Andrew Foxwell: What's up, everybody? Welcome to another episode of the Ecommerce Influence podcast. My name is Andrew Foxwell. This is a little bonus episode for you featuring Molly Pittman and Depesh Mandalia talking about campaign budget optimization or CBO. If you're running Facebook or Instagram ads at this point and you are wondering what is happening with CBO, or how it's working, or why it's working the way that it is, or some tips to improve your campaigns, you definitely want to listen to this bonus episode.
I, also, in case you're curious, have a course on campaign budget optimization that you can find at foxwelldigital.com/courses and it's called CBO: All You Need to Know. I have had really good feedback on that one as well from lot of satisfied customers there. So hopefully you find this episode really helpful and full of a lot of tactical and in the weeds information for you and we kind of get nerdy right out of the gates. If you have any questions, you can email me email@example.com and I look forward to hearing from you. Thank you.
All right, well, Molly, thank you so much for joining me to talk about CBO.
Molly Pittman: Hey, Andrew, thanks for having me. Hello, everybody, that's listening. Happy to be here.
Andrew Foxwell: Let's go ahead and dive right in. Obviously, CBO is coming and everybody's thinking about it at this point in time. I think there's thematic things about CBO that a lot of people have started to see. There's different ways that you can start to transition to CBO. How are you thinking about some of these big things that you've seen and example, more campaigns, bigger campaigns? How were you thinking about some of those transitions and what are the things that you've seen started to really work as it relates to CBO?
Molly Pittman: Yeah, I think first off, CBO, it's the thing that most Facebook advertisers are panicking about right now. It's probably the most frequently asked question that I get from students and it's all about CBO. I think the first thing to understand about CBO... And I think this is mainly because of the way Facebook has rolled CBO out and said, "It's going to become mandatory." There is this panic around it. But the issue that I see is how we're using it. Before we get into the tactical stuff, I think there's just an overall misunderstanding here in terms of where and how I see CBO working and when I see it not working. I think the first thing is not comparing CBO campaigns to ABO campaigns.
Most media buyers that I see they're just taking a campaign that's worked with budget at the ad set level, duplicating it, changing it to CBO, and saying, "CBO doesn't work," or just building a random CBO campaign without realizing that this is functionally operating in a different way than when we had budgets at the ad set level, so our strategies need to change which... I'll tell you guys my best practices, but when I think about CBO the word like amplification comes to my mind like a rocket ship. It's one of the most powerful things that we've been given by Facebook since I've been advertising on the platform the past eight or so years.
But people are using it in the wrong way. And in my opinion, CBO is best used when you want a high volume of results very quickly and what you are doing. The assets that you are using are proven. You're using ad set that have social proof, whatever you're optimizing for. You've been able to generate those purchases. That pixel has a good amount of data. You've targeted these audiences or similar audiences before. That's when I see the power of CBO is using it for scale. I'm seeing a lot of people using it when they're more in the test phase and this is because Facebook saying, "Hey, all campaigns are moving over to CBO."
So I understand why people are thinking this way and I think it is a mistake on Facebook's part because CBO is this rocket ship, it is this amplification factor, and some of the most successful campaigns I've run on the platform have been using CBO, but it's always when I really know what I'm doing and when I'm ready for scale or graduation or whatever we call it as educators. When I'm ready for that scale is when CBO works best. But when I'm testing something new and I'm using assets that aren't proven, you're just amplifying crap or amplifying, maybe not crap, but something that you just don't know if it's going to work and because there is that speed factor, you will see worse results out of the gate if you are using it for testing.
I don't know if this is exactly making sense, but I guess the point that I'm trying to make is the biggest issue I'm seeing with CBO isn't functionally. CBO... It's just people think that it needs to be used in the same way as we set up campaigns with budget at the ad set level which completely isn't true, and because of the way Facebook's positioning this and really forcing us to shift, people think that CBO applies across the board.
For me, I'm thinking, "Okay, so what happens when we do shift to, 'It's mandatory for every campaign?" And my solution is basically, like you said, we're going to have to have so many more campaigns in our accounts because I'm just going to have to set up a campaign for each ad set that I would have had in my test campaign before, so that I can have that same optimization experience and truly get a good test.
I can go deeper but those are my initial thoughts and I just always like to lead with that because I think that's a mindset shift for a lot of people, no matter your experience level with CBO. It's just if you think about it more in that way, it's not, "Does CBO work or not?" It's, "Am I using it correctly and am I using it in the right scenarios?" Which doesn't really match up with what we're hearing from Facebook.
Andrew Foxwell: A number of things to say on that. I think one is really it's a budgetary thing. That's a question we get asked a lot in terms of amplification. I agree it's a scaling technique. The thing that we've set around budget is, and we say this in our course which is at foxwelldigital.com/courses... But we talk about basically that the budget of... If you have four ad sets or four different audiences under a CBO, they need to at least be a... If you're not setting a minimum or a maximum, they need to at least be the CPA in which your CPA goal is. So if the CPA goals is 100 bucks and the CBO budget needs to be $400 and as you said-
Molly Pittman: Yeah, a much larger budget.
