’s like 15 million subscribers. That’s not getting reflected in ROAS, but there is no universe in which that did not help assist or inflate that ROAS number. Whereas media efficiency ratio, if we say, “Okay, we’ve got this content marketing effort, and we’re paying a video editor, and we’re paying a camera guy,” I can include that in my expense portion of the media efficiency ratio, and I know the total revenue that came through, like through an online store, or through Amazon. That’s another nice part about it, is maybe I’m selling on my online store plus marketplaces, and potentially the organic content influenced those purchases, as well, but they weren’t set up for attribution.

William Harris: Okay-

Kurt Elster: You’ve sold me on MER isn’t just like the next best thing, it’s the best thing.

William Harris: Well, and I’m gonna take you back where you said, “Okay, these organic things happened. This PR drop happens or whatever,” right? And saying okay, maybe that wasn’t reflected in ROAS, and I would actually argue it was. It inflated your ROAS, which I think you alluded to, and so you have this big PR drop that happens. All of a sudden, your ROAS on Facebook jumps from 3X to a 6X, and what does your Facebook agency say? We gotta spend more money, you’re killing it! Right?

Kurt Elster: Oh.

William Harris: But is that really what you need to do? Did you get that incremental revenue from the ad spend or was that from something different?

Kurt Elster: It’s hard to know. It’s not like… The agency is not being malicious here.

William Harris: No, not at all.

Kurt Elster: They’re looking at… They’re doing what we were all trained to do. You see that ROAS and you’re like, “Look, you’re getting a return on investment. Why would you not want to put more into the machine?”

William Harris: I call it advertising myopia, right? I’m nearsighted. I take my contacts out. And I think I can see maybe about-

Kurt Elster: Can you see me? I’m over here.

William Harris: No.

Kurt Elster: William! Here!

William Harris: Right now I can. But we’re looking at this very tunnel vision in that, where your Facebook team is looking in Facebook platform, looking at the data that they have, and they’re making data-driven decisions around the data that they have. We need to take back a look of, “But what does this actually mean to the business?” And so, we like to see what is that return on the spend that we did? It’s a good thing. We just have to realize that what Facebook is reporting is not return on ad spend. It is reporting a reported return on ad spend. It’s based on whether or not it was able to take credit for that sale. And as we know, iOS 14.5, it can’t take credit for a fraction of what it used to take credit for, so it’s a made-up statistic now. It’s based on a fraction of the data.

Kurt Elster: And potentially, it was a semi-made-up statistic before.

William Harris: Sure.

Kurt Elster: The thing we miss now, though, is attribution. That’s what we’re missing. And so, I think the… We’re talking about the joy of media efficiency ratio as it relates to a replacement for return on ad spend. But I think media efficiency ratio is saying we’re looking at the entire funnel, everything, but I don’t know within those multiple channels, okay, which is the strong one? Which is the weak one? Where can I do better? Where can I allocate more resources? And I think that’s maybe the problem, the limitation of MER versus having individual channels report these things.

I’d love to hear your answer to this, but the solution that we’ve been using is you use MER, you use ROAS where you can. It’s still there. It’s not gone. But all right, let’s take it with a grain of salt and also be looking at MER, but then there’s one more piece we need to add, and it’s ask the customer how the heck they think they heard about us. I think that’s an interesting one. Ask them. And so, post-purchase attribution survey I think is the way to go.

William Harris: Absolutely. It’s another data point that I think is important. I think about it like an x-ray. In order for them to diagnose where the break is, they’re not gonna take a picture of just one image. They’re gonna take a picture of the anterior, posterior. They’re gonna take a picture of the lateral view. They’re gonna take a couple so they can see almost like a three-dimensional approach to what’s going on. And I’d say post-purchase customer surveys are a one-dimensional view that you need to use. ROAS is a one-dimensional view that you need to use, because it’s still directionally important, right?

If your ROAS goes up in platform, or down, or whatever, it still likely means something relative to itself. MER is another one-dimensional look that you need to look at. When you start combining all three of them, you can start to get a better three-dimensional look at what’s going on with your business. Not one versus another, but all of them together.

Kurt Elster: You say, “Hey, you can look at each one, but you need to look at all three to get a holistic view of what’s going on.”

William Harris: Right.

Kurt Elster: So, like walk me through a working example where we can kind of hear like how you approach it maybe in terms of mindset or framework.

