Meta Ads 101: Mobile Apps

That's already a pure gold Marcos. So the rule of thumb kind of here if you see that in one adset you have kind of mixed audience as you said you would want to consider and create a new adset per an audience. Yeah that's what I would do. I mean you can also make assumptions beforehand and already like oftentimes it comes down to creative type and messaging. So if you for example have short for Hi everyone, welcome to another episode of Tangent AI11. I'm Roman from Tangent and today I'm joined by the guy for Meta Ads, Marcus. Hey Marcus, how is it going? Good to be here. Thanks for the invite. Uh tell us uh why have we invited you to talk about the meta ads? Well, I don't know why you have invited me, but I guess because we had lunch together and

then we thought it would be a good idea. Um, no, I mean, in the end, I've been running meta ads for kind of the past 13 years. Started my career in the gaming industry, moved to subscription ads in 2019, always with Meta as my primary channel. And so yeah, I've been quite active on LinkedIn in the past two, three years now, publishing content around the topic. Um, getting good feedback when it comes to like people implementing my stuff and seeing good results. So yeah, I think it makes sense to go through some of what I talk about on LinkedIn in a little more detail. And that's why we're here today. Exactly. Yeah, Marcos posts a lot of really cool stuff uh in all formats. I already saw a video carousel about meta ads and this is a background for why we're doing this recording and it turned out we never had a video about meta ads oneonone. So this is why uh I thought like Marcus would be a best person to do that with us. Um

we've uh prepared some slides. The slides are based on Marco's uh link in post. We will of course post everything in the description of this video, both Marcus account and the the source post basically. Uh but without a further ado, uh let's do it. Marcos, just share a screen, go through the slides, and I'll ask some questions during the presentation. Sure, let's do that. And yeah, thanks again for putting them together. I mean, I think it's great to kind of have something prepared visually that we can walk through. Um, so yeah, that's our secret sauce. Thanks for reading my stuff and getting it 95% right when transporting what I'm trying to say into into the presentation. Okay, let's jump in and start with a channel level strategy. Um, so one thing I love to always talk about is that uh meta in the end isn't just a singular channel. Um, you have a bunch of different placements in there and each single one of them is as big as a

channel in its own. Um, so like Facebook versus Instagram are two completely different apps that you're running ads in and they have different audiences and both of them are huge. Um, so in the end the the setting of your placements and ads manager can like look quite trivial or you just go for advantage plus placements and don't even think about it. But the inventory behind that is completely different will deliver different results will mean needing uh to run different uh creatives. So, it's actually kind of a big thing to keep in mind when you're running these um these ads on Meta and they're always adding new ones. I mean, uh now recently threats was the next biggest addition to that mix. Um the previous ones were the real placements on on Instagram and Facebook which are their kind of competitor to Tik Tok and kind of that upperfunnel um short form video. Um and I'm excited about threats. I'm I'm just adding them into the campaigns I'm running, trying to see like how can I

actually get some distribution there. For now, I haven't. Um, but in the end, that comes down to um like what creatives are you running and are they a fit for that kind of um audience on the placement. So, I think the first people that are figuring out like what does an ad on threads need to look like in order for the algorithm to distribute it there are going to have a good time because so far I don't think many advertisers are spending a whole bunch of money on that placement. I think the app is at like 400 million uh active monthly activives or so. Yeah. And we can uh first noob question. We can choose the exact placement in meta. We can um but I recommend not to in the end. So like the more you the more you restrict the algorithm and the choices it can make, the higher your prices in the end are going to be. And ultimately where you want to end up is that you're able to run an adset on uh advantage plus placements meaning all

placements are enabled. The algorithm then u makes a decision on where to spend your money and then you analyze if it's doing a good job and how to steer it in the right direction. I wouldn't necessarily do that by um limiting placements in the in the first hand just because it will be more expensive. you are likely not making the right choice because often it's just based on kind of assumptions. While an algorithm can of course always find efficient pockets within a placement. Also, it's just very efficient if they're for example able to show an ad to the same person on first Instagram reels and then on the Facebook feed if they use both apps. Um so kind of they can look into the whole user journey there. So, in the end, the algorithm is going to make smarter decisions, but you need to question it and kind of steer it in the right direction. And I'm sure we're going to talk about that in a minute. Gotcha. And for the threads, for example, how do you like I don't know,

you you run an experiment for the threads, but uh you still want to do advantage plus um I wouldn't even call it an experiment right now. It's not like I set up an AB test or anything. Basically, I'm just adding it into my running assets uh as the placement becomes available in uh in the dropdown. For now, at least from what I saw, Meta hasn't kind of automatically added it. So, you still have to kind of choose it, which is actually quite nice of them. With many of their settings, they just auto enable them and then you have to figure out what's going on. Um so, I'm adding them in now um here and there. And then I just analyze the data looking at am I getting distribution on that placement um and if so what does the performance look like as it's per as it's reported in meta. So really for now it's about figuring out how to deliver there because I think usually as a first step that's the challenge that kind of the algorithm is I mean they have a ton of data on how things convert and perform on Instagram on reals on Facebook. So if you're

running conversion optimized campaigns, they're likely not to directly kind of spend a whole lot of money on a new placement just because the algorithm doesn't know how it's going to do in the end. Yeah. Gotcha. And I also read they starting to show ads on WhatsApp. Have you read that? Yeah. Yeah. Yeah. That's kind of the next addition here. Um yeah, I wonder how how relevant this would be for like app promotion. Um I guess that's really more for like lead generation or so. And then maybe ecom I think they have these kind of branded channels that you can run stuff in. Um yeah, but I mean they're always on the lookout for new placements because they need to show their uh investors that they can increase kind of ad load and revenues. I mean they even added ads recently to notifications that you can kind of run an ad um in like a Facebook notification. So, I mean, they have this these tiny weird placements where I've never seen an ad delivered, but I guess if you optimize, for example, for just brand aare awareness in instead of um