Andrew Foxwell: Yeah, that gets pretty serious pretty fast. And then when you start to consider... I've had a hard time explaining this sometimes to people where there's a hard line between trying to exit the learning phase, which is always something we're trying to get to and then people getting stuck into Learning Limited, etc., where to some degree, I'm not worried as much about exiting the learning phase, unless I'm really ready to amplify that thing. If it's stuck in Learning Limited and it's doing okay, that's just saying, "Hey, it's Learning Limited. We weren't able to exit-
Molly Pittman: We could use more data.
Andrew Foxwell: We could use more data." Right. To me, that's big. I think that the social proof piece is big. That's something that I completely agree with. I think that that has a lot of implications as it relates to CBO. The thing that I would say is... Well, we should talk about mid to lower funnel, but on the prospect there's really two different things, the prospecting side of CBO. It's really a matter of you have to horizontally scale with it like you would have normally. Let's just speak specifically in the United States, 10% lookalikes, you have a cluster of those that's running in a CBO broken out by ad set. You have a 4% or a 6% grouping broken out, you have an interest groupings broken out within the CBO, and I've found just by transitioning into a CBO with some of those things graduating and to some degree it does work better, especially if you have more data.
But as it relates to the structure of how you... I agree that you have to be ready and that it does require more. How do you think about budget and reaction to what I said and then minimums and maximums? I've heard rules like, "You should set your minimum at half your CPA." How are you dealing with that?
Molly Pittman: I totally agree and the reason that I don't believe... What I mean by proof of concept is even more than social proof. It's whether the end goal is lead generation because you're selling info or services or whether you're going direct for the sale and you're more selling physical products. You really need proof of concept that that activity is occurring via a certain set of ads and certain targeting groups, whether they're lookalike or interests or open targeting, whatever. That really needs to be proven in my opinion before we switched to CBO because the optimization is happening at the campaign level as you guys all know, the amount of reach and what's happening at the ad set level is just so different.
For example, when I'm setting up a test campaign, I have budgets at the ad set level. I usually have 10 plus ad sets because I want to test a bunch of different types of audiences, maybe different variations of my avatar, my audience sizes are smaller like half a million to 3 million if I'm not location dependent or a local business. I'm able to test at more of a micro level in these ad sets silos and I have more ads when I have budget at the ad set level, maybe more like six to 10 ads, depending on how many copy and create a variation. This is just the way that I do it, but that allows me to really figure out what's working and hopefully get proof of concept.
It's not that maybe the budgeting even changes that much. Let's say my CPA goal is $50, then even with budgets at the ad set level I try to set each ad set for at least double that budget a day because I want to give it enough juice to actually test, but that allows me to achieve proof of concept or at least the learnings that I need versus CBO. It's a completely different process. A few of those ad sets are going to get most of the reach. Facebook's going to choose one to three of those ads in a much quicker way, but we don't even know if this is going to work yet. We're not ready for that.
I think that Facebook's split testing feature was their answer to, "Here's how you're still going to test even with CBO." But I find that to be lousy and I've told them that and I don't use that. I think they know that CBO just isn't the solution to testing because with CBO what works for me, and I know you too, is I would still probably take the same amount of budget.
Let's say I had 10 ad sets at $100 a piece back to that example that I was using. Okay, that's $1,000. That's a really nice CBO daily budget to start out with. If you've got something that's proven, but I'm squishing, like I call it, and I'm using way less ad sets, no more than five ad sets, because I find that Facebook really can't digest that.
I honestly don't use a lot of minimum and maximum budgets. The reason for that is that I usually find it can allow me to control my CPA and control that reach in a better way of which ad set is getting more reach. But it seems to throw off the volume of results that I'm able to achieve, which for me, when I'm in that moment and I'm trying to generate a bunch of leads or a high volume of sales, I want that volume thresholds to be really high.
So I try to make it so I don't even need to do that and I can just set up this really simple campaign that has no more than five ad sets, but I've essentially taken all of the audiences that worked in the test campaign, squish them together. Now, I'm looking at more like five to 30 million people in each ad set, maybe even more, much more than the half a million to 3 million when I needed to give Facebook a smaller pool.
Then I usually bring over the top two or three ads from those test campaigns that perform the best and have the most social proof and then I'll add two or three new ones just to give something completely fresh in the mix, set the budget at the campaign level, like 1000 bucks, and then that usually I'm able to see a much lower cost per lead, a much lower CPA, higher return on ad spin, whatever it is you're measuring, all of those numbers, I usually see out of the gate that that goes way down.
I'm in the middle of a launch right now for some stuff I work on with Ezra and what we tried to do in this launch is it's generating as many leads as possible and then we go deeper into the funnel. So I ran the test campaign for five to seven days. The cost per lead was around two bucks. I took what works, set up a CBO campaign in the structure that I just described. Our cost per lead went down to a dollar 10 and we're generating more than double the amount of leads that we were in the previous structure. A lot of that is because the cost per lead went down and I increased the budget a bit, but it's just allowing me that amplification. But it's also because I knew it was already going to work.