William Harris: Yep. Okay. So, if we’re gonna go after somebody, I want to go after what’s going to drive the most net new revenue for your business or the overall most profitable revenue for your business, right? So, we’re looking at that incremental revenue. And I’m gonna look at this and I’m gonna say, “Okay, ROAS aside, if I spend more money, I want to know that you made more money.” So, that’s where MER is gonna be my first number that I’m concerned with, right? I want to look at the fact that if I spend your money, do you make more money? Versus the coffee shop illustration where you spent money to me, and you didn’t make money.

So then, there are ways that we could do this. One simple way is you literally just spend more money, and you see if your bank account goes up, right? Does your revenue go up with that? That’s the most basic, simple way that you can do it. Now, if you’re spending a lot of money on other channels… We’ve got clients that are spending $2 million a month on infomercials, a million a month on influencers or whatever, there’s a lot of other things that are going on here, so you wouldn’t know if you just spent more money on Facebook ads if that was the reason why it went up. So, then we look at setting up let’s say a geographic holdout test. Find a pocket of zip codes, or states, or something that you expect to move in a similar fashion to the rest of your audience, to the rest of your buyers, and see what happens when you make changes to that. Does that go up more than the control group? If it does, you now know that what you’re doing here is having an influence outside of anything else that you’re doing nationally to the rest of the stuff, so infomercials, influencers, et cetera.

So, that’s the first step that we look at, right? So, we’re gonna say, “Okay, great. You spent more money. Did you make more money?” And if you go to that blog post I was telling you about, you’ll see there’s a case study that we have there where we show when we do this, we’re charting Facebook spend against revenue, and you’ll see week one, it goes up. Week two, it goes up. Week three, week four, week five, week six, week seven. When you do this now week, after week, after week, after week, you keep spending more, and you keep making more money, and that’s the only change that you’re making, it goes beyond correlation to, “We could probably guess that that’s causative, right?”

Kurt Elster: There’s a statistician somewhere who just passed out, but yes, I’ll go with it.

William Harris: Right. Sure. You’re gonna get to a point, though, where you say, “You know, 12 weeks in a row of I pushed this button and this is what happens, the likelihood that pushing this button causes this is getting a lot higher now.”

Kurt Elster: Yeah.

William Harris: So, that’s the MER part, and that’s where that’s the first part that we have to look at. Now, let’s say that we’ve at least adopted that, and we look at, “Okay, let’s say that incrementally for every $1,000 that we spend, we make another $5,000 of actual money in our bank account.” It’s a 5X MER, right? So, $5,000 divided by $1,000, 5. So, we’re seeing a 5X incremental revenue on every $1,000 we spend. We look in the platform on Facebook, let’s say, and it says 2X ROAS because it’s not getting credit for a whole lot of what’s happening now post-iOS 14.5. We can at least say that there’s a correlation between a 5X MER and a 2X ROAS.

Now, we don’t know if they’re gonna move logarithmically or linearly when they move, but we at least could assume that if my ROAS and platform goes from a 2 to 2.5, there’s a strong chance that I’ve improved the actual revenue coming into the business, as well.

Kurt Elster: So, are you charting this in a line graph over time, like we got MER each week, versus ROAS each week?

William Harris: Yeah. Reported. And I call it reported ROAS. I don’t even like calling it ROAS anymore.

Kurt Elster: Oh, wow. We’re really going hard against ROAS here.

William Harris: Well, because it’s not actually the return on ad spend, so why call it return on ad spend? It’s a reported return on ad spend, right?

Kurt Elster: Yeah. You’re right.

William Harris: Okay, so I look at it that way. The next piece that we’re looking at, though, is let’s look at Google Analytics. Let’s see if this was incremental. And in Google Analytics, a lot of times it’s gonna take last click attribution, but there’s a report in there where you can look at your multipath report, okay? Well, let’s look at the multipath report and let’s look and filter down by Facebook. Do I see incremental revenue there?

Now, again, it won’t take credit for all of the revenue that Facebook drove. But if it went up 100%, and my spend went up 50%, I’m getting more attributed revenue based on click attribution as well here. That’s another point of view now that I’m starting to form this three-dimensional picture that’s saying it’s not just Facebook taking credit for it. It’s not just even my revenue going up independent of itself. There is an actual correlation here that’s taking place. So, I’m starting to see this three-dimensional view. Does that make sense?

Kurt Elster: Yes. Well, and with customer reported attribution, where does that fit in the picture? Is that just another layer to confirm what we’re seeing in our Google Analytics?

William Harris: Yeah. And define the customer reported attribution that you’re looking at.