conversion or like any really upper funnel goals, then you are likely going to see kind of um a more even distribution across placements based on how much reach there is on them. While kind of the more you optimize for quality, the more you're going to be segmented onto placements where it actually makes sense for you to run. And often times that's not going to be a notification or whatever they have like Facebook Marketplace or all these kinds of weird things in there. Yeah. Yeah. Gotcha. Okay. Gotcha. I think Yeah, we can go to the next one. Yeah. Let's move on. Quality versus quantity. Um I mean, yeah, we touched on it already um in the end. um as kind of there is so much that you can learn about this channel and about like the difference placements in it. I always promote that you want to go deep on this channel and I think it's the same on others. I mean also on Google ads there's a lot of nuance to it even though you kind of have less um u kind of settings you can still do and it's a

lot more um already automated. Still you want to get a really good understanding of like where are my ads delivering? how's that performing and how you in the end verify that algorithms are taking the right decisions for you. So it's I think like not the best idea. It always it seems easy today to set up a campaign just because everything is quite streamlined as meta calls it with advantage plus and you have just a few clicks. They take all the decisions and then you run the campaign. But in the end you're likely not going to see like great performance from the get-go. And if you kind of manage uh five channels with a low budget without understanding any single one of them in greater detail, then it's going to be hard for you to scale or even get to profitability in the first place. So I'd say like pick maybe do some um like first exploration in the beginning where you maybe try Google ads, meta ads, other channels just to see kind of

what's the initial performance and if there's any one that you feel like you want to double down on. but then rather go deep and really invest for a few months, put all your marketing budget there so that you kind of know what you're doing. You're driving sufficient volume to these algorithms and can really learn um like how that channel works. And on meta, that's also going to mean doing that for the different placements. Like really try to figure out how you can at least get delivery on Facebook versus Instagram so that you can figure out like if your if you can find the right audience for your app um on their two biggest placements that there are. Um because often times what I see is that people do the usual. They have them some UGC videos because that's been kind of the hot creative type to use these days. So, some talking head of a person talking to the camera saying, "Hey, use this app. It's great." Or making a shocking face and then showing a screen recording. Um, they run that, it gets delivery mainly on Instagram res because that's where these videos um tend to be kind of the native format and

it doesn't perform and then like they feel like meta is not for them. Uh, but they really only tried um like one part of uh everything that is available there. So go deep, test really broadly in within that channel like totally different creative types uh and kind of figure out if there's a placement that can work for you. Gotcha. Do people still use meta as a uh first channel that they start with from your experience? Yeah, from my experience it's the main one. Um on iOS even more so on Android Google ads is a viable option as your first channel just because they have like strong inventory there kind of play store search is embedded within that. So if you run Google ads you're basically running something like a combination of Apple search ads and and meta on the iOS front. So you have play store search you have YouTube for like discovery and video. So that's a good viable option

for Android, but they haven't been as strong on iOS just because like the whole tracking is a lot more complicated for them on on on um Google. They own uh the kind of the system completely um on Android. But I mean, I hear there's updates coming on iOS as well where they're going to use IP address now also for tracking. So that should make things maybe a bit better on iOS as well. Um, we'll have to see, but definitely for now on iOS, meta ads has been the primary and first channel for most apps and then most apps tend to launch on iOS. So that's where kind of the focus on this comes from. Got it. Got it. Let's go on account architecture. So when you're building out um your Meta account in the end, what is the main recommendation from from Meta and from kind of all the stuff that they put out there is that you want to consolidate consolidate consolidate because in the

end you need sufficient data on a campaign and it needs to be able to learn um so that it can take decisions for you and that does make sense to some extent but you definitely want to question it and figure out like how can you um work in sync with the algorithm? How can you steer it, use it to your advantage and not just kind of throw everything in but one bucket and hope for the best because if that doesn't work um then you don't know what to do next. And if it does work uh then you're very reliant on like one campaign that is super consolidated and if it breaks then you're going to be in trouble. So basically um you want to be able to diversify your account in a way that makes sense. Um what that can mean is for example that you're running different campaign types. Um there's app promotion campaigns which are usually kind of the first one that people use um if they want to advertise an app. Um mainly that comes down to either using an MMP like Tangent or others um or the

meta SDK to feedback uh app events to meta and then optimize for something like a start trial. um to then uh target their campaigns. Um there is but also web campaigns um web sales objective that you can use to send people to a landing page um to either convert them directly on the web or even send them then back to the app after that. And just by using these different campaign types already, you're going to be accessing different audiences because Meta kind of has tagged everyone's recent performance against all the different um objectives. And there's just some people in that web sales bucket where they say, "Okay, these people are more likely to react to these ads compared to app promotion, which tends to skew quite young." App promotion, like sending someone directly to the app store tends to be a bit younger for most apps. Um, it tends to deliver a lot on these kind of upperfunnel uh short form video placements like Facebook reels, Instagram reels. So just by diversifying

into these different campaign types, you can already make sure you're accessing a broader audience. And maybe it's the channel in the first place just hasn't worked for you because you weren't accessing that audience. So that can be a big unlock. And the same goes for um how you then do creatives. Um in the end, creative is targeting. We've heard about it before kind of everyone talks about it. Um, so as you uh structure your account, really learn which creative delivers to which audience and if it should be competing against other creatives. Um because if you have an ad that is bringing you like 55 plus year old females and you have an ad that is bringing you 18 year old males then maybe they shouldn't be competing on the same cost per trial target because they're going to be much more or less likely to convert to a paying subscriber uh downfunnel which doesn't mean you shouldn't be using both and try to buy both audiences but they need a different goal and hence you want to then diversify your account by having