I'm seeing the same things over and over. I just wanted to explain you guys structurally how I see the difference and really that's how I play the game, I guess the way to put it, and how I feel about the budget stuff.
Andrew Foxwell: I think that a couple of things to say there from the minimum-maximum thing, I agree. The only time I've seen it'd be used successfully is if something you really know that is an audience and it's performed before and you're going to set up minimum. I think that-
Molly Pittman: But for me, I would usually just put them in their own campaign with their own ad set.
Andrew Foxwell: Yeah. It's not something... First of all, we're going to be forced into it to some degree. So you have two choices. You have a minimum set or you have a standalone, or as I've termed it on Twitter, a standee where it's a standalone ad set.
Molly Pittman: A standee. Yeah, I'm calling it the easy button.
Andrew Foxwell: I feel like you have two choices there, really. And I agree with you. I haven't seen that be a big thing. I think that... A couple of things that you said are very true. One is that the more budget you give a CBO, generally the better it will perform for sure.
Molly Pittman: If you have that proof.
Andrew Foxwell: Yeah. I've even had instances where we've gone with smaller brands and we switched into CBO with not much data and it does prove to be more efficient right out of the gates with even a newer pixel. So I think that proof helps, of course, but even if it doesn't, switching into CBO is big. I think the reason is CBO right now, to me, there's got to be some part in the algorithm. There's always something in the ad algorithm that you are trying to be on the leading edge of that's going to give you incremental.
Andrew Foxwell: An example of this is you have the same thing running like you said it's two bucks, you switch it into a CBO, you give it a little bit more budget, the audience's a little bit bigger, and it's at buck 10. To me, that's a perfect example of what I'm talking about, which is... It's the same thing with IG stories, what they were in the last spring. If you did IG stories with polls, you saw CPAs you hadn't seen in a long time.
So there's always this thing of an adoption rate that I think right now is happening with CBO and why when I've had instances where we've taken clients and we've switched them more into CBO, it's like... Remember the days when people would be like, "I tried Instagram. It doesn't work." Then they tried Instagram again a year later and they're like, "Oh, it's working." It's the same thing where... I've had people be like in October they're like, "We tried it in October, November and CBO just didn't work." I'm like, "Okay, you got to try it again." [crosstalk 00:17:26] works a lot better.
Molly Pittman: I think it's even more than that though, than a feature. To me, CBO represents really how developed Facebook's machine learning is nowadays.
Andrew Foxwell: I completely agree.
Molly Pittman: Because they know way more than we do about what's happening and why and always will. So it's us handing over the control. The issue with that is, and I say this with love, but people that aren't putting time into the marketing, the stuff that matters, the offer, the copy creative targeting, the good stuff. When you hand over shit to Facebook, it's just going to amplify crap. That's what's going to happen. But I think it's even more than an opportunity. Like a new feature, it's the new way. It's how it's going to be, but it does make the game harder for people that might not actually know their market or aren't putting time into the assets, if that makes sense.
Andrew Foxwell: I think that that's true. I think that's a really good point. You have to have the other parts of it figured out. It's not going to be the easy button. I think to your point I've certainly seen a strategy now, maybe two or three times, in the last two months, in the EU specifically, where you're lumping countries and you're really just letting Facebook decide, or I've seen CBO work where you're separating by demographic in the United States with no other lookalikes, no other exclusions, and you're basically using the demographic as the ad set split. So 18 to 25 women, 26 to 35 women that you're letting Facebook decide between that is why.
In some cases, you're using a dynamic creative. Two questions I want to get into on that. One is I've seen about, I would say, 30% of the time we've tried dynamic creative within a CBO, it does better than anything it's done in a long time. And then, the 70% of the time it doesn't work better than an ad that has social proof on it or something, has more of a Q and A and things like that. Have you seen differences with... Has dynamic creative for you changed the game at all in terms of testing or not?
Molly Pittman: No.
Andrew Foxwell: Okay. That's very valid.
Molly Pittman: I think a lot of that, too, though, is the way that I think about marketing, too, is a bit different than a lot of other media buyers in the sense that when I'm writing an ad, I usually have a creative that's very specific to that story and what I'm talking about. It's not just like a set of headlines ads. It's something I can throw up and just test.
Andrew Foxwell: It's very story-based.
Molly Pittman: Yeah. It's built for the platform in terms of the social nature of the platform. I just don't find a lot of use in dynamic creative, but I can see the benefit of it in certain situations. The biggest thing for me is the social proof thing. That's hard for me because even if I'm running ads and the campaign isn't working on, I know I'm at least building social proof on that post ID that I can use in other scenarios. That part for me to digest.