Kurt Elster: So, one of the things we’ve seen people been doing post-iOS 14 is on the order confirmation page, the thank you page, we fire-

William Harris: Oh, sure.

Kurt Elster: Yeah, you fire a popup that’s like, “Hey, how’d you hear about us, buddy?” And then you randomize, like in there is every one of the possible acquisition channels and then you randomize it so that it’s never in the same order, and that way you don’t skew towards whatever is at the top.

William Harris: Yes. You should see a correlation there, as well, right? So, if you’ve increased your spend by 50%, your revenue goes up, and you look at the actual revenue that you have there. Not the reported revenue, the actual revenue goes up. You look in your multipaths report and that also goes up.

Kurt Elster: So, we have in our more traditional way, we say, “All right, we think our reported ROAS… Maybe it’s right. Let’s spend more money on our Facebook ads.” And then we should see revenue go up and then we should be able to go into Google Analytics, check this multistep funnel, see revenue go up there, as well, and then in theory, in this post-purchase survey, more people should be saying, “I heard about you through Facebook.”

William Harris: Exactly.

Kurt Elster: And then, okay, so now we have triple confirmation. If we see that, now we go, “Aha, the reported ROAS may not be exact, but it gives us a trend line.” Good, bad, up, down.

William Harris: Correct.

Kurt Elster: And we’ve confirmed that it is trending up, it was not making this up.

William Harris: Right.

Kurt Elster: I finally wrapped my head around it.

William Harris: And the one nuance that I’ll make to that is that the reported ROAS, if it’s optimized to the same thing, and it goes up or down within there, so within that campaign if that changes, then it is relative to itself and it’s worth looking at. But if you… Let’s say that your total ROAS in Facebook goes up but you flipped the spend from 90% net new prospecting, 10% retargeting, and you flip it to 10% prospecting net new people and 90% retargeting, your reported ROAS will likely go up because it’s gonna get the opportunity to take credit for more sales, but it didn’t necessarily have the same effect in your bank account. So, that’s where I’d say when you’re looking at it in different campaigns or ad spend mixed between different campaigns, there’s a difference there and that reported ROAS changing might not necessarily mean that it’s better or worse for the business.

But within a single campaign, that should be fairly well correlated. If it goes up, you’re likely doing better. If it goes down, you’re doing worse. And the same thing would be true for ads within that campaign. So, you’ll put a new ad into that campaign and that ad is outperforming the other ads in that campaign, it’s very likely that that ad is actually doing a better job. So, ROAS is still a worthwhile metric to use relative to other things.

Kurt Elster: Okay, so really, it’s we’re saying, “All right, ROAS is still good, but it is not the to-the-penny thing that they claimed it to be and that many people suspected was not the case.” Essentially, it’s like a trend indicator. Good, bad.

William Harris: Exactly.

Kurt Elster: Is it going up or down?

William Harris: Right.

Kurt Elster: Right. So, we want that trend line, and that’s where it’s helpful to graph these things over time and in relation to themselves. And if all… We’ve got now three or four reporting sources in the scenario. If they all agree, okay, now I feel good again. And so, this really does… I feel pretty good here. This really does help us get out of that iOS 14.5 terror nightmare that we all found ourselves in, or at least Facebook marketers and people who relied on Facebook ads did.

William Harris: Well, because that’s what’s happening. And so, this is the reason why, so that article I told you about, it’s 7,000 words, which is ridiculous, but it had to be because I’ve run into this now so many times where businesses are coming to us and their agency is trying to do the right thing, and they’re moving from let’s say a 5X ROAS or something along those lines, right? And they’re improving the ROAS in platform by shifting to retargeting or by spending less money, so you spend less money, your overall return usually is gonna get a little bit better, because you’re not pushing as hard as you would need to. Each incremental dollar you spend typically is gonna be a little bit less efficient as you build and grow out, right?

So, they’re trying to do the right thing and they’re saying, “Okay, we’re gonna pull back this. We’re gonna pull back this. This isn’t doing as well because it’s net new prospecting, so pull back on that, and we’re gonna focus just on these core things that have the highest ROAS,” and it ends up being a lot more just retargeting stuff. And so, but they’re trying to do the right thing by you, and this is what the customer thinks they want.

