We Tested AppLovin Discovery for 4 Months. Here's What Happened.

[00:00:00] Connaugh: When we talked last, our average incrementality test score for App Eleven was around like 140, 150%, right? Which is super strong. And, and those tests were actually, you know, partially on universal and partially on prospecting. So super strong results even though you know, it's, even though it's still going after those retention and retargeting audiences.

[00:00:25] So it was my hypothesis before we ran any incrementality for discovery that we would see, you know, higher results from that because we're purely prospecting

[00:00:34] in

[00:00:34] Richard: Hey, folks. Welcome to the Ecommerce Playbook podcast. I'm your host, Richard Gaffan, director of digital product strategy here at Common Thread Collective. And I'm joined today by our, let's say, our resident Applovin guru, Mr. Connaugh McLaughlin, who joined us back in January when I was on with Tony to talk a little bit about some tests on Applovin we were beginning to run pr- I mean, incrementality tests specifically.

[00:00:56] And he's back today to talk about, we're here in May, it's been about four months to talk about kind of how that's gone and what we're seeing work on Applovin. Overall, it's an exciting channel, but I think what we wanna dig into is the specifics of what you can do with this channel to make it work.

[00:01:11] So, Connaugh first off, how are you doing, man?

[00:01:14] Connaugh: I'm doing pretty good. I've been deep in the AppLovin trenches, so excited to share some updates today

[00:01:21] Richard: Awesome, man. All right. Well, let's let's dive right into it. So you put together kind of before we jumped on, like, a pretty good agenda here, so let's just kind of roll through it. So first off I think the thing you wanted to talk about was some new cam- campaign types that they've just rolled out.

[00:01:33] Now, I, I think last time you spoke, maybe they were beginning to, so now-- or roll this specific campaign type out, so we wanna talk a little bit about how that actually works in the real world. So let's yeah, speak to that a little bit.

[00:01:45] Connaugh: Yeah, let's do it. So previously when we talked there were two campaign types out, universal and then prospecting as well. Back when AppLovin first came out, there was only universal campaigns, which essentially included prospecting, retargeting, and retention. So, you know, full, full funnel there.

[00:02:05] And then, you know, they came out with prospecting, which goes after new users as well as retargeting people that have been to the site. Now, this new campaign type that came out a couple months ago is called Discovery, and Discovery is in an effort from AppLovin to go after like 100% new users. It's not perfect, obviously, because of tracking cookies, things like that, but I would say most campaigns that we're running on Discovery, we're seeing like between 85% and 90% complete new visitor traffic users that never interacted with the brand before.

[00:02:40] So that's really cool.

[00:02:41] Richard: what's the difference, maybe define like prospecting versus discovery on AppLovin? What's, how do they differ from each other?

[00:02:48] Connaugh: I gotcha. So, prospecting include is, you know, can range between like 60% and 75% of that new tr- new visitor traffic. Then the other percentage is made up from users that have been to the website but haven't converted, so retargeting

[00:03:03] Richard: Gotcha. So, and you'd written down here in your notes too that like, so layering spend, 10% prospecting on top of discovery. So I guess what that points to is like, so how are you, how have you been using this with clients?

[00:03:16] Connaugh: Yeah. So, most of the time we've been really focusing on e- like just discovery. Like we wanna use this as a channel to, you know, purely prospect to an audience that we haven't been able to get to on, on other channels, right? But now, like we're reaching to a stage on with a few brands where, you know, we're capturing that traffic.

[00:03:36] We wanna use AppLovin as like, you know, a full funnel platform as well like we do the other ones. Like on Google we have brand and non-brand. Meta we have prospecting, retention, remarketing. So essentially I'm try- I'm starting to layer in like a 10% either prospecting or universal campaign on top of the discovery so that we can still focus most of the spend on get- going after new customers, but then you also have a little bit of budget recapturing those users too

[00:04:05] Richard: then let's talk about speaking of how we're using this for clients, like how some of that incrementality testing that we talked about, we were just kind of kicking that off back in January, I believe it was. And so talk a little bit about what the incrementality tests have looked like and so far.

[00:04:20] Connaugh: So this is actually super interesting. When we talked last, our average incrementality test score for App Eleven was around like 140, 150%, right? Which is super strong. And, and those tests were actually, you know, partially on universal and partially on prospecting. So super strong results even though you know, it's, even though it's still going after those retention and retargeting audiences.