Andrew Foxwell: I totally agree with you. That's really good to hear because I think sometimes... I've heard from other advertisers that aren't as DR focus that are more traffic focused or more video view focus where using DCO, dynamic creative optimization, dynamic creative testing, DCT, same thing, has helped it be more efficient. I've certainly seen that. There's a huge publisher we work with and it's made it more efficient where they give Facebook more options and we're able to drop the cost per view down by a couple of cents in some cases which is huge for that.
Molly Pittman: When you're doing branding stuff and you do have a bunch of taglines, headlines you want to test, different messages, and you've got broad creatives, I totally get it. But from a direct response standpoint, we've got to go deep and speak to a certain person and it's just a different ballgame.
Andrew Foxwell: Let me get into... One of the questions that I think you get asked about a lot, and I've been asked in reaction to our course as well in talking about testing, in how do you start to creatively test? I'll tell you the way that we've gone about this is basically we have within testing... I think your best option is to create some grouping of... I gave it a stupid name because it's easier for me to remember, I call it the validation station, where basically you take a 180 page view, your email lists, your previous purchasers, whatever, and you have that in a CBO with one ad set. You lump them all together and you basically rotate through creative and you just see what works that way. That's one method of creative testing we've done.
Another method that I've seen work is where people... You have countries lumped that you're targeting or just the United States, you layer in your broad demographic and then you utilize the ad sets as the creative split to that audience. The creative split is you have created is one ad set, one creative. And that's not something that I would... I don't like that method necessarily unless your audience is massive. So I tend to do the first one. Then the updating of that is you can go through and you can pause the ads as you start to see the first time impression ratio go down or see the CPA rise. You can pause those ads and duplicate them and start the ones that have a little bit more proof.
How do you, in the CBO methodology, go about testing and proving creative?
Molly Pittman: Good question. It depends on the type of business I'm working with and the type of campaign that I'm running. But if it's for cold traffic, I use this thing I came up with a few years ago called the ad grid. If you google it, you guys can find it. Because I work with a lot of people selling physical products and the biggest issue that I see is they really don't understand avatars and they don't understand how to test different messages. For me, if I'm just starting out with a brand and we're talking about cold prospecting, what matters to me the most is testing very different hooks, testing very different audiences of people because I feel like that's the biggest lever I can pull versus blue versus green creative. I'm starting in a very broad way, if that makes sense.
The way I would set that up... An example is a client that I used to work with called Panda Planner. They sell these planners. There are lots of planners on the market. You can get a planner anywhere nowadays. This ad grid was really important because I need to figure out what is really important about this planner and who's going to resonate with this. So we tested for avatars, the entrepreneur, the mom, someone who's dealing with a lot of stress, and then architects were actually someone who was using the planner a lot, which was really interesting.
Each avatar had their own campaign... This is just an example, but this is how I test. Each avatar has their own campaign just to keep it very separate. Since CBO isn't mandatory in all accounts yet, I'm still using ABO and I'm just going deep into interests on where to find that person. All the mom blogs and clothing companies, etc., etc., etc. Then each of those avatars has their own copy and creative. So maybe I'm testing three copy variations and three creative variations, maybe one-
Andrew Foxwell: To those interest lumps.
Molly Pittman: Yes, to that avatar. Because the copy is very specific to that avatar. For example, mom ended up working the best and the reason was because... I use this story from a customer about how her mom life was so much better if she had so much less guilt because she was able to keep up with school activities and she didn't miss doctor's appointments anymore. Like this Panda planner made her feel like a better mom. How much deeper is that than, "Hey, random person, this planner has all these cool bindings and it's the best quality on the market you want to buy?" This is how cold acquisition works on the internet today in my opinion.
What I am most interested in when first starting out and it's easier... I call this a product centric business, a business that has a product that can be sold to different avatars. There are some businesses, like my marketing training business, it's like, "Yeah, there were a bunch of sub avatars but I'm selling it like you are information to other marketers." That's more of a market centric business. So I don't have to worry as much about the avatar stuff. The testing is a lot simpler especially with a product centric business like this, that's so valuable. Because the CPA between those avatars range from like 10 bucks to 90 bucks. It really showed me which directions to go in, which types of interests were working, between the ad sets, which ads were working, and then I was able to move those into CBO campaign.
Molly Pittman: I don't know if that makes sense. But for me, it's about starting really high level with the testing, especially if you had more of a product-centric business and thinking... There are just such big levers we can pull that we forget when we get so caught into the tactics of Facebook marketing.
Andrew Foxwell: I agree. Kind of taking it bigger. Well, Molly, thank you very much for your time. It's good to chat about this with you. I appreciate it.
Molly Pittman: Yeah, thanks for letting me go down a few rabbit holes there.
Andrew Foxwell: Yeah, absolutely.
Molly Pittman: Yeah, great to chat with you guys. Thank you so much, Andrew.
Andrew Foxwell: Yeah, thank you.
Andrew Foxwell: All right, Depesh, thank you so much for joining me to talk about CBO.
Depesh Mandalia: Pleasure.