And what happens is we get into this and we look at their actual… So, we look at P&Ls a lot for our clients, as well, and we optimize around profit, and so when we’re looking at this and we’re saying, “When you spent let’s say $100,000 and your ROAS was 5,” and then we actually are looking at MER at this point in time, but the concept here is let’s just say your efficiency was 5, let’s say you made $1,000 profit, or $10,000 profit, right? So, you’re profitable at that. But you dialed back that spend because you thought a 5 wasn’t good enough for you, so they were dialing back and they were going more towards retargeting. Now, you’re at an 8, so you’re more efficient. The agency’s patting themselves on the back. You’re pretty sure you’re doing good. But your CFO is kind of annoyed with you because you’re losing $20,000 a month, because that money that’s being spent wasn’t going towards driving net new business. It was just kind of taking credit for a lot of stuff that was already happening.

So, you’ve dialed it back, you’re getting a better ROAS.

Kurt Elster: So, we essentially end up playing ourselves.

William Harris: Correct.

Kurt Elster: In service to just that ROAS KPI.

William Harris: Right.

Kurt Elster: And yeah, the way you juice that metric is by shifting the ad dollars away from new prospecting, which is where the algorithm really seemed to struggle, and where we seem to see more impressions and placements happening on Android devices. The algorithm going to where the data is.

William Harris: Sure.

Kurt Elster: But we know historically Android users do not spend as much typically as iPhone users.

William Harris: Right.

Kurt Elster: So, ah, all right, if I’m trying to just juice that ROAS number, well, let’s go to remarketing, because that’s like… I’m preaching to the choir, right?

William Harris: Right.

Kurt Elster: And so, I will see a higher ad spend there, but those people may have bought anyway, right?

William Harris: Right.

Kurt Elster: So, really I’m just skewing the metric and not necessarily doing anything malicious.

William Harris: Nope. Not at all.

Kurt Elster: Just following the system. Now, speaking about that algorithm, our Facebook metrics, you brought up an interesting point in the pre-interview to me. You said, “Hey, what do you set your ad goal to?” Well, why wouldn’t I set it to optimizing for purchase? At which point you laughed uproariously at me at my innocence. Why? Why was I doing that wrong? What should I be doing instead of setting my ad goal to optimize for purchase? Because that sounds really good.

William Harris: It does.

Kurt Elster: I want to optimize for purchase, not like optimize for meet new friends, all right?

William Harris: I’m gonna make up a stat but I’m gonna say 90% of the time we’re gonna optimize for purchase, because it is the better optimization goal. And if you think about the algorithm, the algorithm is going to be looking at 10 million people, trying to decide who is likely to actually make a purchase, and it’s gonna look at data such as, “You know what? Has Kurt ever actually purchased anything online before?” And saying, “Yeah, he has.”

Kurt Elster: Oh, I have. I very much have.

William Harris: Right. And so, it knows that you purchase online. It might know that you’re looking at other things like this. You’re into tech. You’re into whatever, right? And so, it could say, “There’s a good chance that Kurt will make this purchase here within the next seven days,” because that’s all it has the opportunity to take credit for, right? So, it has to be if you’re optimizing for purchase, it needs to make sure you’re really about ready to make that purchase.

Versus, let’s say my mom, who I don’t think has ever bought anything online. It’s not gonna optimize around her in this 10 million people of potential people that are interested or whatever. So, it’s usually going to be the best.

Where that can get tricky, though, is the cost to reach people can sometimes skyrocket when you’re optimizing around purchase, and this is I think where we’re getting to, is around the holidays. So, let’s say Black Friday Cyber Monday, what is every marketer going to optimize their campaigns towards? Purchase.

Kurt Elster: I assume purchase.

William Harris: Right. Facebook’s gonna look at this and say yeah, you all want to optimize to the same thing with this group of people. I don’t know how to optimize that. And so, a lot of times you might not even spend your full budget. Instead, think about it… Again, I like to go back to the logic of this.

Kurt Elster: All right. Well, if you’re in a place where it can’t spend your full budget, you know immediately you have slammed into a supply and demand problem.

William Harris: Sure.

Kurt Elster: As far, like it’s an ad marketplace, right? And so, yeah, if you’re not spending the full budget, there just aren’t enough eyeballs that qualify for your ad dollars.

William Harris: That qualify. And I like that you said that in there, right? Like you had that qualify.

Kurt Elster: Yeah. There’s plenty of people out there, just not the ones that you specifically said you wanted. You’re too picky.

William Harris: And not the ones that the algorithm thinks are going to make that decision, right? Let’s say that Mary Harrington, I’m told that-

Kurt Elster: Mary Harrington, the fake person who I have been receiving mail for for the last decade.