adsets that then um only run ads either for one audience or the other and then you can steer performance based on what you know uh is going to happen downfunnel. Oh so that's already a pure gold Marcos. So the rule of thumb kind of here if you see that in one adset you have kind of mixed audience as you said you would want to consider and create a new adset per an audience. Yeah that's what I would do. I mean you can also make assumptions beforehand and already like often times it comes down to creative type and messaging. So if you for example have short form 9 to6 video UGC content you can be pretty sure it's going to go mainly onto your Instagram and maybe Facebook reels and then on Instagram res the audience is just going to screw a lot younger. Um while if you for example have a static ad um that's mainly going to run maybe in a Facebook feed or Instagram feed where the audience is going to be a bit older higher quality. So kind of over

time like if you look into your breakdowns in the end meta is still very good at giving you data on a breakdown level. So you can always see like where did my ads deliver in terms of placements and what age did that result in. And that's what you want to learn over time, like what ads should I make in order to run them on a specific placement and what audience does that result in? And then that's kind of what you're using to structure that account. And when starting, do you usually go with like, I don't know, worldwide campaigns, all ages, all genders, or you have some sort of brief or understanding for from the client and their product? Yeah, I wouldn't go worldwide, but usually broad. So like within a market, then it's going to be broad, all genders, all ages, and then we learn over time like where where does it make sense for us to target and how do we make those ads that go after these

audiences. Um, most apps are going to be pretty focused on like US as they go. Maybe there's like local apps that are then I don't know French to start with or German and that would be advertising in those markets of course but market or geo is of course another layer to think about within that um diversification and account structure because also um where a person come from will dictate their p purchase power. So again, the downfunnel conversion is going to look different and therefore you want to make sure for example if you're trying to go like broaden your geomix then don't throw everything in one bucket. Like if you have one campaign that is targeting the US uh I don't know Brazil and uh Spain at the same time then your average revenue per user from these audiences is just going to look a lot different. Um, so in the end again you want to use campaign structure to then steer into these different markets towards different goals. Gotcha. Super helpful Marcus.

And last one on here is like creative testing. Always separate that out in the end. Like if you got a new creative first and foremost you don't want to mess with your scale campaign performance and when you upload a new creative that resets the learning phase. So don't go there and like upload new creatives into a functioning campaign every week because it's going to uh tank performance. And on the other hand, of course, you want to ensure delivery on these creative tests because you want to learn like does this work and then figure out why does it or why doesn't it. So therefore, have a separate creative testing campaign that you upload your new batches to on a weekly bi-weekly basis, however fast you can produce. And then um don't put too many in one adset. um like I tend to go with like four to six maybe so that I get at least uh some delivery across most of them and can then in the end look at the data after my testing time frame and figure out okay what had a good hook rate what had a good hold rate which delivered results and what can we do in terms to um optimize them as a first

step uh next step got you let's maybe talk also a bit about this uh learning phase for the campaign I I I I remember that there is a Yeah, obviously one for an account. So sometimes people really like in my experience some people were really hunting for the account with a good uh uh learnings. Let me put it like that. Uh but also you're mentioning the campaigns have a same kind of meaning. I mean the main one people talk about with learning phase is that like on an adset meta tells you to um collect 50 events and then the adset is going to know what it's doing targeting the right people. Um in the end it's an artificial number like the algorithm is always learning and it will have a better understanding of your audience after 10 events. It's going to have an even better understanding of your audience after a thousand events. It's not that kind of 50 like there's a magic threshold and then afterwards everything just works better. It's more

like gradual I would say. But there is a point that like if you're not achieving enough conversion events early on um then that's going to be um a problem. And I think we even have probably we have slides on that. Nice nice transition. We have this one. I'll go back to the other one after. Um so performance signals um so signal volume quality um and the learning phase are all going to um influence the performance in the end. You want to drive sufficient signal 50 a week, 10 a day. That's kind of usually what people say is kind of a good starting point. Then and quality means in the end you want to first of all optimize for an event that makes sense for you. So it's like it correlates with value. Um it's not just kind of a click or anything because anyone can click on an ad. That doesn't mean that you're going to um in the end also create revenue. And the other thing is that the event needs to be enriched with um enough parameters and happen in a moment

where like meta can actually capture it and that often needs some testing. I just did a um webinar with that with Thomas PT on like signal engineering. We talked a lot about like signal quality as well. So in the end I've seen sometimes that meta was only able to capture 10 to 20% of the conversions that were actually happening. So I could turn on a campaign, see there would be I don't know 100 trials happening a day um after turning on the campaign and before there was zero. So I could be sure that's meta. However, meta only reported 10 to 20. And that comes down to like how like when do you trigger that event and how do you trigger it? Um uh I've se seen differences if you use for example automatic logging through the SDK versus custom events using an MMP versus the SDK. So, it makes sense to invest some time there and like figure this out because once you're at scale, it's going to be a whole lot harder to like challenge your setup. You first of all don't want to interrupt anything and also as you have more and more traffic,

it's going to be hard to find out what's the actual incremental value of a campaign. So, early on, I'd say like really kind of do some do some technical tries there. um cross check with the incremental uplift and figure out like what's my match rate of like the events in the account versus reality and you want to get that somewhere near of like 70 80% um that's usually quite good. you're never going to have like a onetoone match, but you don't want to end up losing half of the signal along the way because that just means one uh for once you don't know what's working, what's not because half of the signals are missing and it might be 30% on one campaign and 70% on the other and also the algorithm is not going to learn properly if if you don't have enough signal flowing through. So, signal quality big big topic as you're setting up that account. And then um yeah, the other one I would say when it comes to signal is that start trial is often a good starting