[00:04:47] So it was my hypothesis before we ran any incrementality for discovery that we would see, you know, higher results from that because we're purely prospecting in those

[00:04:57] Richard: 嗯哼。

[00:04:59] Connaugh: incrementality tests on discovery averaging at 284%, right? So like insane. And one of those is kind of an outlier.

[00:05:11] It's at like 570%. So if I exclude that one, we're looking at like 190% for the other three, which is still, you know, that's probably like a closer benchmark to go off of excluding the, excluding the outlier there. But discovery is highly, highly incremental, so super stoked on these results.

[00:05:32] Richard: that's cool. So you talked about there being sort of a wide range of incrementality results, but it sounds like on the whole, or, or actually across the board, they are positive, right? Like you're, you're achieving a positive incrementality factor across all the four brands that we tested. But at the same time, it's still important to have at least some understand-- 'cause obviously 500% is incredibly different than 190.

[00:05:54] So it's important to have some sense of like what that is for your own brand. But at the very least, we could probably say 190 is maybe the starting benchmark. But anyway, I guess my point is it sounds like there's maybe more volatility with this than other platforms or kind of how-- what do you attribute that to?

[00:06:09] Connaugh: mean, I'm kind of looking at it as like, yeah, we can, we can use the benchmark for sure, but I think it's just equally as important to, you know, conduct your-- conduct an incrementality test for each brand on the platform because it's gonna, you know, because of that wide range. So like if we can find what incrementality factor is for your, for your brand, then we can set the correct in-platform targets and, and make sure we're either not over-attributing revenue or under-attributing.

[00:06:39] Richard: Yeah. But the upshot over all those, like this is, sounds like it's fundamentally a great ac- acquisition channel. Right, yeah.

[00:06:46] Connaugh: 100%. So like,

[00:06:47] Richard: let's talk... Oh, go ahead.

[00:06:48] Connaugh: one more thing there, I think, like meta prospecting, our benchmark is around like 120. So like if we're using this new discovery as the, the main prospecting campaign for Applovin and we're seeing 190% with that outlier, like this is 100% a channel we should be testing into

[00:07:07] Richard: Yeah, for sure. So let's talk a little bit then, speaking of testing, about some of the tactical aspects of this or, or just, just sort of day-to-day what you've been doing. So first off, talking a little bit about creative. You mentioned refreshing it once a week, but t-talk a little bit more about how, how Applovin creative has differed from the way that we approach creative on other platforms.

[00:07:29] Connaugh: Yeah, totally. I think the biggest difference is the ad placements. So the ad placements are gonna fit like 30-second videos, and then, you know, there's other ad placements that are gonna be in your 45 to 60-second range. AppLovin came out with a really interesting blog about their share of spend on video length, right?

[00:07:50] So their 31 to 45 seconds is 37% of total AppLovin spend, and then 46 to 60-second video length is another 44%. So that's crazy. You know, that's, that, that, that's pretty much most of the spend is going to long-form videos. So it's really crucial that we're, you know, utilizing like long-form UGC. With AppLovin we have much longer period to, you know, get your messaging across r- as like other platforms you're, you're trying to capture that user within you know, first three to five seconds.

[00:08:28] Richard: Yeah. And for context there, part of that is just fundamental to the platform. It's because those ads are unskippable, right? This is what I'm recalling from last time. So basically, you get, I think it's 30 seconds, right, before you get the skip option. So that just means that instead of two or three seconds, like you're saying, you have 30 seconds that they-- the user has to watch, so you have that time to kind of w-work through your messaging.

[00:08:49] Okay, so, I, I like this. You have one note here that says, "Ad placement, gotta utilize it." Unpack that a little bit. So what does that mean?

[00:08:57] Connaugh: Yeah. So, what I mean by that is like, you know, s- we've tested, we've tested with like creative under 30 seconds before, you know. But if you're-- if we're uploading 50 20-second videos into the platform and testing those when we have 30 seconds to a minute long, well, the more favorable placement is actually 45 to 60 seconds.

[00:09:23] You know, most of the spend is going there. So if we're, if we're only uploading videos that are under that time range, then we're not fully utilizing, you know, the capabilities of the a- ad platform. And we know it likes to favor pushing spend to that spend or to that video length range. So, yeah, that's what I mean by that.