William Harris: I love it. So, let’s say that Mary Harrington sees an ad. Does she know what optimization goal was set when she sees that ad? Does she know if you’ve set that to a reach campaign, a traffic campaign, or a purchase campaign?

Kurt Elster: I believe absolutely not.

William Harris: No.

Kurt Elster: Yeah. What you set your individual ad settings to is totally transparent to me, or totally invisible to me, the end user.

William Harris: Right. What she decides to do after seeing that ad has nothing to do with what you’ve set the goal as, right? And so, if she sees the ad and she’s inclined to click and buy whatever your product is, it doesn’t matter if it was a reach campaign or not. Now, from a reporting perspective, Facebook’s not gonna take credit for likely much of anything out of a reach campaign from a ROAS standpoint, just so we’re clear on that, right? It’s not likely going to get much of that credit.

But if she was inclined to make that purchase, it doesn’t matter what objective you use to set that. So, if the CPMs get way out of hand because everybody’s trying to do the same thing, there’s oftentimes the right thing to do would say, “I’m gonna switch to a different objective.” Same ad, but maybe I’m gonna optimize for add to cart, traffic, reach, something else, to make sure that these people see that. Now, the one thing that I would put in here to make sure that everybody’s at least… If they do try this, they do it in the right way, is you can’t go with an open audience if you’re gonna go with reach. Because there’s a lot of people that it can reach very, very, very cheaply that are not even close to the right person.

So, in this situation-

Kurt Elster: So, it would work against me.

William Harris: It would.

Kurt Elster: Because it’s like, hey, so if you say reach, it’s saying, “All right, you want the maximum number of impressions.” Ignoring everything else. And so, it’s just gonna go straight to these are what we determined are the least sought after? Again, that supply and demand issue, but now working where I have too much supply.

William Harris: Right. But if you narrow that audience down with a lot of really good let’s say inclusions and exclusions to the point where you say, “I feel really good about this audience.”

Kurt Elster: All right, so I’m gonna set it for impressions, just like give me straight reach, but straight reach by itself is not a great idea necessarily, but then if I know through let’s say surveys what my customer is like demographically, and what things they’re interested in potentially, I can layer that over it.

William Harris: Yep.

Kurt Elster: So, I can say like, “You go find me the best audience for this within these confines.”

William Harris: Right.

Kurt Elster: So, I’m gonna say like I know that my audience is typically female, 24 to 38, lives in a major city, like as much info as I know about them, and that’s what’s so cool about the demographic and interest-based targeting in Facebook. That’s so much more fun than letting the algorithm do it. But if you know confidently, you can layer that on there, and maybe that works well. I like this idea. We’re combining the algorithm and what we know that it doesn’t.

William Harris: Somebody is going to listen to this, though, and they’re gonna say, “Yeah, but you can’t get reporting off of that demographic information anymore, right?” And that’s true. So, you can’t get the reporting on who purchased that based on age and other demographic information.

Kurt Elster: I was so disappointed when that audience insights tool fell apart on us.

William Harris: Right. I have my theories on why. Part of me thinks that Facebook might still know, but because there’s so much calculation that needs to happen, it would just be cost prohibitive for them to even try to do that now, because a lot of… So, a lot of what they’re doing, and if you look in your ROAS right now too, you’ll see this. There’s a little eye or whatever that you can click on that says that this is also somewhat modeled data now too, right? So, they’re modeling data.

Kurt Elster: Modeled being a euphemism for, “We sort of fudged this.”

William Harris: Sure.

Kurt Elster: Like if I told the IRS, “Look, I modeled my taxes,” they would not be pleased with me.

William Harris: No, I don’t think so. So, the idea there is that you can’t still get that, but that doesn’t mean that you can’t look at this and say, “I know that these are the right people. I’m gonna go after them and I’m just gonna make sure that my cost to reach them, my CPM is gonna be a third maybe of what it was to go after them from a perspective of purchase.” This doesn’t work all the time. Like I said, it really depends on the big difference, like what is that spread between the reach objective and the CPM on the reach versus the purchase at different times based on, like you said, supply and demand on eyeballs.

Kurt Elster: Okay. I like this idea. I do. Can I run this year-round? Should I just be doing this? Or is this limited to Q4 for the holidays?

William Harris: So, you brought up earlier something about they need to know who you are before they can make a purchase. It was something to that effect, right? They can’t buy from you unless they know that you exist.