Andrew Foxwell: Let's go ahead and just get right into it. I think there's a lot of misconceptions people have with CBO. Misconceptions around how do you get started or how do you roll into a CBO. I think one thing you've done a really good job of helping people understand is graduating into CBOs, which we can talk about. What are the biggest misconceptions you see around utilization of CBOs?
Depesh Mandalia: I think the biggest one is probably something I fell into as well, which is it's no big deal, all they're doing is moving budgets from ad sets to campaigns and I think everyone goes through that process, and it took me six months to figure out that actually there's a whole different algorithm behind it. Because you've gone from this single ad set. You can only see this one audience then distributes budgets across the ads and then the placements and things like that to now this campaign level budget that's spreading its spend across multiple audiences.
Actually, when it comes to option and competition and overlap and things like that, there's far more control than you ever will get with ad set budgets. That's part of the reason why I personally love it so much is it takes away some of the decision making that otherwise you'd need to make as a media buyer. It just frees you up to get better performance. However, it's not literally plug and play, it's not run a CBO and all the magic is going to happen. There's stuff you have to do before that. I think that's one of the biggest challenges people have is just expecting that they're going to change their structures from ad set budgets to campaign budgets and just fire everything in there and it's going to work. There are a few configurations that you just need to get right as well.
Andrew Foxwell: Getting started or moving into or graduating into... How do you go about getting started? Obviously, right now, we're in a situation and time-wise where we still have ad set budget optimization, but as we move into CBO, how do you start to go through that process of warming things up and then eventually moving into a greater CBO in size or scope?
Depesh Mandalia: Yeah, absolutely. My preference is to work top down in terms of account structure. If you think top of funnel, middle of funnel, bottom of funnel, generally for most ad accounts, bottom of funnel is generating your purchases, retargeting, and things like that. When that's performing, you don't really want to impact that too much. However, on the flip side, your bigger budgets are also at the top of funnel. What I'm trying to do is set up those CBO campaigns and have a setup where I can run some testing with CBO, make sure that we've got good ad sets and ads running, and then slowly start dialing down the budget as CBO starts to improve and you're trying to balance that cross and then work down each step of the funnel.
For example, we did one, I think about three months ago. We moved the middle of funnel, which I call the warm audience, into CBO and it just stabilized and gave us far more consistent return on ad spend than they'd seen on their account before and it was because... If you imagine your engaged audiences, which for me are middle of funnel, people that have been engaged or watched a video or anything like that, some audience sizes will be different. Your budgets will depend on audience size and frequency and performance and you're having to adjust that every single day just based on performance. Now that it's in CBO, you're letting CBO decide that and then you let it run and it optimizes itself up and it just frees you up from the day to day anxiety, which is a big thing for me as well.
Andrew Foxwell: Let's talk about that a little bit. I get a question a lot, "How long do you let it run? Do you have to be..." For me, I've noticed you have to be more patient with CBOs and you see more latencies there. Is that your experience as well?
Depesh Mandalia: I'll give you an example. When I'm running my own accounts, I will try and leave it for at least three days, try and go for longer to let it balance itself out. I know the struggle when you're doing client accounts is a bit more difficult because you then need to explain to the client why, for example, performance isn't as good as it needs to be and telling your client we need to just wait and see what happens is not the easiest conversation, but where possible you do have to give it time to stabilize.
We run an experiment really early on when CBO came out to test the theory of running multiple ad sets and giving it time to let it balance out to the best CPA and it worked. We had three of the same audiences in a single CBO, $150 a day, left it for about five, six days, and you could literally see spend was dropping on the ad set that had $60 CPA and spend was increasing on the ad set that had $30 CPA and that was pretty cool to see.
Andrew Foxwell: I'm seeing the breakdown effect and the post break down effect where it's actually working the way that it should be. A tactical question, are you utilizing a lot of minimums and maximums as it relates to CBO or you letting it as much as you can just run on its own?
Depesh Mandalia: A bit of both. For example, if we're introducing a new ad set to a CBO because that can cause two things. It can destabilize performance, but also if you've got good ad sets that are performing, they've got good history, sometimes a new ad set just doesn't get enough spend, in that case we may put a minimum in there just to help it out.
Andrew Foxwell: You actually add ad sets once it's been running?
Depesh Mandalia: Yeah, absolutely.
Andrew Foxwell: Oh, interesting. I don't do that. I put them in their own thing because I'm terrified of destabilizing it.
Depesh Mandalia: I generally don't recommend people do this because you need to know how to recover. Some of the strengths of the top media buys is not knowing what to do when things are well it's how to reverse out and capture that. So I would recommend people don't touch a good performing CBO. I do it because partly I love to experiment and partly, I know what I'm doing with recovery and bringing it back. But if I do introduce a new ad set, I set a minimum and just pull some budget through, but you have to be careful because you're overriding the CBO as well. If you set a minimum of $50, you're a $500 CBO campaign, that $50 might be wasted where it could have got your conversions on another ad sets. That's just something to bear in mind.