point. It makes sense for most subscription apps, but it is flawed in the end because trial conversion um on an audience level can vary massively. You might have some users converting at like 70% conversion rate and others only at 20. So if Meta only looks at your cost per trial, then they're kind of grading them all similarly. So if someone comes in that is very young, doesn't have purchase power, and starts a trial for five bucks, the algorithm is going to be raving about it and feel like, wow, this is great. But then a user that comes in at 10 bucks from an older audience with higher purchase power might be much more valuable to you, but the algorithm doesn't see it. So that's where you really need to cross check what's going on in the account with what's happening in the app. And if you're an onboarding asking for how old are you, what's your gender, then you are basically going to have um downfunnel performance and can map that to what's going on on meta and figure out where do you want to be spending. And that's kind of the main thing that

happened after where that kind of step broke from attributing the actual person to your product analytics. Now you have to use kind of that aggregated data to then look into um how does that map out when uh against the audiences that we're buying on meta. And last but not least, what's also on here is um like different signals can be used for different goals. Um so starting with a start trial, usually a good idea. Um, as you grow the account, you could, for example, look into qualified trial events, um, where you kind of move a bit deeper down funnel and think about like, should we really send a start trial for every user or do we want to exclude certain ones, for example. So, sometimes that can be younger users just because they tend to not convert as good. You might have other other qualifying questions in in onboarding where you kind of try to figure out like if users answer this question then their trial conversion goes up by 40%. So maybe

that's the right audience to optimize for. So kind of going a bit deeper there can be a good unlock as you're scaling. But I would say early on it's rather about like really getting your creatives right. Um knowing which placements to deliver on. So don't get don't make this too sophisticated as you're um just starting out. A start trial is often kind of good enough to to go with in the early days. Gotcha. So it should be like early enough and meaningful enough, right? Yeah. Really early is the big one here. Um so don't try to optimize for events that happen like on day two, three, four. So it should really happen within the first 24 hours or even 16 hours. So in the end you want something that happens in first session um not kind of beyond that because that's going to mean you lose a lot of people. You don't know when they come back for a second session. So really anything in like onboarding still makes a lot more sense than than waiting for them to I don't know start a trial

and then set up their account at some point later. That's just too late and too few. Gotcha. Gotcha. Okay. All clear. We you should go back. Yeah. I think I covered most of this already as well, but okay. Uh let's let's talk a bit about it. Um yeah, we mentioned kind of never optimize for upper funnel goals in the end like clicks, installs. It's just in the end meta is very good at giving you what you ask for and then there is a lot of people asking for uh like revenue and for uh conversion events through app event optimized campaigns. So the ones that are then saying, "I'm going to just ask for a install, get the inventory that is left over because it doesn't fit in the other more qualitative bucket." So I haven't seen an install optimized campaign deliver like value in forever. Like as soon as optimization was launched, um you would have to go with it because otherwise you're just buying the stuff that's left over because it's not event optimized.

I guess it would only drive you installs, right? Yeah, in the end you're getting installs and people are dropping off super quickly. No one is even starting a trial. Um and definitely not converting. So um I mean there are probably business goals where this makes sense. Like I don't know if you're than if you're uh rather monetizing through like ad views. I mean there's a separate um conversion goal for that. So I guess it's also the case there that you need to optimize for ad views and not installs anymore. But also, I don't know, if you have a product that maybe like if you rely on volume in the early days or I don't know, you want to juice your ratings on the store, then maybe there's a way to like get value out of this, but it's it's rarely the case. So, like move further down the funnel, get closer to a business goal, but keep it within the first session is usually what I'd recommend. Um, yeah. I mean, like in the good old days, yeah, you wanted to drive so many installs to

get to the top of the app store because there was some traffic there, but it's not case anymore. I wonder how much you'd have to spend. So, that that's going to a lot. It's probably not going to be all that profitable these days. Yeah. And then, yeah, as I already said, start Starting a trial is just a proxy. Make sure you understand like who is converting and who is not. Like I've seen people kind of optimizing for very low trial start costs uh without considering their conversion. Often times that's also a kind of issue I see when people move to Tik Tok. Um so they optimize for app events there as well. They see half of the trial start costs compared to meta on Tik Tok. But then Tik Tok is just a lot younger. their algorithm isn't as mature. So, you have to expect lower downfall conversion and quality. Again, doesn't mean it won't work. Like, if you can get kind of if

your conversion is 50% lower, but your costs of acquiring users are 70% lower, then it's a good deal. But be prepared that a $5 CPT on Tik Tok isn't the same as a $5 CPT on on Meta. And that's kind of the thing you always need to keep an eye on and use your onboarding questions for to be able to segment that way. So if you run meta and Tik Tok have a question in onboarding where you ask where did you hear about us? So you can get that segment where you're going to uh be able to analyze downfunnel performance. It's never going to be perfect. It's not like everyone that came from a Tik Tok ad is going to answer Tik Tok because maybe they've also seen a meta ad before or they watched a Tik Tok video and then they went to the app store downloaded from there and answer app store. So it's also again just a proxy. But if you kind of layer these different data points on top then you can get gain a better understanding of how to steer like your user acquisition in a way that's going to be more profitable for you. Makes this question mandatory?