Kurt Elster: Yes. You gotta be… I can’t just not advertise and expect to make sales. I need to be top of mind.

William Harris: Right. So, if Timmy and Mary Harrington don’t know that you exist, let’s say Timmy knows that you do and Mary doesn’t, the chances of Mary making a purchase from you without knowing you exist are zero. Timmy has a chance because he knows you exist. Mary’s got zero chance unless you make her aware of you somehow. So, that’s where you look at that and you say, “Okay, there’s an element of how do you still make sure that people are aware of you?” I don’t like running direct response ads necessarily year-round in a reach campaign, but I do like running other types of ads to get people excited.

And I talk about this as the idea of like the romance, right?

Kurt Elster: It is a relationship. It’s a perfect analogy because I think it’s quite directly applicable.

William Harris: When you’re trying to romance somebody, let’s just call it direct response is the idea of like if my wife says, “Honey, can you take out the trash?” It’s not the most romantic thing. It needs to get done. If all you’re doing now, let’s say direct response is like what needs to happen, it’s like here’s this thing, buy it. It’s not romantic, right?

Kurt Elster: Buy this!

William Harris: How do you romance these potential people at a very, very, very cost-effective tactic, like reach? So, this could be video views campaigns, or something else, but some way that you can say how do you get them to align with your company goals, your missions, your things that just are exciting? Engagement posts even, right? It’s a funny joke that they would read, and they say, “Oh, you get me.”

Kurt Elster: Ah. Yeah. Yes. Yeah, there’s a lot more to it than just an ad that says, “Shop now.” Right? You can’t just go straight to that. Just as on date one, it’s not, “Hey, will you marry me?” It’s, “Can I get your phone number?” And then on down the line. And you work your way up to it. So, the path a relationship follows is very much a sales funnel that potentially ends in cohabitation.

William Harris: Right. And optimizing for purchase, there are enough people where if you walked up to them and you said, “Hey, do you want to get married?” There’s enough people if you were to go to 10 million people, there’s probably enough people that would say, “Yeah, let’s do it right now,” right?

Kurt Elster: You’re right. If I asked 10 million people, I would find… We’d find at least a few.

William Harris: And that’s optimizing for purchase. You’ll find the people that are ready to buy this right now. They’ve been thinking about it. Maybe they’ve heard your podcast and they’re like, “He’s got a sexy voice,” right? Julie would be a little disappointed.

Kurt Elster: It’s so high!

William Harris: But yeah, okay, so the idea here, though, is that that’s what that like. Reach, though, is this idea where they don’t know who you are. They don’t know you have a really cool Shopify podcast. They don’t know that you’ve got Star Wars stuff all throughout your office, right?

Kurt Elster: But they should. That’s very important.

William Harris: They should. That’s your reach campaign.

Kurt Elster: I like it. Okay. So, yeah, optimizing for purchase really… You’re right. It’s myopic, potentially. It’s less effective now. And it really sounds like we’re putting the cart before the horse.

William Harris: Yeah. 90% of the time or more, I’m gonna optimize for purchase. But some of your budget needs to go back to, again, that logic of, “Well, how do they buy from you?” They have to know you exist. So, how do we get in front of them and what should you be getting in front of them with? What is that message? How are you gonna romance them?

Kurt Elster: And I don’t think there’s like a single right answer for that.

William Harris: There is not.

Kurt Elster: That’s where it gets into like, “All right, you better figure out your positioning and your messaging,” which is the thing that’s way harder than it sounds.

William Harris: It’s the thing that I think we come into most in trying to help our clients solve, almost more than anything else, is how do you cut through the noise? David Brier, I don’t know if you know him, but he’s got a book called Brand Intervention, and I even just like that term, right? It’s like this thing that it’s like, “We’ve got a problem. We’ve gotta fix it.” But I remember him saying something to the effect of you can’t say the same thing everybody else is saying. If you say, “We’ve got organic eggs,” everybody has organic eggs. That’s not a differentiator anymore. And a lot of people, if you look at their Shopify stores, and I’m sure you see this, because you are looking at a lot of this from the CRO perspective, you’re saying the same thing as everybody else.

Kurt Elster: Yeah. You have to make bold claims. People will say, “I want a website like Apple’s. I like Apple’s.” And they think what they’re looking at is, “Well, I like it because it’s clean, it’s got a lot of white space, it breathes, it’s well laid out,” and it is.

William Harris: Absolutely.