[00:09:40] Just kinda making sure that we're taking advantage of the platform

[00:09:43] Richard: Gotcha. I see what you're saying. So you have 30 seconds, so you have to use them, and in fact, it should be longer than that because that's what the algorithm favors, it sounds like. Okay, so talk a little bit about just sort of, you had mentioned like refreshing creative once a week, but still having success with biweekly or monthly refreshes.

[00:09:58] What's the difference that you're seeing in terms of like, I don't know, volume constraints, creative diversity, like that type of thing?

[00:10:06] Connaugh: Yeah. Honestly, like it's a, it's a wide range. I, you know, I would say we have a couple accounts where the creative refreshes are a lot slower. You know, maybe we get new creative once a month. We're still able to, you know, run a solid amount of spend, get some good results, but we're definitely not able to scale the budget as fast as if we get weekly creative refreshes.

[00:10:29] Especially if we're, you know, leaning into, you know, every week or biweekly getting more video content that's 45 to 60 seconds long. You know, then we can keep filling keep feeding the algorithm that content that it wants.

[00:10:43] Richard: Gotcha. So ultimately it's, it still needs volume and you're gonna have the best performance with weekly refreshes, but you'll still see something with biweekly or monthly.

[00:10:54] Connaugh: 100%, 100%, yeah

[00:10:57] Richard: in terms of like just speaking to the volume point, in terms of like obviously on Meta everybody's thinking about creative diversity right now.

[00:11:03] Is there anything like that on AppLovin or is it just you can just kind of crank out, throw whatever you want in there? Like what-- how does that work for that platform?

[00:11:12] Connaugh: Yeah, it's interesting. What I've seen work the best so far is probably like long-form UGC videos. So that's been like my my main recommendation to brands is to try to crank out more of that content. Like street interview content has-- is also like a new popular thing that's working well too that our teams are cranking out ads on as well.

[00:11:32] So, I would say those two as well as like more product-focused videos. Also like using captions can increase click-through rate. You know, a lot of people are playing their games with their sound off and not, you know. So using captions on the ads so they're not missing the content that's in the video can be helpful too

[00:11:51] Richard: Okay, cool. So let's, let's, let's roll then into some sort of like tactical ideas here. So I'll, I'll... Actually, I'll just turn it over to you. Like, talk to me a little bit about just tactically speaking what you've been doing and what, what you've seen work in your client accounts.

[00:12:05] Connaugh: Totally. My, my main tactical thing right now has been segmenting my creative sets within a campaign by video. So you can actually have up to 10 videos per creative set. So if you don't have too mu- too much content, like maybe a more, more consolidated structure like that, you know, putting all 10 videos in one creative set you know, you could do that and let the algorithm kind of figure out what's, what works best.

[00:12:34] But you can't... If you, if you put 10 creatives in one creative set, for example, it's really hard to see the performance metrics of each video. So like I've been kind of hacking around the system and doing one video per creative set, because at the creative set level, you can see the performance data, conversion rates and everything, and ROAS by creative set, but you can't see it at the ad level, if that makes sense.

[00:13:02] So,

[00:13:03] Richard: interesting.

[00:13:03] so so you'll-- you, you can't get a sense of individual ad performance, but if you're saying you're, you're breaking out essentially ad set by specific video, then you could get-- that's the way you would sort of hack getting those sort of sp- ad-specific metrics?

[00:13:17] Connaugh: right. And that's just-- I don't think that's like 100% necessary, but I like to like, you know, give feedback to clients and say, "Hey, you know, this video is absolutely crushing. You know, let's get more content like this. These videos have spent a l- spent a good chunk of money and haven't had any conversions."

[00:13:34] So that's just something I'm famil- familiar with reporting and like to report. So, yeah, that's why I've been doing that. But one more thing on that, I don't think it's like 100% necessary. Like at the ad level, AppLovin really focuses on showing you spend volume data as well as like clicks and click-through rate.

[00:13:55] I think a good sign for AppLovin is if, if the video is spending, you know, that means the algorithm likes the video. So, and it has a strong click-through rate. But yeah, the creative set is just an additional layer of reporting I like to do.

[00:14:09] Richard: One other thing you mentioned too is, and may- maybe you just, is don't test too much new creative at once. So that's kind of an interesting concept there. So speak a little bit to like what, what, what's going on with that mechanic?