Um I it depends like I have apps where they're able to skip um but then we don't tell them. So sometimes like for those people that just click through like we get some that don't answer at all. Um the more important thing though is like randomize the answers because often times for the people that click through like they just tap tap tap tap tap tap and then it often ends up with like one of the middle ones or so. Yeah. Yeah. So if you randomize the answers then you can make sure that you get a more qualitative data input there. Um but then yeah some clients of mine don't force it others do but we never like we never make it super obvious if you if it's not forced because in the end I I want to have answers there. Okay smart. Um and then last one here. Yeah I mean we talked about conversion events already a lot. So let's let's rather move on targeting. Um, so as we already said, targeting is done by creative these days. You usually target broadly and

then let the creative steer you into the direction that you want to go in. Um, and in the end, like if you were around before before and maybe even before optimizing for app events back then, an account would be super segmented. We would have a 100 campaigns with a ton of adsets in each one of them where you would go after one would target 18 to 24 with a certain interest targeting on top. Another one would do look alikes. Um another one would do the older age group. So it was super segmented and then you would have goals on each one of them bidding for a certain CPA from these group. That thinking doesn't apply anymore at all. But of course, different audiences still have different value and that's what you need to kind of um account for and as I said it still makes sense to try and diversify your account in a way that allows you to access these different audiences. So while segmented segmenting isn't the way to go anymore of course still you want to be able to

address like a large population if you want to scale um and if you're hyper consolidated with just one campaign one adset then meta is making all these decisions and often times I see them not making them so smartly like often times one ad that is the top spender just gets 80% of all the spend one ad means just one targeting and one targeting means you're only going after a certain audience So, as you try to scale that ad, it just gets expensive very quickly because you're just kind of targeting the same uh people over and over. Um, in terms of like age targeting, um, yeah, one thing I talked about before is it got more complicated on meta in the recent years. So, they're pushing more and more towards not doing H targeting just like Google Ads doesn't allow you to. So, they've been hiding the more restrictive targetings. Um, so right now if you are already on the new setup that only allows you to create advantage plus campaigns, then within that if you make an age targeting, it's just a recommendation to the algorithm. So they

say, "Yeah, you can tell us you want to go after 40 year olds, but we're still going to try everything else and if we find it that they have lower cost per trials, then we're going to buy them." So, in order to still do a restrictive age targeting, you'd have to either move back to manual audience, I think they call it, or um yeah, original audience. Yeah, true. That's what it called. Or you can on account level actually set a toggle that says I'm advertising uh age restricted goods, which allows you to move it up to like 25 plus at least. And that's helps you get rid of that usually very poor quality um uh very young audience. And that often times makes sense especially in the early days. I mean you're not looking for massive scale yet. You're trying to prove a channel early on. So like at least excluding the audiences that are generating a lot of traffic but not converting can make sense um as you start out. But be aware meta is trying you not is trying that you don't don't do it. Um

advantage plus placements. As I already said I tend to use it. Um it's kind of an advantage plus setting. I'd always recommend because in the end, if I test an ad, I want to learn where does it deliver and then I can still take the decision whether it makes sense to run that ad on that placement. But if I limit my placements, then I'm I might be forcing creatives onto placements that don't belong there. And I'm never going to learn if there's like profitable pockets within other placements. And I've seen accounts running on very weird placement mixes like uh um I recently came across an account where client client within one adset was spending like 80% of budget at like at a scale budget um on uh audience network which I haven't seen in a while. So it was mainly rewarded video ads in audience network was an Android campaign. So audience network is more of a thing there. But I think most people wouldn't really consider this at least not in non-gaming as one of the adsets uh one

of the placements they can scale in. But they did find in creative that works there. So then the question was okay how can we make more of it? It seems to be working. How can we own that placement? And then we basically access this completely different inventory. It's like a new channel all within from the same UI while you don't have to kind of add the headcount and the processes of running an additional platform where you would be able to buy a rewarded video placements. So definitely um recommend it like figure out um what placements work for you through advantage plus placement. And last but not least, look alikes. Also one of my favorite topics. I think no one is really or like promoting them at least anymore. many more the accounts that I look at aren't using them. Uh while back in the days it was the single most profitable strategy for everyone. Everyone uploaded all their user data to Meta and then created look alikes on a ton of different layers. Like you would have payer lookalikes, high payer look alikes, low payer lookalikes, credit card payer lookalikes, renewals, anything. And that would help you kind

of add a little bit of incremental reach on top of these audience with each lookalike. These days it's not as popular anymore because after you didn't have the IDFA anymore to create them and also it's really hard to figure out is a lookalike actually performing better or worse than another um adset just because you don't have that downfunnel tracking and in your onboarding you couldn't ask a person did you come from a lookalike adset. Um so it's not like you you you're going to know but I am definitely see uh still good results with them. So kind of as you scale try them out like for example what you can do like in like a um geo campaign like create like if once you have enough data that is kind of usually the most important. So you want to have qualitative data so enough payers for example to make sense to create one. Um and then match rate on meta tends to be low. So if you now look uh upload your lookalike data, you mainly going to have email addresses. If

you ask for um that information on boarding, but then meta is most likely not going to have the same email for many of the users. So you have to upload quite a bit um like 5 to 10k is usually what I recommend to get um a decent sized look u um core audience and then build look alikes of it and then you can test them out in like a separate country country for example. So if you then run first a broad campaign, let's say you're mo mostly active in in the US, then you first run a broad campaign in Canada, look at what does my overall Canada performance look like um for that time frame, then you pause everything down, then you only run look alikes in Canada. Again, look at how do metrics behave and that can give you a good insight in whether adding that lookalike actually results in a higher quality audience. Another indicator can be just by looking at uh the breakdowns again. So if you're targeting a lookalike and then you directly see wow now my age targeting school is a lot older um before it was I

don't know 25 to 34 mainly as my main uh audience now it's 45 plus that's also already a good indicator conversion is going to be better and you're going to be knowing that based on your um onboarding data again. Gotcha. And also like Meta just announced the return on the AMM. So basically you can get the raw data for the users attributed by uh Meta. Yeah, I saw that. I haven't dug into it yet what that actually is going to mean for like attribution and how much it's going to help. So I I really don't know yet. Um but it definitely sounded sounded interesting. But yeah, you can uh kind of get the users and uh do the lookalikes based on uh the like device ids and create audiences based on that. Yeah, but you're still not going to really have a unique device ID, right? It's not like they can't give you IDFAs because they they also don't have them.