Kurt Elster: The thing is polished and perfect. But the magic of it, what really makes it sell, is the copywriting. They’ve got these big, pithy headlines that make bold claims, and they’re always set in Helvetica New, of course. Bold, or extra bold, or whatever. But like one that stood out to me was their headline for the Apple Watch when they launched the new series, and it said something like, “It’s the ultimate healthy lifestyle device.” And I think I screwed that up, but that was… It said like, “The ultimate device for a healthy lifestyle.” Something to that effect.

All right, so if we unpack that, they made a claim so bold that it’s like, “Hey, if you don’t buy our watch, you’re gonna die, William.”

William Harris: Right.

Kurt Elster: And I see you’re wearing an Apple Watch and you’re still alive, so clearly it works.

William Harris: Obviously.

Kurt Elster: But it’s because they knew, I’m sure they surveyed customers and said, “Hey, what do you use on your watch?” They have usage statistics. They knew the thing people were interested in were the health features.

William Harris: Right.

Kurt Elster: All right, and they surveyed people, they focused on that, and so now they positioned it as if you buy this watch, you’re gonna live longer. This is a life extension device. Buy one for grandma. All right. Maybe two. One for each wrist.

William Harris: Now we’re getting a little crazy.

Kurt Elster: That one got away from me.

William Harris: Yeah. Well, and I think again, Apple’s, like you said, it’s the copy that makes the difference, and I would even go one step further than that too, right? It’s Apple is the reason why Apple’s website works. Because they already have pent up demand. Because it’s Apple. They’ve already done a good job creating that demand through everything else that they’ve done, so the website itself, they could do almost anything and it’s gonna work, because it’s Apple.

Kurt Elster: Yeah. Well, they launched new MacBook Pros. Did you order one?

William Harris: I did.

Kurt Elster: You did? Had you even looked at it in detail before you… You knew in your heart of hearts you were buying one.

William Harris: Well, yeah. I read plenty of the leaked documents ahead of time, right? But no, not too in depth. I knew that I wanted one. I was ready.

Kurt Elster: I want one. I’ve not ordered yet. But I’m sure I’ll get there.

William Harris: Yeah. I mean, yeah, exactly. They’ve got me. I’m an Apple fanboy, right? Everything Apple. I love it. I’ve got pent up demand for it. They come out with it, I’m like, “Yep. I want it.”

Kurt Elster: Okay. We have unpacked a lot here. This has been… I feel good about this. I like it. I am so glad you were able to come in, to record this in person with me. But there’s one last thing I want to talk about.

William Harris: Okay.

Kurt Elster: Maybe, have you heard about this supply chain shortage thing?

William Harris: No.

Kurt Elster: You haven’t? All right. Well, there’s something like 100 ships piled up on the West Coast who are struggling to unload stuff and then struggling to get that stuff from the port, so now the containers are piling up in the streets.

William Harris: They should put this in the news.

Kurt Elster: Well, and now we hear this every day, like every morning you get updates on the supply chain on the news. That is interesting, because if the national news every single morning is talking about this, now everyone, and the advice, how do you deal with it? The advice is, “Hey, you should buy your Christmas presents early.”

William Harris: Sure.

Kurt Elster: And if you start shopping now, we’re starting to notice certain things are sold out, so you’re probably just gonna be buying what’s available. This year’s hottest subject line, “In stock, ready to ship.”

William Harris: Right.

Kurt Elster: Do I even need to be advertising? Is this all a moot point? Can I just shut the ad account down, we’re done here? If I’ve got the product, we can ship it, and that’s the end of it?

William Harris: It’s a great question. You and I were talking about this a little bit earlier too, the idea that why don’t we… As a Facebook advertiser, generally, going to that myopic view again, you want to do everything you can within the power to optimize for purchases, and go after that, and show the highest return on ad spend you can, and if anything, you’re gonna see that people are buying more, so your ROAS in platform has gone up, so you’re saying, “We need to spend more. This is hot. This is really good. We’re doing a great job for you. Let’s spend more. Let’s spend more.”

But again, that’s looking at the data that you have provided, but it’s not taking a step back and looking at the business data. The business data is gonna say, “You’re gonna sell out in two weeks. Why increase your advertising only to sell out in one week?” You’ve lost margin now on products that you likely weren’t gonna get back in stock soon enough. Instead, you actually should be looking at how do you decrease your ad spend correctly? And so, that is one of the things we are telling some clients.