Depesh Mandalia: Also, sometimes I use max as well. For example, if you've got an ad set that maybe started the day quite bad and you don't want Facebook to experiment in it and you know the history of this ad set is by 12 o'clock today it should be performing well, you can set a max there as well. Facebook doesn't go too deep into spending on that as well. But where possible, it's better to just let it decide on its own.
Andrew Foxwell: Let's go through the funnel of how you think about this. Prospecting CBOs, we generally have seen bigger audiences working better, 10% lookalikes, etc., and we generally break out them in what I'm calling scale ladders, I just came up with that name, I don't even know...
Depesh Mandalia: That's cool.
Andrew Foxwell: Then you break them up by size, like 4%, maybe 1%, 4%, 10%, something like that, in lookalikes then you can break them up by high and low value. So high value of purchasers, ATCs, and then you have low value of page views, etc., and then you have interest CBOs as well where you can break them out individually or you can lump them by type, whatever. How do you approach CBO prospecting? Is it similar in that regard? Or do you do a lot more consolidation as it relates to that? I've seen a lot of people do consolidation now as it exits learning phase as well and I'm curious to your take.
Depesh Mandalia: Yeah, it's actually very similar to that. Prior to prospecting, we have a testing phase where I'm testing different interests and lookalikes and audiences. For example, where possible when I hit prospecting, I want to combine those up. For example, I know some people would like to bulk up interest and do some testing. I test all my interest in individual ad sets of the testing phase. Then I want to group them up in the prospecting.
There are a few different reasons. Number one is audience size. The bigger the size, the better CPO performs. But secondly, if you imagine you've got five or 10 different interests, when you combine them together, they're all intersecting. Facebook would find the best in those group of interests with the caveat that... For example, if I've been through a testing phase and let's say five interests have proven to work well, three might work well with a particular type of ad, two might work well with a different style of ad. In that case, I might not combine them up.
An example is let's say you're selling a luxury watch for example and you're going after watch interests so watch brands for example, high-end cars, magazines, etc., it may be that different ads resonate as well. So you don't want to mix that too much.
Then when it comes to prospecting, I do separate out my broad audiences, my interest, and my lookalikes into separate CBOs. As you mentioned, I split my CBOs by my lookalikes by percentage as well because it's about the quality of the lookalike, so I generally do 1 to 3%, four to seven, and then eight to 10 and that's dictated by my testing. For example, the 1 to 3% lookalike generally the same ads work in there, then I might group them up as well. Obviously, the audience sizes are the same. It's that 1 to 3% or relative to that audience size.
Andrew Foxwell: Right. Your answer... This is why it was funny when I saw the CBO Cookbook for the first time, which by the way if you haven't purchased a copy I'll put a link to it in the comments, because you should get a copy. It's like comical how similar. I read it and I was like, "This is like my class. This is like our CBO course." It's very similar in the way that we approach it in terms of the separation of it. You use range lookalikes more. I don't use range as much and that's a difference, but the separation of it and then the testing of all of the interests in separation is also something that I love doing that. And then combining them, as you said, as time goes on, figuring out which one's going to hit with one which is good.
As it relates to the next part of the funnel of mid funnel, this is where it starts to get murkier of how people go about it. We do, engagers generally, in their own CBO and then if the email list is about the same size, sometimes that'll be in a MOFU CBO, and then sometimes something like a top 25%, last 180, or something isn't the same with CBO depending on the size. How do you think about mid funnel CBOs? Is it similar thinking there, too? Or do you do them more separated and making sure that they're MOFU 1, MOFU 2, MOFU 3 type of thing.
Depesh Mandalia: No, it's small. Similar to what you say. The big factor is audience size. For example, if you have engaged lists and they're into the 50 thousands, hundreds of thousands, and you've got an email list into the millions, then I might consider changing it around. But in most cases the audience sizes are similar. So I have one CBO, the multiple ad sets running, and then just cycle the ads in that way as well.
The goal for me is to try and get those audiences separate out. Even something like engage with Facebook, engage with Instagram, I'd run those as separate ad sets. To give an example on my own info product ad account as well, we consistently see far better ROAs on the Instagram engagers than Facebook. I haven't still figured out why because my Instagram profile... I don't really talk about Facebook ads. That's more personal stuff. Whereas, on my Facebook it's all Facebook stuff but that performs better. So you can see the spend going through and Facebook is self optimizing.
But sometimes what I will do is just trying control the spend as well. Like I said with the kind of max as well. If I see the performance in the day not quite kicking in where it needs to be, I don't really recommend people do that. Partly, I do it, like I said, I just like to keep experimenting so I'm always just tweaking. I've got a Slack channel, for example, of CBO... I can't remember what I called it, like findings, that every time something changes in terms of what I thought would work and it's changed, I just write it down because I'm trying to log all these tweaks and stuff because CBO is still being matured, I think that's fair to say.