So, in the end, yeah, I wonder what that data is going to look like and what you can do with it. Um, but I mean it did sound like the next step in a direction where everyone says at we're just going to kind of building out these uh these solutions again based on IP and fingerprinting and now like Meta is leading the way there. Well, in the beginning they were very well or I wouldn't say supportive of scan but they said hey if this is the way to go then we're going to be investing in scan infrastructure and by now every account runs on AM and now they're going back they're doing this so definitely it feels like the whole AT thing like Apple doesn't seem to care anymore which is which is interesting. I mean, yeah, let's let's see the data at first. But anyway, good good news for advertisers. That's what it sounded like. Yeah, that's for sure. Let's talk a little bit about creative strategy towards the end of this. Um, I mean, I'm in the end I

the way I or like what I usually work on in terms of creative strategy is like the data and the account and like the process behind it. It's not like I'm making videos myself. I mean I have I haven't actually my first client where I'm in a video which was an interesting experience but in the end I'm not a creator or so. So basically um kind of the main topics I tend to talk about for creative production is that I see a lack of like where people take their inspiration from a lack of like really looking at native content on these platforms and again placements. Um, in the end, often times there is the same kind of creative concepts recircled through the industry because everyone just spies on everyone else's ads. It's easy in meta because you can just go to ads library. Um, and then kind of everyone is using the same stuff. And I would say in the end there is just fatigue of a concept overall across the industry. Like recently, for example,

everyone started saying my app is the Dualingo of this and that or the Spotify of this and that, which was like a big trend. Of course, once a user has seen like 20 ads claiming they're the Spotify or Dual Lingo of something, they're going to feel like I don't give a So really like try to not only look at other ads, which is super important. Like if an ad is working for a direct competitor, you can be pretty sure you can make like a similar ad, but look at native content. Look at what's going on organically on these channels. What are users posting that are from your audience? Um, what does that look like on a Facebook reels compared to like the Instagram feed? And how can you kind of tailor your creators to these formats? because that's going to be like the much more um or like you're going to have a lot more inspiration there and it's going to be a lot more novel than what you see in other people's ad accounts. And in the end, Meta puts a quality score on ads that's often a bit forgotten. Like everyone thinks about like conversion and their bid or like if

like oftenimes people don't use bids. But um then still the kind of auction uh um uh function is that they're looking at uh relevance as well as what you're bidding and how well you're converting. And the relevance piece is something you can work on for free. like it's like if you become more relevant and you have like high quality content that people love to engage with because it fits into their feed then that's going to be the thing that really sets you apart. So make sure you don't forget about that. Um also engagement for me a really important topic that is often a bit under the radar. I often see that kind of ad targeting especially in the early days um of an ad is like super driven just by engagement. So if I mean in the end it's what meta is going to be able to collect on a user level or must. So it makes sense for them to use that signal to target. So if you see an ad

that gets a ton of likes, comments, shares um from a certain audience, it's most likely going to start targeting that audience more and more. And I've seen cases where this just went wrong completely. Like in the end meta takes any engagement as good engagement, but there is a lot of bad engagement happening as well. So if there is a certain audience that is just hating on you the whole time saying like you don't you don't get me this product is uh I hate this meta is going to start targeting them which is not going to be great for performance. So yeah kind of like look into that um especially for your top ads like it makes sense to dig into the comment section a bit you can look into like who's liked. Um, what I like to run for at least my one, two, three top ads is you can even uh kind of uh just copy copy paste from like the Instagram web view into like chatbt and ask it kind of around sentiment have it estimate whether a certain username might be male or female. It's not going to be perfect, but it's going to give

you a good idea like how are different people kind of engaging with this and understand like how that is influencing target targeting. And it can be very obvious where lots of people just write negative comments and you target the wrong audience. But even in a more subtle um way like if you have a ton of people even in a positive way commenting on the ad or engaging with it, but it's not the ideal audience, then that can be a problem over time. I've for example seen like account shift from something like uh 80% targeting male and 20% targeting female to kind of flip that over time just because females have been engaging with the ads more. But then from my downfunnel performance I've seen that they're not converting as well. So kind of over time we've trained the algorithm. Hey these uh women are engaging so it goes after them more and more and more but that drove down performance over time. So then you have to make a tough decision which is do I take those ads down because they seem to be attracting more females or how can I

in the end adjust them in a way that also is going to be more resonating with males so that I can get that engagement signal on top of just the conversion signal. And last but not least here um social proof is important for targeting but also for conversions. So, as soon as an ad has a ton of engagements, comments, uh, on them, it's going to do just a lot better because it shows people, hey, you need to care about this because this is kind of it's going viral. It's important. Um, so it's going to stand out in the feed. So, make sure that you um, for your top performing ads, duplicate them with the post ID. Um, I know Meta in the ad account has this little toggle saying duplicate with social proof or like with engagements or so they frame it. In the end, that only works if you duplicate an ad within the same campaign. Um, as soon as you duplicate something into a new campaign, then you're going to lose it. And there is ways around. Um, one is through the post ID, which you can access if you kind of open up your ad in uh in the