[00:14:21] Connaugh: Super interesting. I think that's a actually a lesson I learned over the last two weeks because I would upload maybe like, you know, 10 videos at once to test. So I have like, let's say I have 10 creative sets that I know perform really well. If I'm adding in more, if I'm adding in 10 more, AppLovin is really good about pushing spend to new creative so that you get learnings fast on those new creative sets, whi- which is a great thing, you know, but definitely see like, some performance dip while I do that because I'm giving so much spend to new creative I don't know is gonna perform.

[00:15:01] So my now, my new thinking is like, okay, I'll upload like two to three, you know, every three days or so. You know, so it's doesn't have too much of an impact on performance, but I can still get a good read on the new creative.

[00:15:15] Richard: So it seems like there's-- I mean, obviously there's a trade-off there. Like, obviously for many accounts, one thing that people often worry about is their new creatives aren't getting any spend on Meta anyway. Sounds like that's less of a problem, but the trade-off, of course, being that is it that Apple Event will basically spend against non-performing creative whether you want it to or not?

[00:15:35] Connaugh: Yeah. Yeah,

[00:15:36] Richard: Yeah.

[00:15:37] So the, the hack then or is to drip it out. Like, so say you have a weekly creative drop or whatever. You're saying put out, you know, one or two videos a day, and that will mitigate the efficiency problem?

[00:15:50] Connaugh: I think so. You know, one or two a day or every other day, depending on how many, how many how many, how much new content we get in a week, we can, you know, space that out over every other day or so.

[00:16:01] Richard: Yeah. Cool. Okay. Any, any other sort of tactical pieces that you wanna speak to here with AppLovin?

[00:16:06] Connaugh: Yeah, I would, I would say one more thing. I think utilizing the product catalog you know, this may seem kind of basic, but utilizing it in the correct way at the campaign level is really important. So you know, your, your catalog syncs directly from Shopify and within AppLovin you can kind of create different groupings, kinda like a custom label on Google.

[00:16:27] So for example, if you have like, a perfume campaign, you can specifically create a label for all the perfume products and then, you know, put those in a c- in that campaign so that when you get to the product catalog section of the ad after the video and the static, you know, the products that you're advertising are the ones showing in the product catalog.

[00:16:49] So yeah, segmenting by campaign type and then at the product or at the campaign level, making sure the correct products are selected.

[00:16:57] Richard: Gotcha. So actually, and maybe for the sake of visualization then, like a, a catalog app or the use of the product catalog, I guess, on AppLovin looks like video, like static end card, and then products? Or like just sort of like a, what, static images of your product photos or something like that?

[00:17:15] Connaugh: Yeah. So like, I mean, that's, that's another really unique aspect of AppLovin is you get all three aspects in one ad, right? So you get your 30-second unskippable video, and then you get your static ad, and then you get your product catalog directly after those, right? So it's a, it's a, like a full process the user goes through for sure

[00:17:36] Richard: Interesting. Okay, cool. Well, I think, yeah, if there's no- nothing else that you wanna leave us with. Actually, I mean, you mentioned that it's a no-brainer test with the 5K ad credit. So I think that seems like it's overall sort of the message here is that what our tests have shown is that AppLovin is absolutely worth it to find for sure incrementality, incremental customer acquisition.

[00:17:57] So check it out. You get a 5K ad credit. What, what are the, what are the sort of stipulations on that?

[00:18:03] Connaugh: Yeah, exactly. So, CTC with partnership App-- with AppLovin, we get a 5K spend match. So spend 5K in the platform, AppLovin will automatically apply 5K to your account and it'll automatically start spending. You know, we've been able to we've been able to ramp up brand, brands to, you know, 30, 50K plus within a month.

[00:18:22] However, like if you wanna do a more conservative test, I think, you know, t- that 10K, the 5K initial investment and then the 5K free ad credit, that's a 10K test over a month or two, and then you can get a pretty good signal on, hey, this is like gonna work for us.

[00:18:38] Richard: Yeah, that makes sense. Okay, well, if, if you wanna access that, if you want us to or rather if you wanna have our AppLovin ad guru Mr. Connor McLoughlin, on your account, you know where to find us, Comethroughcode.com. Hit that hire us button. Let us know you're interested in talking. We would love to work with you.

[00:18:53] So, I think we'll leave it at that. Connaugh appreciate the time. We're gonna check in, back in with you probably in a few months to just see where we're at with AppLovin overall how that's trending. But until next time, folks, take care.