So Facebook is still doing tweaks and things like that. One example is we tried to scale heavily in November in one of our CBO campaigns, which 2018 you could scale it 100%, 500%, and the budgets continued and ROAs was good. This year in November, when we were scaling aggressively, the budget just wasn't shifting. It was almost like we just didn't make a change. When I spoke to Facebook they said, not in these words, but they put a dampening in there because prior to that people were upping their budgets and CBO was just spending it so fast and people are losing a ton of money. So then they went the other way to try and control it, which I think they said, "If the budget increases more than 30 or 40%, it applies a lot more pacing than it did before versus 30, 40%, then it's more likely to have a direct impact."
Andrew Foxwell: That's interesting. That's good. That makes sense. Because we've seen similar things. I've had a number of inquiries from people saying, "Why isn't my CBO spending all the way to the cap?" I'm like, "Well, I guess that's what was taking place." But I haven't confirmed that. So that's good to know.
In terms of looking at remarketing, we tend to, with remarketing CBO, separate them by size. Same thing. Really recent or like hot and then like warm remarketing. Is that similar to how you're doing it as well?
Depesh Mandalia: Yeah. One of the exciting things for me with remarketing with CBO is combining it with both DPA, dynamic product ads, and also, DCO, dynamic content optimization, is what I call the infinity retargeting. In the sense that if you, for example, target the people that have added to cart in the last three days, every three days, that audience refreshes. Brand new people are coming in every three days. When you're running dynamic creative, Facebook optimizes that further to serve the right note to the right person as far as it knows.
It gets to a stage where generally you don't need to touch the ads because the audience is refreshing every three days or whatever your window is. And then it's always testing these dynamic ads out. It's just a cool thing to see that you can get your bottom of funnel working really well and it almost becomes hands off. Then you can focus on mid and top funnel.
Andrew Foxwell: You're using dynamic creative in lower funnel more often?
Depesh Mandalia: More often, yeah. We've tested it in top of funnel actually and it has also performed well there. As long as we've gone through the testing phase and tested the different variants to see what works and then push it into prospecting and just put a single DCO in. Obviously, the big downside is social proofing and I like to use my page post engagement and use the social. But there is the added benefit that with the cold prospecting, with a top of funnel, the DCO just keeps cycling and delivers the right ad format.
The other thing, I don't think a lot of people understand, is Facebook has propensity models based on what type of ad people will respond to. Image, carousel, video, all this stuff, and when you're able to run a DCO and you've got a video and a static and a few different variants, Facebook knows that this particular user will react to this format. So it will select out the right format for them. And I think that's a really cool thing as well.
Andrew Foxwell: Yeah, I think that's incredibly powerful. We've tried it in prospecting, we've tried it in all parts of the funnel, and it's 50/50 whether it works or not. It's 50% of the time it's fine, 50% of the time it doesn't do anything, and I think it's because of social proof. If it's a product that people get easier something, then it works. But if it's something that requires more proof, then it's harder to get that to go. That's good. I think we've integrated a few times and coach people to integrate DCO in the mid to lower part of the funnel, but it hasn't been as big of a part of it, probably could be. That's a really good suggestion for people as well.
I guess there's a lot of different other things we could talk about. One thing that comes to mind for me is asset placement customization, which is sort of related to CBO. Mostly, customizing Instagram stories is where we've done most of this. Is it helpful for you, asset placement customization, or do you just let the consolidated ad run and if it doesn't look perfect on IGS then it's fine? There doesn't seem to be a consensus on this and I think it's obviously because we're all not sure. Sometimes it's fine, sometimes it's not. I'm curious of your take.
Depesh Mandalia: I start off and just let Facebook do its thing without customizing and it's a catch 22 because if your ad format isn't great for stories to start with... I look at the breakdowns and see if I need to optimize further. You could end up having bad performance in the stories, but it's probably because your ad isn't really designed for stories.
Then it's also down to the brand as well. For example, in eComm in particular, stories can do a really good job of just 15 seconds short ads that focus on just one core aspect of the offering and make it cool and interesting and just get them to engage with that. The story ad format still has so much more potential because there's less competition and it's still cheaper, I think, than a lot of the other placements out there. I just haven't pushed enough on that. That's something for me for 2020 to do a lot more testing on stories because I just think there's so much more to be had there.
Andrew Foxwell: Just to give credit where credit's due, last March is when Florian Litterst from Germany from adsventure.de, he was the one that suggested for me to try stories of polls. Since then... It's amazing. If you have an account that has an Instagram centric audience, scaling with stories horizontally and doing a 10% LLA, CBO with just stories can add incremental for us. It works really well. But sometimes... Then I tell the client it's going to work and then we try it and then it doesn't do that well. Classic Facebook issue. But I still like it and I think stories is really big.
I was trying to read between the lines on... There's a big comforter brand in the United States that's really bullish on stories and I've got to think based on their ad library, that 80% of their spend is going to be stories at this point. They've still have feed, but so much of their creative that's running it looks to be stories. Who knows? That's just a guess. I know that they're spending a lot, so it's interesting to me that there's enough stabilization there.