browser like show with Facebook comments or so it says, and then in the URL, it's going to give you the post ID and you can use that for kind of recreating new ads with that same social proof. But that's then only going to be on Facebook. Um, the way to also keep Instagram engagement is through the creative ID. And I'm using a tool there called ad manage AI built by two guys that know what they're doing. And within that, it's super easy. In the end, any ad you duplicate through ad manage AI will always use the creative ID so that social proof is consistent. And over time, that means on one ad, you're going to have the same creative ID in each and every campaign. And social proof just keeps building up. And that's going to become really really powerful as you grow. Nice. Nice. Any questions or do you schedule any time to browse through Instagram just to get any sort of inspiration and see how it works or you use all of the tools? Um, I haven't found like really good

tools to look into organic content like I'd love for someone to build like an AI agent that kind of sets up for example new accounts on Tik Tok or so and then starts to tailor to a certain audience and then look at the content it serves just to learn what's working. Um, I do look into these apps every now and then when kind of I feel like hey let's take some organic inspiration then I look at them. I'm not I'm actually not a big social media user besides LinkedIn when in my free time. So it's not like I'm on Facebook reels all the time. Yeah. Like I haven't opened the Facebook app in a while before I then just for creative inspiration really. Um but yeah, in the end so I have I have certain it's not scheduled but in the end I do take time to look at these placements and try to figure out like what styles could we use for for a new advertiser that I'm working with. Yeah. Um some more on the creative topic. So besides production um it also comes down to like how you're running your creative tests, how you are um making use of the

wins that you are generating. One thing I tend to recommend is like don't feel like it's necessary to move a winner into your business as usual scaling structure right away. Like if things are working well right now, you're hitting your targets and you find like a decent winner that's maybe a 5 10% uplift. it looks fine. It might not be worth to kind of uh mess with your running campaign just to upload that. If it's not like an ad that is totally going to change the game for you because it drove like 30 40 50% better results. I mean that's rare but it happens then it's probably not worth it. like rather keep that ad for when performance goes down which happens with fatigue and then you have something in your backlog to upload because oftentimes what can happen is like you find a few decent wins in today's batch you upload everything and then you don't find anything for the next three four badges and then at some point there's going to be a shortage so

I'd like to have at least some ads in my backlog where that I know I can swap in as soon as things don't look as good anymore Um yeah, creative fatigue in the end to me just like the learning phase, creative fatigue is kind of a scale like um and if you have an ad that performs exceptionally well, it's going to fatigue slower just because if you're like 100% above target and you go to 50% above target, it's still going to be an exceptional working ad. Um so you're going to have a lot more time until it actually gets to a point where it might be hurting performance. If your ad only starts at decent performance, which maybe is like on target, then it's going to be quicker that it gets to be below target. So, kind of keep that in mind. It's not like any ad will have fatigued in a certain time frame. I have some kind of winning ads that have been carrying accounts for like over a year and they're still doing well. Yeah, kind of don't don't rate them all similarly

there and say I need to kind of swap them out after a month or so. And it also makes sense to relaunch like ads that have shown fatigue um after a while because in the end fatigue just means kind of the same people have been seeing the same stuff over and over. So click rates go down, view rates go down on videos and then it gets less less effective. They might not remember anymore within 3 months from now. Also your audience might have shifted just because the account also learns over time. So for example, as I said, like if the account now has shifted to a more female targeting than a male one because that's where the people that have been engaging more and more with ads, then relaunching that ad might already kind of surface it to new users, which again will have higher engagement rates with it. So it's worth keeping this stuff in mind, but of course also always iterating on it. Like if you have a winning ad concept, it should never die. Um like in the end you should always iterate on the hook the messaging swap in new styles kind of figure out how to keep it alive because if the whole kind

of the ad creative framework is working then you can be sure it can work for a long long long time. Um yeah, so even like years and years um I've had ads running like when I still worked in gaming like that we had like a city building strategy game and like in the end showing like progress in an ad and like in the end before afters of cities that are big and small that was kind of the main thing and we did it in a thousand different styles and a thousand different ways with a thousand different cities but the whole idea behind it was the same and we never said okay let's I mean we tried other stuff of course but we never said let's let's not do that anymore and try something completely different. And the other thing which I learned like when I worked with ecom clients more is that I mean it's kind of clear but I think app advertisers tend to not think about it that much is that ads work in sequences. So in the end in ecom that thinking is still a lot bigger where you have like topfunnel um like ads or or campaigns that are like warming up an

audience on like an idea and a new product and that they need it. And then you have bottomfunnel um ads like I don't know aesthetic with like a discount on it that then scoops up all the people that have been sitting on the fence. They show they they viewed like 10 of your videos. they feel like, I need it, but it's too expensive and then they see a discount and they finally convert. Um, but that's how it works everywhere. I mean, most people won't only have seen one ad before they convert into your app as well. So, it's it makes sense to keep that in mind. Like, for example, you can think about like how how you can make sure that people actually notice if an ad belongs to the same advertisers. So if you kind of are running um your um your videos with a certain message at scale and lots of people are seeing them then really think about like okay all the people that have seen that video like what might they need as a next kind of touch point to then maybe decide to finally convert and you can make sure that basically people recognize um that the

next ad they're seeing is actually the one from the same advertiser app. So kind of reusing the same creator can can help. So, if you have like a top performing creator in a video, you could put that same person in a static ad or like do follow-up videos with them. Um, so really kind of get into that thinking a bit. I would say I've seen good success with it that someone like if you have a ad a video that has been running for a while and seen by millions of people, then only 2% of them click through. So, there's 98% of them that watched it that have heard about your app, but you're not talking to them directly. And I'm sure there's ways you can do that a little more smartly than just showing showing them a completely different ad. Gotcha. And one question about the fatigue rate. Is it like automatic? Can you see it? Like do you receive an alert like, oh, the conversion rate dropped significantly. Probably it's a fatigue. There's actually reporting on this by now. I'm not sure when it was added, but Meta surfaces a creative fatigue rate.