Haven’t had to do that yet this year, but we did do that a little bit last year when… You know, we like to work with them on their inventory forecasting. How are you doing on inventory? When do you plan on having more coming in? And right now, if we sell at the current rate, how long do you have before you sell out?

And if that is a short period of time, we’re gonna look at starting to decrease your ad spend. You need to make as much profit off of each one of those as you can from a margin perspective. And the flip side of this is limited inventory a long time ago, for a lot of things, charge more for it. Well, a lot of people don’t want to charge more for it, and there’s a lot of good reasons not to. It kind of annoys people. But at the very least, you should make as much profit off of it as possible.

Kurt Elster: Certainly, you shouldn’t be discounting, is what I’m hearing.

William Harris: Right.

Kurt Elster: Which is interesting for Black Friday, that we’re… and not everyone discounts, of course, like luxury brands obviously don’t. So, moving away from discounting this year because of that awareness around supply chain shortages. But also trying to get a handle on, “Okay, here’s how much inventory we have. Maybe we adjust ad spend down to try and stretch it out.” Because I don’t want to sell out of all my bestsellers week one of November and then I’m sitting on my hands for the next six weeks. I want to try and stretch that into ideally like December 10th.

William Harris: Right. Shipping cutoffs, right? You want to be able to get the most out of that as you can. And to your point about, well, should we just shut it off? Usually, no. Unless you have really low inventory. Maybe. Maybe that is the case. But I also think that there… I don’t know if the algorithms would actually go and say that there’s officially this, although I think Google has gone on record and said this. If you shut your account off and then on, and then off and then on, and then off and then on, think about this from an algorithm’s point of view. The algorithm’s goal is to find let’s say good buyers, but also good advertisers. And the algorithm’s gonna look at this and say-

Kurt Elster: Oh, good point.

William Harris: … “You’re unstable, and so, the likelihood of…”

Kurt Elster: The algorithm is going, “Those are red flags. You’re a bad client.”

William Harris: Correct. The algorithm doesn’t know if it can rely on that ad spend because Facebook’s job is to make money. And it does that by helping you make money, but its job is to make money, right? And so, if it’s not able to rely on the money that you’re likely going to be sending to it because you’re up and down, there has to be something built into the algorithm to favor stability.

Kurt Elster: You know, I’d never thought of that. That’s a good point.

William Harris: So, I would never recommend shutting off, but I do think that there is reason to dial it back a little bit if you’re running into inventory issues. And like you said, just really being aware of where you’re at with that inventory forecasting, knowing that even if it says it’s gonna get here in two weeks, do you know that it’s gonna get here in two weeks? Because if you don’t know that, it may be worthwhile to pull back just a little bit as long as you’re still selling through what you need.

Kurt Elster: Yeah. We have several clients now who are like, “Yeah, we ordered early. We got our stuff into the port. We thought we were good. Now, it’s stuck in customs.”

William Harris: Right.

Kurt Elster: We had several people where it’s like, “Oh, we don’t have anything stuck in transit. It’s just sitting in customs, and we don’t know why.”

William Harris: So, we’ve been talking to our clients about this. This is something we’re doing year-round, always talking to them about where their inventory is at and how are they feeling as far as inventory projections. So, a lot of them have purchased enough early on that I don’t feel too concerned with most of our clients in respect to that. But there are a couple that they’re hot. We’ve got one client, I think they’re just about to go on… Well, they just filmed for Kelly Clarkson’s show. They’re gonna be on Ellen here I think pretty soon too. And when these drop, they’re gonna sell a lot, right? So, we are at a point right now, we’re not scaling up for them right now. Even though we have ample opportunity to, we’re just looking at their inventory saying, “We don’t know now if there’s enough inventory for them to satisfy demand all the way through into December.” So, why should they spend money to acquire customers if they’re going to get them anyways?

Kurt Elster: You know, at the same time, good problem to have.

William Harris: Great problem.

Kurt Elster: Let’s leave it there. William Harris, where can people go to learn more about you?

William Harris: Yeah. is our website. As far as learning more about me, I’m fairly active on Twitter and LinkedIn. Twitter is @wmharris101.

Kurt Elster: This guy doesn’t know his own Twitter handle.

William Harris: Yeah. I think my closing remark would just be that ROAS is… It’s just pretend. It’s a made-up statistic, so focus on the real. So, all these things, focus on the real, like real cake versus not real cake. Podcast in person versus a made-up phone call.

Kurt Elster: A Zoom call. Yeah. All right. I love it. William, thank you so much.

William Harris: Thank you, Kurt.