Depesh Mandalia: I wonder what the demographic is and whether that has a play as well.
Andrew Foxwell: No, it's definitely young, probably under 35 in the U.S. To be conscious of your time, I'll only ask one more question which is. if people are sitting here and thinking about, "Okay, I've tried CBO, I'm trying some of these different tactics but I'm looking to also review the quality of what I've done," how would you guide them? Clearly, the cookbook's going to have some of these things and it definitely has it in there. Many of its pages is... It's big. It's very long, very comprehensive. But what would you guide them towards?
Depesh Mandalia: This is an interesting time because right now you can at this point still choose between ad set budgets and campaign budgets. And I want people to understand this is not us versus them. It's not CBO versus ad sets. If ad set budgets are working for you, that's great. Continue running with it. But use this time to test and experiment because the latest from Facebook is CBOs are going to be a mandatory on X date, whether that keeps changing or not, who knows? This is just Facebook. They're the type of people who will land a massive change right before Black Friday. This is how they operate.
What I'm trying to get people to understand is don't get caught out in the cold because at some point CBO will be mandatory. Start experimenting and see what works, use your own budgets, use free knowledge from groups like yours, Andrew, and just see what works for you, or if you want to fast track that and get paid program, then you just start testing the theories out. But the thing is just start testing. Don't just sit there with your ad set budgets and put your head under the sand and say, "Look, I'm in a good space here." That's my key thing is just start to do that experimenting. Just start with one part of the funnel. If it's just top of funnel, bottom of funnel, whatever it is. Start testing and see.
When I'm looking at my kind of ad accounts and what I want to do, I solve it with a hypothesis. That's all it is. If I do this, will this happen? What's the outcome? That's how you need to start off. CBO, for me, the reason I'm such a big fan, is because it does give a lot more stable results and it allows you to step back from a lot of the worrying about daily performance. I went CBO only around, I think, early 2019 maybe, mid 2019 before Facebook announced it because it was working so well for us.
Obviously, then Facebook announced it's going to be mandatory and that's why myself and you, Andrew, we're trying to get people to understand not only is CBO powerful, it's obviously mandatory as well, but just start testing. There's no reason not to do that. Whether you're taking listing campaign, your assets, just copying it over to a CBO format and just put some small spends behind it. So that's one side.
The other side, which I don't know if you've heard a lot about, is a lot of people are worried because the small budgets don't seem to work as well with CBO. That's actually not true and I'm actually trying to get people to focus on how different it is. For example, we've taken on a client, they do a food delivery service and their budget's right now $30 a day and we're having great results with CBO.
How are we doing that? First of all, we're not putting too many ad sets and ads into the campaign because the more variants you put in, the more money you need and secondly, the funnel is quite long. So someone clicks an ad, they select the location, delivery type, menu items, they add to cart, they register, the whole process right to purchase is quite far. If you were to optimize for purchase, you need far more budgets to make that work. We found that optimizing for complete registration, which is halfway down the funnel, gives Facebook the stability it needs because we're hitting enough conversions a week.
Then we're also getting good performance because people that register are more likely to go and purchase as well. So when you're working with small budget CBOs, that's the balance to be had. Give Facebook the data it needs and think about minimizing your variants. For example, for that $30 CBO, we only run three ad sets and in each ad sets two ads and that's it and it works.
Andrew Foxwell: Condense and consolidate, that's been my advice. We also actually have a low budget food delivery service.
Depesh Mandalia: Oh, cool.
Andrew Foxwell: Same exact thing basically because they're not going to small geos and so in theirs it's basically each geo has its own budget. We've just condensed and consolidated and we're trying less things. Actually, we're trying a couple of variants on, actually DCO, which has been productive for them. Even at a very low span. They're also about $40 a day max. It's exactly the same thing.
Thank you very much for your time. I really appreciate it. If people want to be involved, I recommend that they join your group as well, of course, and check out and purchase the CBO Cookbook. Thank you.
Depesh Mandalia: I appreciate it. Cheers, Andrew.
Andrew Foxwell: Hey, one last quick thing. Did you know that Austin and I are now taking your calls with questions? We can talk about a previous episode and you can ask us any question you'd like on any sort of topic related ecommerce or just leave a comment. We're taking these questions on a phone number that you leave a voicemail on. The number is (512) 710-7578. Give us a call and leave us a message and let us know what's on your mind. Again, phone number's (512) 710-7578.
Austin Brawner: Hey, guys, it's Austin and if you've been loving the podcast you got to go check out brandgrowthexperts.com. That's where I work, one on one with my clients to help them build faster growing, more profitable online stores. I've got coaching programs and workshops that we host all over the world. Would love to have you come check it out.
If you're a fast growing eCommerce business or you want to be a fast growing eCommerce business, you got to check it out. That's the spot for you. We go more in depth than we do in the podcast with comprehensive trainings and coaching to help you scale up.
Check it out, brandgrowthexperts.com. See you there.