I've I'm not sure yet how they are calculating it. I mean in the end you will see creative fatigue based on performance and then a kind of very well functioning winner will just not do as well anymore in a few months from now that's going to come down to like lower click rates which then drive up click costs. So it's more I usually look at it through the secondary metrics but then yeah Meta has this creative fatigue right now. But I've seen that I don't know like I launch a new ad and then I'm scaling it a little bit and even just through the time frame of the creative test it shows that creative fatigue was 2% early on and is now at 30%. I'm like how did that happen in a week? Not sure how they gather that data or how they kind of calculate it. Um but yeah definitely something to keep an eye on like on your top level dashboards. Um just look at kind of click rates of top ads over time and that is going to be a pretty good indicator. Gotcha. Gotcha. I think is it Oh, we

have a operation. I think that's the last section. That's the last section. Let's do this a little more quickly. We're early in one hour. I thought I didn't think I would be talking about uh the stuff so much. Time flies of course. But good good good stuff. Um I think we had most of this already. I mean, yeah, kind of use all available reporting metrics or kind of I love these breakdowns because they can give you so much more context and it's never time wasted if you're digging into like a campaign ad for an hour. Like you're going to learn so much if you figure out like where is this delivering? How does placement correlate with certain age groups, audiences? Um hour of day can be really interesting just to look into um their uh how they're in the end serving the ads. Um, sometimes there's just going to be quirks where you have a day where all your budget gets spent very quickly in the early hours and that's just because the algorithm is kind of doing weird things has been happening more and more often. Um, or also that

you kind of shift your budget from, I don't know, you used to be 80% on Facebook reels and now they're putting everything on Instagram feed and that stuff you just want to be aware of. So if a campaign tanks then I mean it can be that just performance is bad because you're not doing a good enough job. But very often it's also that the algorithm isn't doing a good enough job and then that's no um that's not the point where you're going to have to say I'm going to switch up my whole strategy and do different ads and do everything differently. If you feel like okay this actually looks funky in terms of ad delivery then rather wait it out see when you can start scaling again. So keep a good eye on that. naming conventions super important I think like just for keeping a good overview of the account and being able to break down on additional metrics. So as soon as you like add an additional data dimension in the name of an ad for example I don't know like the usual ones they have are age gender placements that I'm looking at. Um, but then you could

for example implement something like creative type on your ad level so that you call it ad named UGC or static or whatever you have kind of your buckets then that allows you to pull data and use that as a breakdown level because you can then kind of in in an Excel or your analytics tool um have it um aggregate data on that naming piece and that can give you like a lot of insights on really kind of learning what's driving performance, where does my money go? Um, so I would say kind of implement a basic one from the get- go, and you can even kind of um make it longer and longer over time because you're going to have kind of additional data pieces that you want feedback on. And when a consultant like me comes in, it's also just very helpful if the account is at least thought through a little bit and you don't have just campaigns that are called one, two, two, copy, copy. Um, I don't know. So, I've

seen some pretty terrible uh terribly named account setups. And last but not least, yeah, kind of always always ver verify the uh recommendations that you're given by the algorithm, but also meta itself. Uh, it's be it feels like this is becoming a bit of a running gag that I'm ending presentations on uh bashing the marketing pros. Um, but I mean, I don't I don't want to bash them. Um, I've gotten help before with like some issues from their team. Um, but in the end, they're not um marketers. It's like a sales team that is uh kind of an external team to Meta. They're not kind of they don't have access to all the tools someone that is working at Meta would have. And in the end, their main job is to kind of call as many people a day in order to like make them use additional campaign dives, make them spend more at budget. and that's not necessarily aligned with your performance. So you would never have a meta marketing pro telling you, hey, scale down because this doesn't seem to

be working. And of course, yeah, kind of be mindful of that. I see. Gotcha. Um, I have a question, but let's see. It's not related to the slide. Let's see what else do we have. Is it the last one? That's the last one, I think. Yeah. All done. Yeah. Nice. like precisely an hour. But then I have two questions super quick. Uh first question, I'll put it in the comments also to increase the engagement for this video. I'll ask Marcus to respond in the comments after we publish the video. If you guys have questions for Marcus, please also put it in the comments. That also helps that will not only help you, it will help other people and it will help this video to get uh more views. Also, give us a like. Like, it's always super useful since I write LinkedIn content regularly. It's just super useful to me when people ask questions in the comments and I get to

answer them there and then that's usually inspiration for what I write about next. Show me, okay, there is something that people don't understand and I haven't even thought about. Exactly. Uh and of course if you like to support uh what we are doing here with this educational videos please give us a like. Um last question Marcus how can people find you? Mainly on LinkedIn. Yeah. Um Marcus Burke. Um I'm pretty sure I'm the first one that comes up by now. I just crossed like 9K followers. Um and that's my main channel that I interact uh people with. We will leave the link in the description. uh to save folks some time. Um this was super helpful, Marcus. Uh I think that's a very good video to start with meta ads. Thanks a lot for joining and yeah, until next one. Cool. Thanks a lot, Roman. Really enjoyed this. Thanks, man. See you around.